Biology of Aging: Observations and Principles, 3rd Edition

The Biology of Aging

This page intentionally left blank

The Biology of Aging Observations and Principles Third Edition

Robert Arking

1 2006

3 Oxford University Press, Inc., publishes works that further Oxford University’s objective of excellence in research, scholarship, and education. Oxford New York Auckland Cape Town Dar es Salaam Hong Kong Karachi Kuala Lumpur Madrid Melbourne Mexico City Nairobi New Delhi Shanghai Taipei Toronto With offices in Argentina Austria Brazil Chile Czech Republic France Greece Guatemala Hungary Italy Japan Poland Portugal Singapore South Korea Switzerland Thailand Turkey Ukraine Vietnam

Copyright © 1998 Sinauer Associates, 2nd edition Copyright © 2006 by Oxford University Press, Inc. Published by Oxford University Press, Inc. 198 Madison Avenue, New York, New York 10016 www.oup.com Oxford is a registered trademark of Oxford University Press All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior permission of Oxford University Press. Library of Congress Cataloging-in-Publication Data Arking, Robert The biology of aging : observations and principles / by Robert Arking. — 3rd ed. p. cm. ISBN-13 978-0-19-516739-9 ISBN 0-19-516739-2 1. Aging. 2. Physiology, Comparative. I. Title. [DNLM: 1. Aging—genetics. 2. Aging—physiology. WT 104 A721b 2006] QP86.A75 2006 612.6'7—dc22 2005030674

1

3

5

7

9

8

6

4

2

Printed in the United States of America on acid-free paper

To Lucy, who encouraged and For David and Deanne, Jonathan and Carolyn, Ben, Jared, and Rachel; Joshua and Emily— who will know. Theory without fact is fantasy, but fact without theory is chaos. C.W. Whitman, 1894

Seek simplicity and distrust it. Alfred North Whitehead (cited in Gilbert, 2000, p. xvii)

The effort to understand the universe is one of the very few things that lifts human life a little above the level of farce, and gives it some of the grace of tragedy. Steven Weinberg, The First Three Minutes, 1977

This page intentionally left blank

Preface to the Third Edition

There were two inspirations for writing this third edition. The first was the qualitative change in our understanding of the genetic basis for longevity and senescence which occurred since 1998 when the second edition was printed. When that text could no longer serve by itself as an adequate text for my own classes, I realized that the time had come to take on that task once again. The second inspiration was the generosity of the reviewers and colleagues who commented on the second edition in a mostly positive manner. No good deed goes unpunished, and so many of those same individuals have had to trudge though another review or critique. Experienced readers will, I hope, note that there are at least six important conceptual changes that have been made in the presentation of the data. Perhaps the most important is the stronger distinction drawn between the biological mechanisms involved in longevity determination (mostly described in chapter 7) and those involved in senescent processes (mostly described in chapters 9–13). And these processes are defined in a time-independent manner in chapter 1: “If aging is a series of increasingly different and less functional molecular and physiological signatures, then senescence comprises the processes that are responsible for the changes in those signatures.” That first chapter also sets up the beginning of an integrated theory of aging over the life span, which is updated as the text progresses. A second change emerged from my wrestling with the interesting and insightful genetic data published in the past half-dozen years. The sheer mass of this information forced me to develop a conceptual framework on which to mentally hang all these facts, if only to avoid the deterioration of this book into a laundry list of unrelated items. And so the organization of chapter 7 is based on

the matrix of public mechanisms via which longevity seems to be regulated (i.e., metabolic control of several types, stress resistance, genetic stability, reproductive effects). This framework seems to accommodate diverse data without obvious strain, and so it offers an obvious pragmatic value in assisting readers to conceptualize the information and make it theirs. Third, our increased knowledge of aging cells allows me to put forth in chapter 9 some ideas as to how a cell transitions from a healthy state to a senescent state, but in such a manner as to still allow for high levels of intra- and interspecific variability. Fourth, the renaming of aging mechanisms as senescent mechanisms, which is the emphasis of part IV of the text, is a name change that I think will help the reader understand that aging is a nonprogrammatic loss of function that is, however, somewhat plastic and that can be modulated. As before, I have grouped these losses under the rubric of damage to the gene interaction networks arising from stochastic damage, from mitochondrial damage, and from degradation of both intra- and intercellular regulatory systems. Recent data have forced a reevaluation of the roles these various senescent processes play such that we may see mulitple mechanisms at play here. Fifth, the standard evolutionary story does not fully explain the evolution of social organisms, and so I have incorporated recent work that deals with intergenerational resource transfers into a discussion of human longevity. I am intrigued by the possibility that extended longevity may well have played a mostly unappreciated role in our becoming quintessentially human. Sixth and last, if both longevity-determining mechanisms and senescent mechanisms are plastic and can be significantly modulated in the laboratory, then the

viii Preface to the Third Edition demand to move these anti-aging interventions into the human arena will inevitably grow. Discussions about longevity extension are as old as Methuselah, but the past few years have witnessed a growing discussion of this topic in diverse venues. These public discussion have rambled all over the field, with the consequence that arguments and questions get muddled. I have included here a full discussion of my own opinions on the biological, social, and ethical aspects of one highly focused aspect of this debate. My striking a clear position is not meant to propagandize but rather to encourage readers to argue with me and reach their own conclusions, but without abandoning what we can all agree are the demonstrated facts. There are many words in this book (too many, some will say). And yet I think that the main ideas underlying this text are summed up in three graphics (figures 1.6, 9.6 and 14.9). If that last statement is true, then why did it take so many words and chapters to describe these central ideas? One reason is that this is a databased textbook, and so the basic statements must be well supported by robust data. In an emerging field such as is biogerontology, capable people may obtain differing results or interpret the same results differently, and so the disagreements of the field should also be laid out for students to read and ponder and reach their own decision. Another reason is the need to point out things that might not be fully clear to a person just entering the field, such as relationships between apparently unrelated variables or to illustrate interactions between the different complexity levels of the cell or organism. Finally, I believe some of the many words are necessary to describe the broad scope and deep richness of the data now available and the predictive hypotheses that now exist. I fervently hope that my readers will more or less agree with this rationalization. Even so, not all the words could fit into this book, and so I have constructed a website where both supplemental and updated material may be found (http:// bio.wayne.edu/profhtml/arking/textbook/ supplement.html If this text offers a useful view of the processes and mechanisms underlying the biology of aging, then that view was made possible only because I

have built on the shoulders of my colleagues who have conceived and executed informative experiments that made things clear and/or wrote useful reviews that astutely summarized an area and made it easier to grasp. I thank them for their efforts and point out that the references cited clearly reflect more the limits of my mind and my desk rather than the limits of what has been done. And so I apologize to those whose work was overlooked, and I look forward to them bringing me up to date. I will say again that I am fortunate in the acuity of my critics. Their sharp eyes have saved me once again from embarrassment or errors, whether of omission or of commission. I accepted willingly and gratefully most of their suggestions and criticisms, for they were made in a spirit of collegiality. These colleagues were generous with their time and knowledge in critiquing the draft of this third edition. I must particularly thank the reviewers who so capably critiqued these chapters and made many useful suggestions and corrections: Robert Avila, Daniel Callahan, James Carey, Vince Cristofalo, Aubrey de Grey, Michael Fossel, Mark Greene, Barry Halliwell, Nikki Holbrook, Don Ingram, Nicole Jenkins, Tom Kirkwood, Christiaan Leeuwenburgh, Jeff Leips, Gordon Lithgow, George Martin, Gawain McColl, S. Jay Olshansky, John Papaconstantinou, Serguei Scherbov, and Ron Woodruf. I must also thank Craig Giroux and Michael Fossel, among others, who shared with me their logic and passion in the quest to understand the cell. But stubbornness dies hard: I did not accept every suggestion, and so I must accept full responsibility for any errors or inaccuracies in the work. And of course I must once again acknowledge the sometimes vocal assistance of the students in my Biology of Aging classes at Wayne State University, who pointed out to me the strengths and weaknesses of the revised text as I was preparing it. The text is again more readable because of their efforts. I thank Kirk Jensen for his enthusiasm for the idea of publishing with Oxford University Press, and Peter J. Prescott for his skill in bringing the project to reality—in large part by gently reminding me of the virtues of brevity. Some may believe

Preface to the Third Edition

I didn’t listen, but that is not the case. Only Kaity Cheng’s organizational capabilities kept this project from becoming an embodiment of chaos theory. I am truly grateful to Andy Sinauer for making available to me the text and artwork of the second edition. The current work owes much to his generosity and goodwill. Matthew Garin deserves full credit for translating my crude sketches and marked-up photocopies into distinctive figure files. The organization of the book has changed again, for it is now divided into the answers to the five questions contained in the Table of Con-

ix

tents. Although a full understanding requires the students to address each of these questions, there is sufficient material in the several key chapters so that each instructor can, by means of judicious choices and emphases, impart their own interpretation to their course. The point is to tell a coherent and realistic story that will reliably guide the students’ future thinking on the topic. If this book achieves that goal, I will be satisfied.

Robert Arking October 2005

This page intentionally left blank

Preface to the First Edition

If we are truly fortunate, we will age. Each of us will struggle with this fate in our own way. There has been much attention focused on the biomedical, economic, social, and psychological aspects of human aging, but until recently serious biological attention was given to this topic by only a few farsighted innovators. In part, this was because our attention was mostly focused elsewhere—perhaps on the triumphs of molecular genetics in deciphering the genetic code and in unraveling the molecular mechanisms that regulate gene action. These biological insights are traditionally viewed in the context of embryological development. The reasons for the neglect of the rest of the life cycle are not clear, but they probably had a lot to do with the scientific and cultural prejudice that aging is not a fundamentally interesting or attractive biological process. But our concepts have now changed, in part due to the demographic changes taking place in society and in part because we are now beginning to understand that our present biological views will not fully explain aging. Perhaps it was necessary to attain our present level of understanding of embryological development before we could appreciate the complexities inherent in the biological problem of aging. I view aging as a fundamental biological process that can be defined, measured, described, and manipulated. My researches and readings have led me to suggest that aging is a genetically determined, environmentally modulated, eventdependent process. I have tried to construct this book so as to serve the reader as a guided tour through the literature that has led me (and others) to this conclusion. Although I have taken care to present the conflicting data and opposing points of view that characterize this unsettled field, the book is not intended as a monograph addressed to other specialists. I have written this book for

students of aging (be they formally enrolled or not) who have a level of biological knowledge no more sophisticated than that provided by any good introductory biology textbook. I believe it is important for people conversant with the sociological and psychological aspects of gerontology to also be knowledgeable about the biological aspects of aging and the implications of the current research for their own fields. I have tried to explain in a clear manner the logical bases of the arguments and have veered away from overwhelming the reader with too much unnecessary jargon and details. But interpretations cannot be made without data, nor without thought, so I selected what I believe to be pertinent facts and observations. I hope the reader will think about them and not just accept them uncritically. Let me explain the organization of the book. I believe it is important to first be sure of what it is we think we know. Accordingly, we begin the journey with a rigorous definition and exposition of exactly what does—and does not—constitute an aging change. The CUPID (cumulative, universal, progressive, inherent, deleterious) definition developed here guides us through the thickets of facts, interpretations, and complexities that beset the path. I then discuss the several ways of measuring aging. It is a difficult task and one which is often skipped, particularly by those of us bearing childhood fears of mathematical thoughts, but it is important to master the concepts involved (if not the numbers) simply because of the old axiom, If you cannot measure what you are studying, then you do not know what you are talking about. I do not propose to weigh the human spirit, but there is no reason not to assay our bodies or measure our molecules. I have deliberately adopted a comparative approach to the study of aging. If aging is a fundamental biological process, then we can learn

xii Preface to the First Edition much from the study of diverse and even exotic laboratory animals, some of which will, if we are fortunate, age in such a way as to illuminate some particular aspect of the aging process. I have surveyed comparative aging research by using those examples that illustrate particularly well some special aspect of aging that might otherwise have been overlooked or not appreciated. An anthropomorphic approach is an illogical process to use when trying to understand basic biology. Nevertheless, I have not ignored humans. I have used humans as the examples of vertebrate aging, and so a comprehensive (but far from definitive) chapter on the normal and abnormal aspects of human aging has been included. Instructors, students, and other readers may choose to go over this in detail, amplifying it as necessary, or else choose to read the highlights at the beginning and end of the chapter, depending on their background and goals. I then round out our survey of the known facts by examining the proven genetic and physiological predictors of longevity as well as the various tested methods of modulating the life span. Once we know what it is that we know, then we are finally equipped to discuss and critically evaluate the several different theories of aging. This task fills up the second half of the book. All the theories are plausible; what I hope the reader will come away with is a sense as to which theories have been critically disproved, which are still untested, and which appear to be both plausible and probable. The proper assignment of the probable mechanisms of aging now will have much to do with our eventual success in better understanding it later. The facts presented in the first half of the book play an integral role in this analysis. I conclude the investigation by examining whether or not there is a fundamental aging mechanism.

At one level, of course, it is obvious that there are a multiplicity of aging processes. Yet this cannot be taken to foreclose the existence of common processes any more than the existence of the multiple ways in which different species progress from an egg to an adult cannot be used to obscure the fact that there are probably only a small number of fundamental developmental mechanisms, mechanisms common to broad groups of organisms. The existence of common fundamental aging mechanisms might make it easier for us to more fully understand and perhaps even manipulate the aging process in future. Of course, such a proposal is only a target at which future researches may take aim to disprove. If aging is a process that baffles all of us, then how presumptuous is the ordinary professor who believes that he can write something of value about it? Especially since the sage has lamented that “Of the making of books, there is no end.” Well, yes, but . . . My motivation in writing this text grew out of my own need to understand the field of biological gerontology, an area toward which my research was leading me, and to organize it within a conceptual framework that made sense to me. I enjoy teaching and sharing with students whatever it is that I know. The text evolved and grew out of the lectures I wrote for such a course, a course I volunteered to teach because it struck me as an efficient and easy way in which to learn and understand and organize the literature. With the wisdom of hindsight, I can say that it was mostly enjoyable; it may even have been efficient, but it was certainly far from easy! It was only because of the efforts of many other people that I was able to complete the task I had so presumptuously set for myself, and it is to them that much credit is due.

Contents

Part I What Is Aging? 1

Perspectives on Aging

3

2

Measuring Age-related Changes in Populations

26

3

Measuring Age-related Changes in Individuals

54

Part II Why Do We Age? 4

Evolutionary and Comparative Aspects of Longevity and Senescence

95

Part III How Do We Age? 5

Human Aging

135

6

Altering Aging: Interventions That Modify Longevity and Senescence

202

7

Genetic Determinants of Longevity in Animal Models

237

8

Genetic and Social Aspects of Aging in Humans

319

Part IV What Is the Mechanistic Basis of Aging and Senescence? 9

Mechanisms Underlying the Transition from Health to Senescence

347

10

Stochastic Theories of Aging

363

11

Systemic Theories of Senescence

395

12

Senescence as a Breakdown of Intracellular Regulatory Processes

419

13

Senescence as a Breakdown of Intercellular Regulatory Processes

449

Part V An Integrated Theory of Aging 14

A Theory of Aging over the Life Span

483

xiv Contents

Part VI What Can We Do about Aging? 15

Aging-related Research and Its Impact on Society

505

References

527

Index

591

Amount

Part I

Optimal level

What Is Aging? Time

This page intentionally left blank

1

Perspectives on Aging

1.1 Introduction Through the centuries, sages have pointed out that many of the more profound aspects of human culture, the sometimes tragic struggle of humans against fate, originate in the fact that we all must die. Great art and major religions flow from the contrast between our boundless dreams and ambitions and the realities of our temporal prison. It is unclear when this concept appeared; indeed, it is unclear whether any other species shares with us a recognition of the inevitability of death, although some primate cousins share our sensation of an individual consciousness. Our Neolithic ancestors almost certainly were aware of our common fate and felt the same tension, for 50,000 years ago at Shanidar in what is now Iran they buried their dead on a bed of wildflowers. Then, as now, senescence and death were likely to have been accepted by most people as given conditions of existence. The few dissenters searched for a magic potion or fountain of youth in attempts to escape their fate. Most people just searched for an explanation to justify their fate and were satisfied with a supernatural or religious interpretation. All were aware that humans age and, if they lived long enough, succumb to fraility, senility, and death. It is likely that this recognition by our ancestors that an altruistic life does not avert aging and death underlies the origin of religions (Holliday 2001). Our preference for the new is not due solely to the efforts of the advertising industry to sell us the latest consumer item. Each of us absorbs

as we grow up the undeniable truth that old things tend to wear out and break down: old toys, old cars, old machines—and old people. Our reaction to this reality takes at least four forms, three of which have been best expressed by the artists among us. First is the acceptance and celebration of our mature years, freed of lingering diseases, as penned by Robert Browning: Grow old along with me. The best is yet to be, The last of life, for which the first was made . . . “Rabbi Ben Ezra,” 1864

Second is a refusal to accept aging. Many have fought senescence and death, knowing it to be a struggle they must lose but nevertheless fight because they could do nothing else. Dylan Thomas perhaps best echoes their feelings in these lines: Do not go gentle into that good night, Old age should burn and rave at close of day; Rage, rage against the dying of the light. (“Do Not Go Gentle Into That Good Night,” 1953)

The difference between these two views is due, in part, to how one sees life. Perhaps Browning’s proponent celebrates mature love and companionship, secure in the belief that mortality makes life and the enjoyment of it precious; that the

3

4 Chapter 1 Perspectives on Aging sense of not having world enough and time enough is the spur to our achievements, not the least of which should be to master the art of living well. To these arguments the advocate of Dylan Thomas might reply that he rages precisely because there is neither world enough nor time enough for a short-lived human to know what can be known or to explore what is not yet known. As long as aging was, like the weather, something that we could not control, then the Browning–Thomas debate remained a philosophical dialogue without a resolution. Now that we are beginning to exercise our new-found knowledge regarding the manipulation of aging, this leisurely debate has metamorphosed into a high-stakes ethical, scientific, and political contest. We will revisit this dialogue in the last chapter. A vital and vigorous life is precious to us; that is why we both celebrate and rage at its finite length. A cooler, more intellectual reaction is to describe the events; this approach constitutes the third form of response. An important advance was the explicit recognition that each human follows the same path of growth, development, maturity, and senescence—a process that has never been described better than by Shakespeare in his famous passage of the seven ages of man from “As You Like It” (1600, act 2, scene 7). The regularity is what catches our eye, for it suggests an underlying and predictable mechanism. The fourth form of our reaction to the reality of aging is the scientific investigation of the biological mechanisms responsible for the predictability of our aging. It took the three centuries following Shakespeare, during which classical biology was established, before August Weismann (1891a,b,c) could even begin to formulate the first mechanistic questions relating aging and evolution. These questions were reformulated through the experimental efforts of various investigators, such as Élie Metchnikoff in Russia and Raymond Pearl in the United States. Together, Metchnikoff and Pearl demonstrated that a characteristic similarity of aging and senescence transcends species boundaries, and they postulated mechanistic theories to explain and predict the aging process. But the complexity of the topic defeated these initial attempts at

understanding, and the attention of biologists was otherwise captured by the more promising prospects being developed by Thomas Hunt Morgan in genetics, Hans Spemann in embryology, J. B. S. Haldane in physiology, and Otto Warburg in biochemistry. Shortly after the molecular biologists had begun unraveling the mysteries of the gene, the publication and widespread acceptance of Alex Comfort’s book The Biology of Senescence (1956), led to an affirmation of research on aging as an important aspect of basic biological inquiry. Comfort achieved this affirmation by summarizing the available data with a critical eye and wellturned phrase and by being succinct: The primary assignment of gerontology— that of finding an accessible mechanism that times the human life-span as we observe it— remains undischarged. But it is nonetheless far closer to that objective today than when we last reviewed the subject—partly because, through the growth of experimental evidence which the pretheories of the past have generated, the possibility of a hierarchy of aging processes integrated by a life-span “clock” has come to be reorganized and the nature of that clock is becoming clearer. (Comfort 1979, p. 16) This statement very specifically defines the problem and how to face it. But the increase in knowledge that Comfort was instrumental in effecting has altered our concepts and redirected the problem. Many biogerontologists would today dispute the idea of a “life-span clock” that measures our time and would instead advocate in both fact and metaphor a more multifaceted and diffuse type of mechanism. For example, Finch (1990) has pointed out that biological time is functionally equivalent to cascades of specific physical or chemical events and thus is fundamentally independent of absolute sidereal or calendar time. Time does not directly measure the changes we each undergo. The evidence to be presented throughout this book will strengthen the idea that aging is fundamentally an event-dependent, and not a time-dependent, process.

1.2 On the Nature of the Puzzle: The Difficulties in Studying Aging

The fact that we continuously modify our concepts is not surprising, given the variety of disciplines from which the knowledge required to solve this problem must be drawn. It should be evident on reflection that gerontology is a field of inquiry, not a fully autonomous academic discipline. Gerontologists, then, are people from a variety of backgrounds who share an avid interest in aging. To understand the biology of aging requires posing testable and reasonable questions: Why do two species so closely related as mouse and man have such very different life spans? What causes the deteriorative changes during the life span of each of these species? Are the causes the same in both species? Can the deteriorative changes be postponed? Reversed? Is it possible to reliably predict the life span of an individual? Is it possible to prolong the life span? Is it worth it? It is the task and the goal of gerontologists to find the answers to these and similar questions. Our knowledge about the nature and causes of aging and senescence is accumulating increasingly rapidly. This knowledge is not just an accumulation of depressing facts. In the last decade the rigor of gerontological thinking has increased remarkably, as evidenced by the shift from a purely descriptive to an increasingly analytical approach, as well as by the correspondingly more detailed quantitative examination of various cellular and physiological mechanisms. We do not yet know the answers, but the fog that obscures them is lifting, and we can now see at least the outlines of the answers. A brief model of the mechanisms underlying the aging process is presented at the end of this chapter to provide a conceptual framework into which the myriad facts and evidence presented throughout the next dozen chapters may be coherently fitted. The model will be expanded as we progress through the evidence, and an integrated version will be put forth in chapter 14 in an effort to summarize our knowledge. One question has already been answered. The quest for immortality is biologically hopeless. Given our present state of knowledge, it is more beneficial to opt for a healthy and vigorous, albeit finite, life than to search in vain for the elixir of immortality. Thus, gerontology is committed not to a search for immortality but to the elimi-

5

nation of premature disability and death, to the deciphering of the mechanisms that regulate our longevity and our aging, and to the extension of the healthy portion of our life. Our increasing knowledge makes this latter goal appear more attainable with each new biological insight. Yet some serious observers of the contemporary social scene object to the marriage of gerontological knowledge with modern biotechnological techniques. They believe that the goal of that union— to bring about the significant extension of the healthy portion of our life (the “health span”)— constitutes an ethically suspect manipulation which demeans human dignity. These are serious concerns which should not be ignored. But it would be premature to address them before we have studied the biological information. Only then, in chapter 15, shall we delve deeper into this and other questions related to the general topic of aging research and societal goals. It seems that the debate between Robert Browning and Dylan Thomas has not really been settled but merely shifted its venue from the philosophical to the political arena.

1.2 On the Nature of the Puzzle: The Difficulties in Studying Aging Apart from the historical and philosophical blinders that make it difficult to visualize the topic of gerontology as a whole, a fundamental problem of causality impedes our progress toward understanding the mechanisms of aging and senescence. A normative scientific inquiry such as gerontology usually begins by a more or less systematic description of the functional and structural changes that accompany aging. These descriptions initially may be based on human studies and later involve animal models. They are usually qualitative at first, as Shakespeare’s description was; later they become quantitative, as the longitudinal studies reported by Shock (1985) demonstrate. Because aging is a complex process that affects a wide variety of functions, even the most casual investigator is soon overwhelmed by the

6 Chapter 1 Perspectives on Aging large number of apparent associations between the aging process and various phenomena. The complexity of the topic gives rise to a major epistemiological problem: How shall we judge which of the many changes associated with aging are actually real and which are spurious artifacts? And how shall we objectively sort out the actual associations into those based on strong data and those based on weak data? Many people view science as an activity dealing with the collection of more and more obscure facts until their sheer number alone lets us see the “truth.” Nothing could be further from the mark. Biology, like any other science, does not deal with mere facts; rather, it deals with evidence, which is quite different. Scientists try to understand what is happening in and to the object of their studies, and so they construct hypotheses regarding the mechanisms at work. In the present context, we want to know what mechanisms cause the aging of the body. Once a hypothesis has been formulated, then most known facts are irrelevant to the question at hand. The only relevant facts are those that can conclusively disprove the theory in some logical manner. A fact that disproves the theory is strong evidence against its acceptance; facts that are consistent with the theory can argue for its tentative acceptance. Even though a scientific theory cannot logically be proven to be not wrong (i.e., correct), it is the consistent support of the theory by independently obtained evidentiary facts which establishes and solidifies a conclusion. When consistent efforts to disprove a theory all fail, then there is a high probability that it is correct. But not all classes of evidence are equivalent. The three types of evidence are (1) correlative, (2) loss-of-function, and (3) gain-of-function evidence. Correlative evidence arises from the observed temporal or spatial correlations between two or more events, and carries with it the inference that one event somehow caused the other. In most cases, these correlations furnish few direct clues regarding the nature and identity of the underlying causal mechanisms, despite the high degree of statistical significance. For example, graying hair in humans has a very high coefficient of correlation with chronological age, yet no one

would seriously propose that gray hair is a cause of aging. It could just as reasonably be argued that the reverse is actually true: that aging caused one’s hair to turn gray (and this is actually closer to the facts as discussed in Chapter 3). In that case, we are no closer to an understanding of aging than we were before. So correlations give us a starting point for an investigation, but they do not offer convincing evidence of a causal relationship. Most of the theories that result from this correlative approach appear to be more plausible than this extreme example, usually because they involve important changes and postulate a physiologically reasonable mechanism that could bring about the desired effect. Gerontologists are ingenious, and consequently the field has never suffered from a lack of theories. For example, consider the large differences in rate and timing of the age-dependent decrements of different physiological functions in humans, as shown in figure 1.1. These longitudinal and cross-sectional data clearly illustrate that different functions decay at different rates, even within a single individual. The heterogeneity of these age decrements has been used as an argument against the idea that the rate of aging is controlled by any single basic process. It is only reasonable to suppose that the difference in the rate of aging of each organ reflects the fact that different processes are at work in each one. Conversely, the same data have been used to buttress the idea that there is one central pacemaker process, for the heterogeneity is exactly what would be expected if different systems were responding at different rates to the same stimulus. With enough ingenuity, descriptive or correlative data can be argued both ways. Persuasive arguments are available; solid evidence is needed. One such approach is to obtain loss-of-function evidence, perhaps by disabling a gene or other molecule thought to be important in the process being studied. “Knocking out” a gene involved in the cell’s defenses against oxidative stress might reveal that the organism does not appear to differ in its somatic damage rate or age at death from the controls. Such a result might strongly suggest that the gene in question plays no role in defending the organism against oxidative stress. Although stronger than correlative data, loss of

1.2 On the Nature of the Puzzle: The Difficulties in Studying Aging

7

100

Percent of functional capacity

90

▲ ▲

80

▲ ▲

70

▲

60

▲ ▲

50

▲

40

▲

30

▲

20 10 0

0

10

20

30

40

50

60

70

Personality Nerve conduction Cardiac index High-frequency hearing ▲ Maximal breathing capacity ▲ Maximal work rate Renal blood flow Accomodation capacity

80

Age (years) Figure 1.1 Age-dependent changes in some anatomical and physiological factors in humans, as reported in various reports of cross-sectional and/or longitudinal studies from the Baltimore Longitudinal Study on Aging. For most factors, the level at age 30 was taken to represent the optimal response and assigned a value of 100%; the other age-specific data are expressed relative to this base value. Younger baseline ages are used for measures of highfrequency hearing and of personality. The data are presented here as schematic linear projections that omit the inter- and intrapopulation variability inherent in the original data; however, the overall trend is not obscured. (Data assembled by G. T. Baker III and J. Frozard on the basis of Gerontology Research Center studies.)

function still doesn’t unambigously rule out other inferences. For example, perhaps other molecules within the cell could compensate for the disabled molecule and increase their activity so as to maintain the cell’s overall ability to deal with oxidative stress. The observed results might well be evidence of an integrated cell defense network rather than evidence that the gene in question is not involved in defense against oxidative stress. On the other hand, suppose an investigator knocks out some particular gene and then observes that the animal lives significantly longer. This result can be construed as indicating that the gene in question acted in some way as an inhibitor of long life. Finally, the third possibility is that the knock-out mutant might express a very short life span. Such a result would indicate that the particular gene might play an essential role in the animal’s ability to express a normal life span. All three types of results need to be confirmed by appropriate follow-up experiments. All three types of results will be encountered in the literature that I survey in this volume. The fact that one experiment can yield three different outcomes, each of

which is interesting, means that our judgment as to the validity of the initial result depends on the incisiveness of the several different follow-up experiments. Thus, it will be useful to pay attention to the entire chain of evidence rather than to be swayed by one dramatic result. The strongest type of evidence is gain-of-function evidence. In this case, one might use some technique to specifically increase the cellular activity of the same gene discussed above which is involved in the organism’s defenses against oxidative stress and observe that the organism has a lower damage rate and dies at a later age than do the controls. In this case, initiation of the first event (up-regulation of the gene) causes the second event (extended longevity) to happen under conditions where it would otherwise not occur. The gold standard of evidence that an experiment has identified an important aging mechanism is to extend the animal’s longevity by some sort of gain/loss-of-function experiment. Of course, the experiment could have two other outcomes: the animal might live a shorter life, or there might be no obvious effect on its longevity. In the

8 Chapter 1 Perspectives on Aging former case, the increased dosage is deleterious to the organism. In the latter case, it appears to have no effect. Again, confirming any of the possible outcomes of a gain-of-function experiment will require appropriate follow-up experiments; and attention should again be directed to the entire chain of evidence. The willingness of biogerontologists over the past decade or so to apply critical tests using evidence capable of disproving specific hypotheses has fundamentally transformed the field from a descriptive survey of interesting phenomenon to an analytical study of biological mechanisms bringing about the loss of function we know as senescence. The identification of specific genes in laboratory animals and in humans that have significant effects on aging and longevity has transformed the field, as has the simultaneous molecular analysis of behavioral or physiological interventions long known to affect aging and longevity. Teasing out the networks of mechanisms that alter aging from the gene up and from the organism down is, in the process, yielding an integrated view of the subtle regulatory mechanisms that operate in the laboratory animals and presumably in humans as well. There are two very interesting findings, to which I will return in future chapters, but simply state here. One is that although some anti-aging mechanisms are species specific, others appear to be highly conserved across phylogenetic lines. In the latter case, insights obtained from investigations on one species of laboratory organism (worms or flies, say) can be translated to rodents or to primates and, perhaps someday, to humans. Although all aging mechanisms are of some interest, conserved mechanisms have the greatest significance because they may suggest possible interventions into human longevity. Second is the idea that although aging is undoubtedly complex, it may in fact be regulated by common mechanisms that are simpler than the effects they produce. It is important not to let our hopes sway our judgment. As you read this book, probably the best advice to keep in mind is the instruction of Alfred North Whitehead, who wrote that every scientist should “Seek simplicity and distrust it” (cited in Gilbert 2000, p. xvii).

People have been aging at least since our species came into existence. What, then, accounts for the recent interest in aging and senescence? Certainly the interest in the social, psychological, and medical aspects of gerontology had a dual genesis: first, in the demographers’ realization that the elderly would soon become a significant aspect of the population; and second, in the federal training, service, and research programs established in the mid-1960s in response to this awareness. The biological interest came a little later and a little differently. It is a commonplace observation that scientists appear to possess a collective awareness that causes many of them to ask the same sorts of questions more or less simultaneously. The cause is not a metaphysical process; it is simply the summation of numerous individual assessments of recent advances, measured against the kinds of questions to which the new knowledge might be most appropriately applied. Traditionally, biologists usually attempted to explain the aging process in terms of general biological phenomena that were under intensive study or that seemed most important at the time. Much of the recent progress in fields such as genetics, evolution, developmental biology, and ecology has been integrative, in part because many phenomena are now known to have multiple causes. Thus there is a need to formulate a synthesis of ideas that will extend our understanding to processes that are simultaneously rooted in each of these diverse disciplines. These integrative approaches have necessarily made many people more receptive to a study of the interactions involved in aging. The people researching the various diseases have come to realize that an underlying risk factor (cause?) for many illnesses is age. This synthesis has been facilitated by the enormous accumulation of empirical data concerning aging, as well as the organization of those data into meaningful and accessible review articles, symposia, and (most important in the long run) computer-based relational databases. As a consequence, the divisions between gerontology and the rest of biology are gradually being blurred by the realization that all parts of the life cycle are continuous with one another in process and mechanism, if not in detail.

1.3 Defining Aging and Senescence

As the interest in the biology of aging spreads outward, attracting scientists from other disciplines, it also spreads downward, engaging the interest of people other than practicing research biologists and gerontologists. Numerous best-selling how-to books, by combining an incomplete description of human aging with a favorite set of putative interventions, have made their authors wealthy and their readers believers. This book will do neither. But it may contribute to an increased understanding of the questions we face and an appreciation of what we know and what we don’t know. This book then is written for the biology student who wishes to learn the essentials of the subject, for the clinically or social science oriented gerontologist who wishes to learn about the mechanisms that count our days, and for any who are interested in how we can intervene in the process.

1.3 Defining Aging and Senescence The fact that I have used the terms “aging” and “senescence” thus far without defining them suggests that these familiar words have a universal definition. They are familiar, but they are also imprecise in that they may mean different things to different people. Different authorities use different words. Costa and McCrae (1995, p. 25) take the broad view and define aging as “what happens to an organism over time.” Their reason for adopting such an all-inclusive definition is to draw our attention as much to functions that are preserved as to those that change. Understanding the mechanisms that underlie stability may provide insight into the processes that promote loss of function. This is a good point. But this broad view does not allow us to distinguish aging from anything else that happens to the organism, so it is not useful for our purposes. And, in fact, Kohn (1978) did draw a distinction between developmental changes and age-related changes that justifies the rejection of any all-inclusive definition: “By teleological criteria, development can be viewed as consisting of early processes that enhance the functional capacities of a system, whereas aging consists of later processes that di-

9

minish or have no effects on ability to function” (p. 10). I provide evidence for this distinction in the discussion of mortality kinetics in chapter 2. Comfort (1960, p. 8) proposed that aging is “an increased liability to die, or an increasing loss of vigour, with increasing chronological age, or with the passage of the life cycle.” In a similar vein, Maynard Smith (1962, p. 115) defined aging processes as “those which render individuals more susceptible as they grow older to the various factors, intrinsic or extrinsic, which may cause death.” Frolkis (1982, p. 4) wrote, “Aging is a naturally developing biological process which limits the adaptive possibilities of an organism, increases the likelihood of death, reduces the life span and promotes age pathology.” And Rothstein (1982, p. 2) stated that “the changes from maturity through senescence constitute the ‘aging’ process.” These different definitions give the initial impression that they each describe the same phenomenon, albeit in different words. Does the similarity of words imply that the underlying concept is accurate? Strehler (1982) tried to formulate an answer to this question. He pointed out that aging is not simply the sum of the aggregate pathologies and of disease-induced damage, and that, conversely, not all the changes in structure and function that are correlated with age may be appropriately considered as fundamental agerelated changes per se. These two concepts, unlike the preceding definitions, impose limits on what we may regard as the fundamental aging processes. In an effort to incorporate this rigor into an operational definition, Strehler (1982) suggested that fundamental age-related changes must meet the following four conditions: 1. They must be deleterious; that is, they must reduce function. 2. They must be progressive; that is, they must take place gradually. 3. They must be intrinsic; that is, they must not be the result of a modifiable environmental agent. 4. They must be universal; that is, all members of a species should show such gradual deficit with advancing age.

10 Chapter 1 Perspectives on Aging For a long time these criteria were thought to define aging processes adequately and to allow us to distinguish them operationally from non-aging phenomena such as diseases and accidents. As new data have developed, however, it has become clear that the concept of universality is the Achilles’ heel of this definition. Chapter 3 presents these data; for now, it will suffice to say that there is so much individual variation in aging due in part to our genetic heterogeneity and in part to chance alone (Finch and Kirkwood 2000) that it is not possible to talk of all members of a species aging in an identical manner. The concept of intrinsic change, while valid, is being narrowed as we appreciate how much various lifestyle practices modulate what were once thought to be completely intrinsic events. Nonetheless, Strehler’s concept of deleterious, progressive, and intrinsic changes is still useful today. More recently, Masoro (1995a, p. 3) proposed that aging refers to the “deteriorative changes with time during postmaturational life that underlie an increasing vulnerability to challenge, thereby decreasing the ability of the organism to survive.” This definition is similar to Strehler’s, but not all would agree with the inclusion of time in a definition of aging. The role of time in aging is worth discussing, particularly since chapter 3 presents evidence suggesting that physiological biomarkers are a much more useful index of aging than is the simple passage of time. Finch (1990, p. 5) points out that “aging” is generally used to describe a host of time-related alterations that biological entities from molecules to ecosystems undergo. Is there a theoretical or empirical reason to assume that time itself plays a causal role in the progression of an organism from birth to death? Biological time is measured by interlocking cascades of specific physical or chemical events, and the underlying mechanisms are now understood in some detail. For example, the biological clock in the fruit fly Drosophila, the mold Neurospora, and presumably other organisms depends on the cyclic interaction between at least two specific gene products, which then unstably repress their own transcription on exposure to light and thus provide the rhythmic

circadian output characteristic of a biological clock (Gekakis et al. 1995; S. A. Kay and Millar 1995; Sehgal et al. 1995). The fundamental units of the biological clock are transcription cycles. The fundamental units of the sidereal clock are day and night, which are based on the relationship of Earth to the sun. The involvement of light connects these two otherwise disparate clocks and loosely connects time to aging. Thus, although it is customary to view aging as a time-based process, this approach is flawed if only because the translation of chronological time or planetary rhythms into biological rhythms subjects it to myriad biological controls, which act to interpret it differently for each organism. We each know individuals who are the same chronological age but appear to be very different physiological ages. Physiological age is not a simple time-dependent phenomenon. Something is missing. As Arking and Dudas (1989) noted, one indication of a more sophisticated understanding of aging would be our ability to remove time from the analysis of aging, since time is only an imperfect correlate of the currently unknown physiological processes involved in aging. Only when we can substitute the operation of the actual physiological mechanisms for time will we have a firm idea of what we’re talking about. In other words, we need to make time an independent rather than a dependent variable in our analyses. Instead of using the calendar to measure aging, we need to be able to use the changes in important physiological variables to measure aging. This goal was initially accomplished in the studies of Finch (1988) on the neuroendocrine control of reproductive aging in the mouse (discussed in chapter 13) when he showed that the onset of menopause was a hormone dose-dependent process and not a timedependent phenomenon. This relationship was later demonstrated in human physiological studies (see Manton et al., 1995, and the related discussion in chapters 2 and 6). More recently, this finding has been widely demonstrated in many experiments using the high-throughput techniques of microarray based gene expression analysis (discussed in chapters 7 and 14). This technique will likely allow us to track the changes

1.3 Defining Aging and Senescence

in expression of each of the thousands of genes involved in the aging process. Although we do not yet understand everything that is happening in these microarrays, it is clear that aging cells, tissues, and animals can be characterized by their gene expression patterns. Since the gene expression changes likely arise as a direct result of changes in the organism’s internal environment, including the effects of preceding genetic changes, then it follows that the gene expression changes are not dependent on the passage of time but rather on the inputs from these other variables. Aging has its molecular signature, and it is not tied to a sidereal clock or a calendar. Because much, perhaps most, of the data I examine in this book is traditionally presented in terms of time and age, we cannot realistically remove time from our discussions. But we should be cognizant that the genetic and physiological changes that allow our bodies’ foundations to crumble are not time-dependent but show at best an imperfect and misleading correlation with the clock and the calendar. As a result of this survey, we may define aging as the time-independent series of cumulative, progressive, intrinsic, and deleterious functional and structural changes that usually begin to manifest themselves at reproductive maturity and eventually culminate in death. A simple mnemonic for this definition is CPID (cumulative, progressive, intrinsic, deleterious). Having defined aging, how can we measure it? It is important to be able to measure aging, for otherwise we would not know whether an organism was aging faster or slower than another, nor would we know if any anti-aging intervention was working. In practice, we use two different measurement standards, depending on whether we are measuring aging in a population or in an individual. The rationale and technical details of each metric are presented in chapters 2 and 3, respectively. For now, just note that we measure aging in populations by using the observed agespecific mortality rates to calculate the number of surviving organisms that will likely die in the next time period. If the number of deaths increases, then the population is composed of aging individuals and may be considered an “aging

11

population.” The rate of aging can be computed based on how long it takes for the mortality rate to double. Humans are a slow-aging species, and our mortality rate doubling time is about 8 years. In individuals, however, we measure aging by measuring changes in physiological traits, or biomarkers, known to be important to normal functioning and capable of predicting remaining longevity. If the observed changes are altered in the direction of loss of function, then the individual is aging at some calculable rate of loss of function. It is highly likely, although not yet proven, that the microarray-based gene expression and proteomic patterns will one day be able to serve as a molecular marker of an individual’s aging status (see chapter 7). They already can serve as a reliable indicator by which to distinguish young from old organisms or healthy from diseased ones (Bronikowski et al. 2003; Pletcher et al. 2002; Tan et al. 2002). Eventually, we would like to be able to map senescence onto physiological processes, onto molecular composition, onto gene expression patterns. It is in such a manner that the population and individual measurements will converge on one another. If the changes in these biomarkers and/or gene expression patterns have been previously correlated with population longevity, then knowing the rate of change in any single individual’s biomarker values should allow one to calculate the probability that the person will survive to some specified age. Evidence presented in chapter 3 suggests that this convergence is now beginning to occur. Senescence is the other word we need to define. Although “senescence” is often used interchangeably with “aging,” Lamb (1977, p. 2) suggests that “senescence” and “senescent” should be reserved for instances “when talking about the changes which occur during the period of obvious functional decline in the later years of an animal’s life-span.” This usage is in agreement with the earlier suggestion of Strehler (1982 p. 11), who defined senescence as “the changes which occur generally in the post-reproductive period and which results in a decreased survival capacity on the part of the individual organisms.” Thus, senescent changes are those that

12 Chapter 1 Perspectives on Aging most noticeably occur during the latter part of the life cycle and that are somehow associated with the increased mortality characteristic of the last stage of life. I pointed out that microarray data gave aging a molecular signature and freed it from time. Some of the observed gene expression changes may be the (or a) cause of some physiological alteration; other gene expression changes may be the consequence of some preceding environmental shift. It may never be possible to draw an unambiguous causal connection linking every important physiological change to every observed alteration in the gene expression patterns. The causal connections are probably too interconnected in complex circuits to allow such a simplistic conclusion. But the causal nature of the gene expression patterns does not really matter. What does matter is that the gene expression patterns for a given organism in a given environment be recognizable, repeatable, and bear at least a correlative relationship to the physiological states characteristic of senescence. If such conditions really do apply (and the experimental data to date certainly suggest such conditions do apply), then aging will have its series of molecular signatures, and we will someday be able to use those signatures to measure the progress of aging in individuals. We can now define senescence as those processes that bring about the changes in an organism’s gene expression patterns and/or physiological biomarkers from those known to be consistent with health and somatic maintenance to those patterns consistent with aging and failure to maintain oneself. If aging is a series of increasingly different and less functional molecular and physiological signatures, then senescence comprises the process that is responsible for the changes in those signatures. I have engaged in a rather long and thorough derivation of the definitions of aging and senescence. These are key terms in the biology of aging. If we satisfy ourselves with imprecise or sloppy definitions, then our understanding of the field will suffer, for we will never know exactly what it is we are talking about. A terminology committee under the aegis of the Gerontology Society of

American is developing precise definitions of these and other terms used in gerontology. Before reading further, make sure you understand the reasoning behind the definitions given here. Some of the concepts contained in the definitions presented here are worth emphasizing: 1. Not all time-dependent changes should automatically be considered fundamental agerelated changes. Time should be a dependent variable. 2. Age-related changes usually manifest themselves beginning at reproductive maturity, although their genesis may have been earlier. 3. Age-related changes are cumulative, progressive, intrinsic, and deleterious (CPID). There is so much individual variability that it is difficult, if not impossible, to conceive of any one aging pattern as being the normative pattern for every individual member of the species. 4. The death of the individual organism is the ultimate end point of aging. For the individual, it is a sudden and acute transformation from one state to another; yet the process of aging involves a progressive increase in the probability of dying within a population of individuals. Different measurement methods are used to measure aging in populations and in individuals. 5. Aging and senescence are fundamental and intrinsic properties of most living organisms. Comparative studies provide valuable insights into which mechanisms are specific and which are conserved. 6. Aging can be described as consisting of the progression through some series of different biomarker values and/or gene expression patterns which mark the transition from a state of high bodily maintenance and normal functioning to a state of low bodily maintenance and increasingly abnormal functioning. Aging has its molecular signatures. 7. The changes in these biomarker and/or gene expression patterns are brought about by senescent processes of various types.

1.5 Is Aging a Universal Trait?

Using the points emphasized above as a working definition of aging or senescence has the advantage of allowing us to be precise in categorizing a particular process as a normal age-related change. For example, we can easily distinguish deleterious changes due to aging from changes due to infectious disease (the latter is the result of a parasite and is not intrinsic), or from changes that have no obvious deleterious effect (for example, gray hair). However, this precision does not come without a price. Rigid adherence to typological thinking might cause us to reject any age-dependent physiological or genetic change that does not occur in all individuals. This would be a serious error, because the underlying assumption is contradicted by the data (see chapter 3). Age-related changes are not universal within a species; different individuals may age differently; this is likely due to differences in the senescent processes at play in each individual. Probably the best approach is to use the CPID criteria as a general guide and to resolve questionable cases on the basis of the evidence available. Some internal contradictions may result, but consistency is not the highest virtue.

1.4 Two Conceptual Models of Aging Since the study of aging has its origins in medicine, it is not surprising that the initial views of aging made the assumption that all deaths are attributable to either overt or covert disease. Implicit in this viewpoint is the assumption that the elimination of disease will result in an increase in life expectancy. Because this is basically just what occurred in the 20th century, then many people concluded that this “medical model” of aging was correct. But during the last half of the century, another point of view, the “biological model” of aging, arose, which disputed this older interpretation. The main evidence for the disagreement was the observations that first, the increased life span in the 20th century involved an increase in the mean life span but not in the maximum life span (see chapter 2); and second, even healthy and disease-free organisms aged.

13

These criticisms, backed up by experimental data, led to the conclusion that aging is not the outcome of disease or pathological processes, but rather results from the evolutionary tendency of organisms to allocate more energy to reproduction than to somatic (self ) maintenance and disease (see chapter 4). Disease and aging are considered in the biological model to be related but different processes, and this difference distinguishes biogerontology from geriatrics. The data in this book are much more consistent with the biological model of aging than with the medical model, and the reader is encouraged to draw his or her own conclusions after surveying all the data.

1.5 Is Aging a Universal Trait? Is there such a thing as a non-aging system? In a fundamental sense, such a system cannot exist, for cosmologists generally agree that our universe (probably) and our solar system (certainly) have a finite life span. At the other end of the scale physicists agree that most subatomic particles (perhaps even the proton)—decay. If both the universe and its component particles age, then so must all the intermediate organizational levels. Nothing is forever, for there is no forever. Despite this exercise in logic, it is reasonable to ask if, on a more familiar time scale, non-aging systems do exist. Non-aging systems would be systems that, when periodically examined, exhibited no changes. Are there any? Kohn (1979) claimed that nonaging systems do exist, and that they are always composed of dynamic processes. The simplest type, a chemical system at equilibrium, can be depicted as follows: A+B
C+D The chemicals A and B interact to yield products C and D; similarly, the products C and D may interact to yield A and B. The reaction will proceed until predictable amounts of the four chemicals are present. The amounts depend only on the initial concentrations and the amount of free

14 Chapter 1 Perspectives on Aging energy available. Once the reaction reaches this equilibrium point, it will stay there forever, provided that no work is done on the system and that environmental conditions remain constant. Given the stringency of the conditions, it is understandable that such systems are not found in nature, remaining only laboratory curiosities. One type of non-aging process that is common in nature is a steady-state process, such as the one depicted in figure 1.2. In this case we have a series of several different sequential reactions, each of which may be reversible. The reaction as a whole is driven in one direction by the continuous addition of component F and the compensating continuous removal of component E. When the net flows of E and F are balanced and the rates of the various reactions are steady, the amounts of components A, B, C, and D will not change with time. This is a non-aging system. It could be converted to an aging system by a progressive and irreversible change in the rates of utilization of F and/or the production of E. It may not appear sensible to talk about agerelated or non–age-related changes in a steadystate system because the system contains different populations of molecules at different times. Since the system is continually being renewed, what is there to age or not to age? The answer is that the identity of the system is independent of the turnover of its components; rather, it depends on the interactions of the components. A more familiar physiological counterpart of the steady state is illustrated by the homeostatic process depicted in figure 1.3. The fasting blood glucose levels in human males did not change during the 2-hour period of examination, although the number of both glucose molecules and insulin molecules in the systems increased substantially. This is a nonaging system in which the identity and the numbers of the insulin and glucose molecules change from one moment to the next, but changes in one component evoke changes in the other such that the variable affected—in this case the blood glucose level—remains constant. The measured levels of molecules usually do not remain identical from one measurement to the next even in this non-aging system. The regu-

E

D

A

C

B

F

Figure 1.2 A steady-state process in which the concentration of components A, B, C, and D will not change as long as the inflow of F and the outflow of E are balanced. In the absence of change, this constitutes a nonaging system. (After Kohn, 1977.)

Figure 1.3 Measurements from an actual homeostatic process: the maintenance of glucose levels in the blood. Plasma glucose levels stay almost constant for 2 hours in healthy volunteers during the hyperglycemic glucose clamp technique (a method for quantifying insulin secretion and resistance), despite large alterations in the glucose infusion rates, which cause corresponding alterations in the plasma insulin levels. (After DeFronzo et al. 1979.)

1.6 Measuring Age-related Changes

latory components of the system consistently undershoot or overshoot the optimal value, leading to a long-term fluctuation about the optimal level. Statistics represents this situation by showing the mean and variance, as in all three curves of figure 1.3. However, if the interactions of the regulatory subcomponents change, the system might be transformed into a non–steady-state system characterized by a progressive alteration in the mean level and/or in the variance. This phenomenon is nicely illustrated by the data contained in many of the figures found throughout this book. Now let’s answer the question we posed at the start of this section: Is aging a universal biological trait? The answer is simple: It is widespread but not universal. There is some evidence suggesting that aging as we have defined it occurs in at least some bacterial species, but strong and broad evidence of aging is found only among eukaryotes (see chapter 4). Even then, aging is not found in individuals of all species, and not in the same manner among those that do age. A bewildering array of lifestyles and life spans confronts the investigator who seeks broad generalizations, from mayflies that live 1 day to shrubs that live 11,000 years or more. Finch (1990) has sorted this cacophony into three classes: (1) organisms that show no or negligible signs of aging, (2) organisms that display a gradual progression of aging, and (3) organisms that exhibit a rapid onset of aging. I deal with these classes in more detail in the discussion of plasticity later in this chapter, as well as in chapter 4 and at other points throughout our tour of the topic. However, we should keep in mind the nature of the mechanism(s) that distinguish these three groups, and we should consider whether the differences between them are qualitative or quantitative. These three categories crosscut and otherwise disregard the common phylogenetic and evolutionary relationships—a phenomenon that a complete explanation of aging must encompass. If evolution is the fundamental theorem of biology, what enables a trait as fundamental as aging to escape being phylogenetically constrained? I answer this question in chapter 4.

15

1.6 Measuring Age-related Changes Aging is a deteriorative process that manifests in two distinct ways. First, aging increases the probability with time that the individual will die. Second, aging decreases the ability of an individual to withstand extrinsic stresses. It follows, then, that either the timing of death or the age-related decrease in functional properties may be used to measure the occurrence of age-related changes. Death is a singular and acute event in an individual’s life span. Simply knowing its chronological time for one individual gives us no useful information with which to determine the rate of aging of that individual. Knowing the times of death (or the lengths of the life spans) for numerous individuals raised under similar conditions will allow us to determine whether the probability of any single individual’s dying is constant. A constant probability throughout the time period studied implies that the chance an animal will die in any given period is not related to the age of the animal. Given this observation, we must conclude that age-related changes are not taking place and that the deaths are probably a result of accidental, stochastic causes. This mortality pattern results in the survival curve depicted in figure 1.4a. In contrast, an increase in the probability of dying as the animal grows older, is empirical evidence that age-related changes are taking place. This mortality pattern results in the survival curve shown in figure 1.4b, and in a clustering of the ages at death about the values of the mean life span, as shown in figure 1.4c. The determination of the presence or absence of age-related changes within individual animals is based on an assessment of population data. Such data are often presented in the form of two-dimensional survival curves (fig. 1.4). I deal with the measurement of aging in more detail in chapter 3. For now, keep in mind that these two-dimensional plots abstract the actual information, and in the process they both highlight and obscure various kinds of information. The information obscured in the plots of figure 1.4 has to do with the individuality of the rates of aging. It is all too easy to assume

16 Chapter 1 Perspectives on Aging

(a)

(b)

Number of individuals

Age (c)

100

Percent surviving

Percent surviving

100

Mean life span

50

Age

Mean life span

Age at end point

Figure 1.4 Aging and the probability of dying. (a) Percentage of individuals surviving as a function of time for a population in which the probability of dying remains constant with time. (b) Percentage of individuals surviving as a function of time for a population in which the probability of dying does not remain constant with time. (c) The age at death of the individuals in the population depicted in (b). The number of deaths is low at first and increases to a maximum value late in life. (After Kohn 1977.)

that every member of the population loses its vigor in a constant manner, as in figure 1.5. The real situation is much more complex; figure 1.5b and c attempt to identify some of these complexities. Would it be more beneficial, then, to study the age-related decrements that take place within the life span of one individual and to determine their importance by comparing them with the mean population values? This has been the goal of the longitudinal studies undertaken by Shock and colleagues (1984). This type of study design is very powerful and has contributed much valuable information that otherwise could not have been obtained. There are two problems here. First is the obvious one of determining beforehand which physiological or biochemical parameters are adequate and determining predictive measures of aging. This task is difficult but not insurmountable. The second point is that the longitudinal

studies have led to the conclusion that many individuals do not follow the pattern of age-related changes predicted from the averages based on the summed measurements made on different subjects (see figure 1.5). The differences between individuals become even more pronounced as the individuals age. Aging is so highly individual that average curves give only a rough approximation of the pattern of aging followed by individuals. Thus, knowing that a certain individual has a measurable decrement in a particular physiological function may or may not be a sufficient basis for saying anything reliable about that individual’s rate of aging, much less predicting his or her longevity. Nonetheless, the widespread dissatisfaction with using time to measure aging has given impetus to the development of biological markers of aging. Such biomarkers, as they are called, have shown some promise of being able to measure individual rates of aging. I discuss biomarkers in some detail in chapter 3.

1.7 Models for Studying Aging

(a)

(b)

17

A B C D

Vitality

Vitality

Z

Death threshold Death threshold

Z

BD C

(c)

A

Time

Times of death A B C

Vitality

D E F * DE threshold ABC threshold B

F D Time

E CA

Figure 1.5 Representations of aging, based on different assumptions: (a) that all members of the population lose their vigor in a constant manner, and there is a single and consistent death threshold; (b) that individuals age at different rates, and there is a single but variable death threshold; (c) that individuals age at different rates, and there are multiple and variable death thresholds. All values are arbitrary. (Panels a and b after Lamb 1977.)

1.7 Models for Studying Aging 1.7.1 The Choice of an Organizational Level Age-related changes may be measured on a population level, on an individual basis, or on a cellular or even subcellular level. Information that may be correct and valuable at one organizational level may be meaningless at another level of complexity. For example, one school of thought views organismal aging as due to autonomous changes taking place in individual cells. Consequently, much effort has focused on deciphering the age-related changes

taking place in individual cells both in vitro and in vivo. One can describe and catalogue these agerelated changes that take place in a short-lived cell, such as a fibroblast or an intestinal cell. Does this knowledge then assist us in understanding the biology of a long-lived cell, such as a neuron? What is the relationship between the longevity and senescence of any component level, such as molecules, organelles, cells, tissues, or organs, and the longevity and senescence of the whole organism? Will the knowledge obtained from any single organizational level be sufficient to provide us with a model of aging in the whole animal? The answers to such questions will probably be constrained by the following three observations.

18 Chapter 1 Perspectives on Aging First, mere turnover of a component, whether molecules or cells, does not necessarily constitute an age-related change as defined here. Second, transferring a term from one organizational level to another often results in semantic confusion. For example, we talk of the life span of cells. At the organismal level, “life span” is an unambiguous term because it has a well-defined end point— death. Yet not all cells are destined to die before the organism does. Many cells (such as neurons) exist as viable functioning entities throughout the life span of the organism. Other cells end their individual existence by dividing mitotically into two daughter cells, which will themselves undergo mitosis later. Does the term “life span” refer to the individual cells or to the whole clonal line of cells? If, for example, we choose the former alternative, then we are equating organismal death with mitotic reproduction. If we choose the latter definition, then what does this definition mean when the organism dies even though the clonal lines of cells of which it was composed are still alive at the point of death? Forcing the identity appears to lead us to equate very different processes. And finally, are the age-related changes observed at one organizational level intrinsic and autonomous, or are they a secondary consequence of deteriorative changes occurring elsewhere in the body? Despite these cautions, it must be said that the senescence of the whole animal ultimately must be caused by changes in the lower levels of organization. I pragmatically adopt a reductionist approach, believing that only by studying aging at the lower levels of organization will we ever be able to understand the nature and causes of the age-dependent declines in the survivability of the whole organism. But we must not be too enthusiastic in this approach, lest we become too simplistic in our interpretations.

1.7.2 Choosing an Experimental Organism The choice of an experimental organism is dictated both by pragmatic considerations and by one’s perception of the goal of gerontology. If one perceives the goal to be the study and understand-

ing of human aging only, then the value of any proposed experimental organism will be in direct proportion to its biological similarity to humans. The more distant the evolutionary relatedness between us and them, the less useful are the lessons learned and thus the less desirable they are as models of aging. The use of less desirable organisms might be justified on the basis of pragmatic considerations of cost or time or because of their special utility in answering a particular question, but they are a poor substitute for humans in such studies. Another approach views the study of aging as leading to the eventual understanding of a fundamental biological phenomenon, one that is an intrinsic evolutionary part of the life history of most organisms. In this viewpoint, aging is as worthy of study in its own right as is developmental biology. Much study has shown that there is no more reason to believe that the underlying mechanisms of aging must be the same in all organisms than there is reason to believe that all organisms develop in exactly the same way. At both the anatomical and the biochemical level, the same functional requirements may be met in very different ways. Thus the information obtained from invertebrates may not be directly applicable to vertebrates in general or to humans in particular. However, fundamental biological processes are usually quite similar to one another in different organisms, once one makes allowances for the structural and/or functional differences in the different systems. Thus the study of different organisms wisely chosen is likely to be of great value in identifying the diverse mechanisms that underlie aging and senescence in our own species. No single animal or plant model is the best one in which to study aging processes. An investigator’s choice of an organism depends on various criteria, not the least of which is the nature of the question being asked (e.g., see Masoro 1999). For example, a genetic analysis of the mechanisms of aging almost by necessity focuses the choice on organisms, such as Saccharomyces cerevisae (yeast), Drosophila melanogaster (fruit fly), Caenorhabditis elegans (nematode worm), or Mus musculus (laboratory mouse), that are well suited by virtue of the genetic tools and

1.7 Models for Studying Aging

knowledge that has been deliberately accumulated about them over the years. Physiological questions might be more easily investigated using rats or mice. Other experimental goals might be (1) describing the changes with aging within one or more of these species, (2) describing and analyzing the causes of longevity within or between species, or (3) describing and analyzing models of accelerated or decelerated aging (Weindruch 1995a). Each of these goals would impose its own constraints on the choice of the organism to study. Birds, for example, live much longer than mammals of comparable size (Holmes and Austad 1995a,b); hence they might serve as one component of the comparisons referred to in the second or third options outlined here. There are pros and cons for every choice. The April 2004 issue of Aging Cell presents a spirited debate on the utility of using short-lived animal models for the study of aging in humans (http://www.blackwellsynergy.com/tcc/ace/3/2). An analysis by Weindruch (1995a) of animal usage patterns in gerontology studies from 1972 to 1992 reveals that most of the 2476 reports sampled studied either rats (48%) or mice (28%). Drosophila, the third most popular organism, accounted for only 5.3% of the studies. Another large gap separates this number from the percentage of studies done on all other organisms (a total of 18.7%, which include nematodes, houseflies, rabbits, hamsters, fish, protozoans, birds, dogs, nonhuman primates, other insects, lizards, cows, and a herd of other rarely studied beasts, at frequencies ranging from 0.2 to 2.0%. During the past decade, there has likely been an increase in the proportion of studies done with nematodes and flies, as will be apparent in chapter 7. Nonetheless, there is an obvious risk that our knowledge will be overly dependent on aging processes easily studied in rodents and that consequently we will miss an important insight that might be best obtained in another, rarely studied animal model. The situation is even worse than one might imagine, since half the rat studies were based exclusively on the use of male Fisher 344 (F344) rats. The dangers of this overreliance on one animal strain are well illustrated by the fact that the older F344 rat develops

19

nephropathy (a kidney defect) when fed a diet containing casein as the protein source but not when fed a soy protein diet or a calorie-restricted casein diet. In the absence of the latter two facts, it would have been (and was) easy to conclude erroneously that kidney failure is a normal component of aging in rats. Only by comparison of that conclusion with data taken from other strains was the error recognized and corrected. Fortunately, the National Institute on Aging quickly recognized the need to develop and support rat and mouse strains well suited for research on aging, along with developing sophisticated genetic techniques to best probe aging mechanisms in these animals (Harrison and Roderick, 1997; Sprott, 1997). Similar traps await those who base their knowledge of aging solely on human geriatric studies. A comparative biological approach to the study of aging and senescence is the most prudent and conservative course available to us. Approaching this field of inquiry with a less anthropomorphic point of view may not tell us directly how to manipulate the human life span, but it does promise to be the most efficient approach for showing us which physiological systems and which organizational levels we should investigate. This comparative strategy has borne fruit over the past decade. A more or less rational system of model organisms (laboratory species which preferentially lend themselves to a genetic and functional genomic analysis of their aging processes) has evolved. Such organisms are not only amenable to classic genetic tools, but their genomes have now been sequenced, and tools and procedures have been specifically developed for each species, which enables researchers to use the high through-put techniques necessary for simultaneously analyzing multiple genes or proteins or other interesting molecules. The model organisms now in play include the yeast Saccharomyces cerevsiae, the nematode worm Caenorhabditis elegans, the fruit fly Drosophila melanogaster, and the mouse Mus musculus. To this list we may add certain long-lived birds such as pigeons (Columbus livia) or budgies (Melopsittacus undulatus), the attraction of which is that in certain ways they age even slower than long-lived mammals such as humans. Organisms such as

20 Chapter 1 Perspectives on Aging torotoises, lobsters, and deep-sea fish age so slowly that they are considered to have negligible senesence; they too are being viewed as potential model organisms. We can also add to this list rhesus monkeys (Macaca mulatta), which are primates and thus very useful in a limited way for translating findings from these other organisms to animals very similar to us. It is likely that any putative human anti-aging interventions will need to be tested first on rhesus monkeys, and it will be essential to have a good knowledge of their aging pattern. And, perhaps surprisingly, we should also add Homo sapiens to this list of model organisms. Humans obviously are not being experimented upon in the laboratory (at least not without their informed consent), but the focused analysis of extraordinarily long-lived or short-lived groups of humans, the growth in vitro of human cell lines, and the analysis in silico of the human genome allow us to rather quickly determine which of the mechanisms and processes identified in the model organisms are also present in humans. This knowledge assists in identifying phylogenetically conserved aging mechanisms as well as in enabling the translation of information from simpler organisms to complex ones. (The longevity determination and senescent processes taking place in these six model species are discussed in more detail in chapters 4–14.) Finally, Rueppell et al. (2004) have argued that the social insects (ants, bees, etc.) should be included as model organisms because their division of labor into reproductive and nonreproductive individuals is analogous to our body’s organization into reproductive cells (germ cells, mitotic cells) and nonreproductive cells (postmitotic cells). The social insects offer the opportunity to experimentally alter selected variables in different individuals and observe the growth and demographic changes that ensue, a task that is difficult to do with human cells in culture. In addition, the study of social insects will provide the opportunity to discover and characterize aging mechanisms operative over a wide range of different demographic and social configurations. These classic laboratory model organisms are well suited for genetic and physiological investi-

gations of the biological mechanisms of aging and longevity. But one central human life-history characteristic missing in most of the classic organisms is sociality, in the sense that laboratory animals do not live in complex age-structured societies. A complete understanding of human longevity must include the role of our social structures and the identification of socially dependent mechanisms of human aging, which I discuss in chapters 14 and 15.

1.7.3 Problems Peculiar to Gerontology Regardless of the species chosen for study, many animals of good quality are needed for studies on aging. “Good quality” is a nebulous term that may mean different things to different investigators. Certainly, investigators should comply with all applicable regulations, not merely because of legal implications but because any investigator worth his or her salt would want to use healthy animals so as not to confound aging with disease. The details of animal husbandry will depend on whether the animals must be germ-free, free of a specified pathogen, and so on. Once these criteria are decided, the next decision involves numbers. Aged animals are, by definition, survivors. If an investigator devises an experiment that requires applying some anti-aging intervention to young rodents, then he or she must be prepared to set up a large starting population of young animals and pay the cost of rearing the colony for 2 or 3 years. If another researcher wanted to use 10 very old rats in some other experiment, the same financial costs would apply. The prices could easily reach several hundred dollars per animal. Old rats and mice of defined strains can be bought at reasonable prices only because of the collections developed and subsidized by the National Institute on Aging (Sprott 1991). The same problems arise regardless of the species being examined; old flies are proportionately expensive as well. The age of the animals illustrates two of the major problems confronting gerontology research: the need to plan experiments months or even years ahead, and the extraordinarily high cost of aged animals.

1.8 The Plasticity of Aging

For any animal model, then, the life span should be short, as a matter of convenience and expense. Yet the short life span should not be due to the animals dying prematurely of an infectious disease or other preventable pathology. And we should not let the convenience of a short life span cause us not to investigate the mechanisms promoting a long life span. In fact, since we humans are, despite our lamentations to the contrary, a very long-lived species, then we really need to identify those factors that have made us thus (see chapter 14). In addition, the environmental conditions necessary for an optimal life span should be known and defined over the entire life span. Failure to control these conditions, as in the case of the F344 rats fed casein, may lead to incorrect conclusions. There is no substitute for animal experimentation in biogerontology. Because we do not know the biological mechanisms that underlie the aging process, clearly we cannot use substitutes such as computers, which can show us complex interactions but cannot synthesize missing knowledge. Only animals studies can do that. Computer simulations are proving to be very valuable, but as an adjunct and companion and guide to animal studies, not as a replacement for them.

1.8 The Plasticity of Aging 1.8.1 Intraspecific Plasticity The longevity of an organism is a phenotype. That is, it is one of the observable properties of an organism produced by the interaction between the organism’s genetic potential (its genotype) and its environment. The life span can be affected not only by changes in the genotype alone or in the environment alone, but also by changes in the manner in which these two variables interact. This change in the expressed phenotype of a genotype as a function of the environment is called phenotypic plasticity (Scheiner 1993). Mice and flies provide examples of this phenomenon. One might expect inbred mice (which are so genetically similar that they can accept skin grafts from one another) raised in a constant and

21

defined environment to exhibit identical life spans. But they don’t; there is always a significant variance about the mean (Ghirardi et al. 1995; Witten 1994). In this case, we are forced to conclude that neither the genetic nor the environmental factors were completely controlled. Something is missing. That missing factor may well be the role of chance. Random or stochastic events that inevitably happen during development may well account for these observed differences in ostensibly identical animals (Finch and Kirkwood 2000), and I discuss these events in terms of reproductive aging in chapter 13. Members of a species may be able to express several different longevity patterns, depending on circumstances. But any individual organism can only express one particular life span pattern. Clearly, the nature of the environmental signals inducing the animal to express one of several possible longevities is another important variable. As shown in figure 15.3, three different environmental pressures induce three qualitatively different patterns of extended longevity in the same normal-lived strain of fruit flies. The animal contains the mechanisms necessary to bring about at least three different extended longevity phenotypes; the nature of the environmental signal specifies which one is actually expressed. Both chance and signal specificity contribute to the plasticity of the longevity phenotype. An example of phenotypic plasticity in which we have at least partly defined one of the environmental parameters and its interaction with a particular genotype is the case of the F344 rat and its diet, as described earlier. The environmental signal is caesin, which when present in the diet has an effect on renal tissue such that almost all the rats will develop neuropathy by 27 months of age. This premature mortality, and its concomitant effect on the life span of the population, is due to the interaction between the F344 genome and this specific dietary component. If it were a simple interaction, all the rats would show the phenotype of renal neuropathy at the same age. The fact that there is some variability in the incidence and age of onset suggests that the interactions between the genetic and environmental factors are complex and that we are far from precisely defining each of them.

22 Chapter 1 Perspectives on Aging Two methods of dealing effectively with a multitude of complex and ill-defined interactions are to employ statistical analyses (Scheiner 1993) or to use computer modeling or simulation studies (Kowald and Kirkwood 1996; Witten 1992). Both caloric restriction and ambient temperature are examples of stringent and defined environmental variables that bring about widespread and partly defined changes in the patterns of gene expression within the organism and in all levels of the longevity phenotype expressed by the affected organism (see chapters 6 and 7 for a more detailed discussion). Different genomes often respond in unique ways to each of these variables. Chapter 3 offers a more detailed description of particular environmental effects. The life span phenotype of an organism, or of a population, is modulated—often quite significantly—by environmental factors. The phenotypic plasticity, as defined here, arises from that common observation. Such plasticity rests, at least in part, on two different types of genetic effects (Via et al. 1995). First, some alleles may be expressed differentially in particular environments, with varying effects on the phenotype. Second, regulatory loci that are sensitive to environmental perturbations may cause other genes to be turned on or off in particular environments. The major result of this interaction between genotype and environment for biogerontology is that we can, even in principle, speak of a particular longevity as being characteristic of the organism only in a defined but limited set of environments (see figure 15.3). There is no single life span for all seasons.

1.8.2 Interspecific Plasticity We tend to judge what we do not yet know by an extension of what is already familiar to us. This method often helps us grasp the unfamiliar, but it can lead us into difficulties. Humans and the various domesticated animals with which we are familiar age in a similar manner. After a developmental period that culminates in sexual maturity, adults maintain physical vigor for a relatively long time before beginning to manifest progres-

sive dysfunctions in various physiological systems over a relatively extended period of time. Despite the substantial differences in absolute life span, the pattern of aging in humans, dogs, cats, horses, mice, and other placental mammals follows this progression. Life spans range from the 1 year of the shrew to the 120 years of the human. But not all organisms age in this familiar manner. What are we to make of the mayfly, which lives but 1 day? Or the Pacific salmon or octopus, each of which spawn once and die? And how are we to understand the senescence of an organism such as the bristlecone pine, which can live as long as 5000 years? Between the mayfly and the bristlecone pine lies an almost incomprehensible millionfold difference in life span. We need to impose some order on nature’s exuberant and untidy range of longevities and thus begin the abstraction necessary to understanding. Finch (1990) has proposed that we characterize senescence by viewing it as a continuum with three general subdivisions according to the observed rate of degenerative change: rapid, gradual, or negligible. This approach has proven to be a most useful organizing principle, and the description that follows is drawn from Finch’s description (1990, pp. 9–10). The first point is that these patterns of senescence are not intended to represent discrete and absolute categories arising from the operation of three different mechanisms. Environmental effects such as temperature or nutrition can shift the rate of senescence of certain species from rapid to gradual, or vice versa. The categories are plastic and depend on the interplay of environmental and developmental factors with the organism’s genome. Rapid senescence is characterized by the rapid onset of major pathophysiological changes at a particular common time after maturation in most or all members of a birth cohort. These changes quickly cause exponential increases in mortality rates (see chapter 2), as well as the death of most members of the cohort within a relatively short period of time, usually a year or less. Thus senescence and death occur almost synchronously throughout the population. Mayflies and other short-lived invertebrates are considered to exhibit rapid senescence. But rapid senescence is evident

1.9 A Conceptual Model for Data on Aging

also in other species, some of which are quite long lived but are characterized by a long developmental or juvenile phase that culminates in a short but intense period of reproduction, after which the organism dies. Examples of such semelparous species (see chapter 4), as organisms that reproduce only once in their life span are called, include the Pacific salmon, the octopus, the marsupial mice, and most species of bamboo. The reproductive fitness (a measure of physiological functioning; see chapter 4) of such species reaches a maximum value once and then vanishes. Gradual senescence, which characterizes almost all placental mammals, is the familiar pattern of aging sketched earlier. One important and diagnostic difference between rapid and gradual senescence patterns is that the latter does not display synchronous senescence and death. Another difference is that the reproductive fitness of organisms displaying gradual senescence generally reaches an early peak or plateau and then gradually decays to zero. Here again, human reproductive patterns can serve as a familiar guide (but see chapter 5). Negligible senescence is assumed to operate in long-lived species for which it has not yet been possible to describe dysfunctional changes. Since no individual is immortal, such senescent changes must take place in all organisms. But given the fact that many of these long-lived species live in habitats that are inconvenient for scientists (e.g., the underwater habitat of lobsters and octopuses) or substantially outlive the scientists studying them (such as sequoias, bristlecones, and other trees), it is not surprising that our knowledge of such species increases only very slowly. However, one indication that we have correctly categorized these species as showing negligible senescence is that many of them show an increase in reproductive fitness as they grow older. This would not happen if they were senescing. There is a website devoted to the understanding of these very longlived species (www.rockfish.org). Finch (1990) has speculated that the tetrapod ancestors had a slow rate of senescence as a primitive trait and that the other patterns represent derived or secondary traits. These three patterns of senescence are not evolutionarily distinct in the

23

sense of being restricted to only certain related phylogenetic groups; rather, they are scattered among a variety of different such groups. The lifehistory characteristics of any particular species— and not its phylogenetic niche—appear to play a deciding role as to the type of senescence pattern the population will display (see chapter 4).

1.9 A Conceptual Model for Data on Aging This book is premised on the demonstrable fact that the most comprehensive explanation of aging is that based on the evolutionary theory of aging (see chapter 4). An organism must reproduce, and it must maintain its body. Longevity and aging flow out of the interaction between these activities. The first parts of this book describe various aspects of longevity measurement and determination, while the explicit discussion of particular senescent mechanisms is dealt with in the last sections of the text. But the myriad data facts and concepts that I describe in this book can be overwhelming and confusing if they are just piled up in a heap, and if we do not have some sort of conceptual framework within which we can sort and hang the various facts until we need them. And so I offer the following bipartite model to fulfill that organizing role for the reader, until you get your own bearings and begin to interpret the data in your own way. It is a reasonable anticipation of the data to surmise that there are a variety of mechanisms that determine how long an organism will live in a more-or-less healthy state. These “longevity determinant mechanisms,” as I call them, are very effective during the first half or so of our life span and act to appropriately integrate reproductive and somatic maintenance processes in such a manner as to keep our bodies at a high level of function. The senescent mechanisms, as defined above, are those processes that are increasingly active during the last half or so of our life span and act to eventually alter physiological and genomic signatures such that our ability to function is degraded, we become less resistant to various

24 Chapter 1 Perspectives on Aging

stresses, and eventually die. The relationship of these two phases is shown in figure 1.6. As we learn more about the topic, we will periodically update this basic diagram (once in chapter 9 and again in chapter 14) until it incorporates all the essential components. Thus, most of the facts I describe will easily fit into one or the other of these two compartments. Those provoking few facts that fit in both or neither should not be discarded but should rather be set nearby and periodically reconsidered. Hans Spemann, a pioneering developmental biologist, supposedly said that we should treasure our exceptions for they may be important clues.

1.10 Sources of Information Many of the statements and data mentioned in this book are referenced to specific articles in the scientific literature, and the reader is encouraged to look up and refer to those particular references which are of interest. Many articles are now online,

and the PubMed web site (http://ncbi.nlm.nih.gov) of the National Library of Medicine is an excellent portal to access this literature. There are quite literally thousands of interesting articles out there. It may be best to start with a recent review article (usually these are so identified on the abstract). The single most inclusive website on the general topic of aging is that operated by the American Association for the Advancement of Science, the major scientific umbrella organization in the United States (sageke.sciencemag .org). The articles are written by both scientists and science writers; thus, the novice need have no fear of being drowned in jargon. The site offers reviews or perspectives on various aspects of aging as well as summaries and links to the current literature. Students are entitled to a highly discounted 6-month membership. A related free website (www.sageke/crossroads) deals with the ethical, social, and political debates now taking place in the larger community, and this site may be particularly informative when reading chapter 15. Finally, I have constructed a website for this textbook on which I will post,

An Integrated Theory of Aging LIFE SPAN

=

HEALTH SPAN

+

SENESCENT SPAN

processes:

Longevity determinant mechanisms

Senescent Degradation of Gene Expression Patterns

based on:

conserved gene based mechanisms

Stochastic breakdown of individual's genes which interact with environment network; influenced by heredity & environment

Figure 1.6 To help understand the aging process, the life span of gradually aging animals such as humans can be viewed as composed of two interdigitating phases. The health span begins at the end of development and ends when the age-dependent mortality rate begins to increase significantly. As an approximation, this occurs at about the time when 10% of the population has died.. The senescent span extends from that point to the time when all members of the cohort are dead from nonaccidental causes. In individuals, the transition time would be when agerelated losses of function begin to exert a noticeable effect on function. The health span is marked primarily by the operation of mechanisms that enhance the health and maintenance of the body. The senescent span is marked primarily by the operation of mechanisms that bring about increasing damage and loss of function. This duality allows us to sort out the data regarding longevity-determinant mechanisms from that regarding senescent mechanisms and to interpret the data in a more coherent manner.

1.10 Sources of Information

by chapter, supplemented material as well as significant new updates and/or interpretations of the material. The address is http://bio.wayne .edu/profhtml/arking/textbook/supplement.html These websites and this book should represent the minimum tools in your learning kit; you should

25

use them so as to expand your sources and your knowledge. I hope you enjoy the process, for the biology of aging is turning out to be at least as fascinating as is the biology of embryonic development, and it will likely prove to have a far larger impact on human life and societies.

26 Chapter 2 Measuring Age-related Changes in Populations

2

Measuring Age-related Changes in Populations

2.1

2.2

Introduction

Life Tables and Survival Curves

Senescence is a deteriorative process. It is difficult to predict the increasing probability that any given individual will die. The estimates we do have are based on the statistical analysis of a population of like organisms. When constructed in the form of a survival curve in which death is the end point, this procedure is informative, although it is subject to all the simplifying assumptions mentioned in the previous chapter. However, the mere fact that a cohort of organisms eventually dies does not necessarily mean that the population underwent aging and senescence. Death does not require aging. All the organisms in a population may have died of accidental causes before any of them had the chance to display senescent changes. Clearly, then, aging and death are not the same thing. All populations die, but not all of them die of age-related causes. We must have a method of reliably distinguishing aging from nonaging populations, if for no other reason than to keep us from wasting our time examining populations that cannot help us to understand the biological bases of aging. The analysis of survival curves will accomplish that task, and in the bargain it will give us some additional useful information about the dynamics of the aging process.

Let’s begin with an example. Assume we have a population of 1000 mature individuals who do not deteriorate in any way as chronological time passes. They are potentially immortal. Let’s assume further that the only causes of death are predation and accidents, and that these random events have an equal chance of happening to a young organism as to an older one. Finally, let’s assume that the predation rate is 20% per year. What would the survival curve for such an odd population look like? What would be the values of the various statistical parameters? Constructing a life table is one way to answer these queries. A life table is a concise and standardized summary of the survival statistics in relation to age and was originally developed to meet the needs of the insurance industry. The tabular format has no theoretical basis but reflects an empirical approach to the measurement of mortality. A survival curve is a graphical representation of the data in a life table. A good detailed discussion of life tables may be found in Carey (1999). As table 2.1 shows, seven different kinds of numerical relationships are found in most life tables, as follows:

26

x

the age interval, with the time units specified by the person constructing the table. The ini-

2.2 Life Tables and Survival Curves

Table 2.1 A Life Table for a Hypothetical Population of Organisms with a Constant Mortality Rate Age 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

the population. As such, it represents the raw data. All other values in the table are derived from this number.

lx

dx

qx

Lx

Tx

ex

1000 800 640 512 410 328 262 210 168 134 107 86 69 55 44 35 28 22 18 14 11 9 7 6 5 4 3 2 1 0

200 160 128 102 82 66 52 42 34 27 21 17 14 11 9 7 6 4 4 3 2 2 1 1 1 1 1 1 1

0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2

900.0 720.0 576.0 460.8 368.6 294.9 235.9 188.7 151.0 120.8 96.6 77.5 62.0 49.5 39.5 31.5 25.5 20.5 16.0 12.7 10.5 8.0 6.5 5.5 4.5 3.5 2.5 2.0 2.0

4990 3990 3190 2550 2038 1628 1300 1038 829 661 527 419 333 265 210 166 130 102 80 62 48 37 28 21 15 10 6 3 1

5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 4.4 4.3 4.1 3.9 3.7 3.3 2.7 2.0 1.3 0.7

Note: these equations work well for situations where the product of (qx)(lx) is a whole number. It makes no biological sense to have fractional deaths. Rounding off these numbers introduces errors into the subsequent calculations, particularly when the population size is small as in days 20 and beyond. Thus the calculated ex is approximately 5.0 until the population numbers become very small. Using simple arithmetic forces us to choose between biological sense or mathematical accuracy. For this population, the time interval from 1 to 19 days illustrates the normal use of the life table.

tial age is represented by x, the following ages by x + t, x + 2t, x + 3t, and so forth, where t is the time interval (days, weeks, months, years) used in the particular case. lx

27

the number of organisms alive at the beginning of each interval. In most but not all cases, this number must be obtained by counting

dx the number of animals dying during each age interval; that is, the number of deaths between age x and age x + t. This number is obtained either by counting or by subtraction of the second census number from the first. If it is obtained by counting, in some cases it could represent the raw data from which the other values are derived. qx the age-specific probability of dying, or the proportion of the animals alive at the beginning of the age interval that die during that interval. This number is obtained as shown by the following equation: qx = dx /lx. Lx the average number of animals alive during the age interval x. An approximate value of this number is obtained by taking the average of the two succeeding time intervals (t), as shown by the equation Lx = lx + (lx + t). Tx the total number of organism-age units to be lived by the total number of organisms alive at the beginning of the age interval. This number may be obtained by summing the values of Lx, as shown by the equation Tx = l(x) + l(x + t) + l(x + 2t) + . . . . ex the mean further expectation of life at the beginning of the age interval x. A close approximation of this value may be obtained by using the value of Tx , as shown in the equation ex ≅ Tx/lx. Note that the accuracy of the estimate of ex depends on the values of x (column 1 of table 2.1) and of qx (column 4), and it increases as these values decrease. In cases where the age interval, x, may be a value other than 1 unit (1 day, 1 year, and so on), then ex ≅

(Tx) (length of age interval) lx

One convention of note is that the number of organisms involved is always adjusted so that the value of lx in a life table is 1000 (or 100,000 for

28 Chapter 2 Measuring Age-related Changes in Populations human populations). In most cases the data on which the life table are based represent only a fraction of this nominal number. Remember that the values in all the columns can be derived from the given values of x and lx. Having briefly described the tabular arrangement of a life table, let’s construct one for our hypothetical population of potentially immortal organisms. Plotting the survival data, lx, of table 2.1 yields the survival curve shown in figure 2.1. The curve shows an exponential decrease of survivors with time. A distribution plot of the absolute numerical value of dx also decreases with time, simply because in each succeeding age interval there are fewer survivors left to die (figure 2.2). As specified in our initial assumption, the value of qx is a constant (in this case set equal to 0.2) and therefore plots as a straight line with time (figure 2.3). The values of Lx and of Tx do not give rise to graphical plots; rather, they are used to calculate the value of ex, the further expectation of life at the beginning of age interval x. Since qx has been defined as a constant in this population, it follows that the average expectation of further life at any age for our hypothetical organism is 5 years. Therefore, a population that dies as result of random predation rather than of senescence generally displays the following characteristics: (1) The number surviving is a decreasing exponential function of time; (2)the age-specific death rate is constant at all ages; and (3) the further

expectation of life is constant at all ages, assuming a large enough population size. As a result of these three characteristics, the probability of any individual living long enough to age is almost zero, as is implied in figure 2.1. Thus, unless the original population were quite large, it is highly improbable that there would be any aging survivors. Even if there were one or two such survivors, their presence could not alter the fact that the population as a whole died from non–agerelated phenomena (see Witten 1994 for a more detailed description). Suppose we alter the assumptions underlying this life table such that a constant number—not a constant proportion—of our hypothetical population would die in each time period. The life table for this altered population is shown in table 2.2. The graphs corresponding to the values of lx, dx, and qx are shown in figure 2.4. The survival curve (figure 2.4a) describes a linear decrease with time because the constant number of deaths represents a larger and larger proportion of the dwindling number of survivors. The value of dx (figure 2.4b) is constant by definition. The value of qx rises sharply with time, also because the constant number of deaths represents a larger and larger proportion of the dwindling number of survivors. Note that the values of ex, the expectation of further life at birth, have different trends as a function of time in the two populations of tables 2.1 and 2.2. These are interesting theoretical distributions. Do any real populations have life table charac-

1,000

Figure 2.1 A survival curve, based on the lx data of table 2.1, for the population of individuals that do not senesce but die as a result of accidental events that affect 20% of the survivors each year. See text for further explanation. (After Lamb 1977.)

Number surviving (lx)

900 800 700 600 500 400 300 200 100 0

0 1

2

3 4

5

6 7 8 9 10 11 12 13 14 Age (years)

Number dying (dx)

2.2 Life Tables and Survival Curves

29

200 150

Figure 2.2 Distribution of the ages at death in a population of individuals that do not senesce, based on the dx data of table 2.1. (After Lamb 1977.)

100 50 0

1

0

2

3

4

5

6

7

8

9

10

11

12 13 14

Age-specific death rate (qx)

Age (years)

0.4 Figure 2.3 Plot of the age-specific death rate in a population of organisms that do not senesce, based on the qx data of table 2.1. (After Lamb 1977.)

0.3 0.2 0.1 0

0

1

2

3

4

5

6 7 8 9 10 11 12 13 14 Age (years)

teristics similar to those of either of these two hypothetical populations? Inanimate yet breakable objects might be a good real-world substitute for our hypothesized non-aging and supposedly immortal organisms. A life table for cafeteria tumblers was constructed from empirical data; the corresponding survival curves are shown in figure 2.5. The survival curve for ordinary (annealed) tumblers approximates the exponentially decaying curve that is characteristic of a constant age-specific death rate (as in figure 2.1). The survival curve for toughened tumblers approximates the linear decay characteristic of a constant, ageindependent number of deaths (as in figure 2.4). In this case, the toughened tumblers have a con-

Table 2.2 A Life Table for a Population of Organisms with Constant Number of Deaths X

0–1 1–2 2–3 3–4 4–5 5–6

lx

dx

qx

Lx

Tx

ex

1000 800 600 400 200 0

200 200 200 200 200 —

0.20 0.25 0.33 0.55 1.00 —

900 700 500 300 100 —

2500 1600 900 400 100 —

2.8 2.3 1.8 1.3 1.0 —

Source: after M. J. Lamb (1977).

stant but much lower number of “deaths” per time interval, and consequently have a higher e0 (Witten 1984, 1987). Such curves are often found in the biological world as well. Figure 2.6 shows the survival curve for wild lapwings in Britain. It is a clear-cut exponential survival curve with a value of ex that is constant between 2.2 and 2.6 years throughout most of the life span. Similarly, it is not uncommon to find linear curves in the biological world, as figure 2.7 illustrates. Should this observation be interpreted as meaning that these organisms are not subject to senescence? Obviously the answer is no, for most animals raised in captivity typically have an average expectation of life that far exceeds that observed in wild populations. The explanation for this apparent paradox is that in wild populations, the death rate from predation and other random events is so great that senescence has no chance to appear. Almost all members of the cohort are dead before vigor has declined significantly. The onset of senescence is impossible to detect from a life table if the mortality in early and adult life is very high. No individual has a chance to grow old. If we measured the survival of a biological population maintained under laboratory conditions,

30 Chapter 2 Measuring Age-related Changes in Populations

(a) 1,000

(b) 250

800

200 dx

600 400

100

200

50

0 (c)

150

lx

0

1

2

3

4

6

5

0

0

1

2

3

4

5

6

1.0 0.8

qx

0.6 0.4 0.2 0

0

1

2

3

4

6

5

Figure 2.4 Plots of data (from table 2.2) for a population in which a constant number of individuals dies in each time period. (a) Survival curve, based on lx data. (b) Distribution of the ages at death, based on dx data. (c) Plot of age-specific death rate, based on qx data.

Figure 2.5 Survival curves for cafeteria tumblers. Each scale division equals 2 weeks. The lower curve depicts the (exponential) survival of 549 annealed glass tumbers. The top curve depicts the (linear) survival of 241 toughened glass tumblers. (Based on Comfort 1965, from data of G.W. Brown and Flood 1947.)

Percent surviving (lx)

100

Toughened glass tumblers 50

Annealed glass tumblers

Time (x)

we would construct a life table similar to that shown in table 2.3. In this case, a cohort of 750 newly hatched adult male Drosophila was reared and maintained under controlled conditions of temperature, light, humidity, and so forth. The animals were transferred to fresh food every 4 days, and the number of flies that died in each

time interval was counted. Note that this is a longitudinal study. The number dead at each age was multiplied by 1.33 (1000/750) to normalize the dx values to a standard population size. All subsequent calculations were based on these scaled dx values. Note that the investigator chose to count the dead animals, but could equally well have

2.2 Life Tables and Survival Curves

31

1,000

Percent surviving (lx)

800

Figure 2.6 An exponential survival curve for lapwings, based on 460 birds banded as nestlings. Note the similarity of this empirical curve to the hypothetical curves depicted in figures 1.5a and 2.1. The curve is based on data from Lack (1943). (After Lamb 1977.)

600

400

200

0

0

1

2

3

4

5 6 7 8 Age (years)

9 10 11 12

Number surviving

70

50

30

10

12 20 30

100

150

166

Age (months) Figure 2.7 A linear survival curve for 77 mouflon sheep housed at the London Zoo. The data begin after the first year of life and combine male and female mortality information. (After Comfort 1979.)

32 Chapter 2 Measuring Age-related Changes in Populations Table 2.3 A Life Table for Adult Male Drosophila Raised in the Laboratory x (days)

nxa

lx

dx

qx

Lx

Tx

ex

0–4 4–8 8–12 12–16 16–20 20–24 24–28 28–32 32–36 36–40 40–44 44–48 48–52 52–56 56–60 60–64 64–68 68–72 72–76 76–80 80–84

750 750 749 746 741 737 732 719 702 648 561 421 334 254 170 108 79 41 19 6 0

1000 1000 999 995 988 983 976 959 936 864 748 570 445 338 227 144 105 54 25 8 0

0 1 4 7 5 7 17 23 72 116 178 125 107 111 83 39 51 29 17 8 —

0 0.001 0.004 0.007 0.005 0.007 0.017 0.024 0.077 0.134 0.238 0.219 0.240 0.328 0.366 0.271 0.486 0.537 0.680 1.000 —

1000.0 999.5 997.0 991.5 985.5 979.5 967.5 947.5 900.0 806.0 659.0 507.5 391.5 282.5 185.5 124.5 79.5 39.5 16.5 4.0 —

11864 10864 9864.5 8867.5 7876.0 6890.5 5911.0 4943.5 3996.0 3096.0 2290.0 1631.0 1123.5 732.0 449.5 264.0 139.5 60.0 20.5 4.0 —

47.5 43.3 39.3 35.5 31.9 28.0 24.2 20.6 17.1 14.3 12.2 11.4 10.1 8.7 7.9 7.3 5.3 4.4 3.3 2.0 —

Source: after Lamb (1977). an

x is the actual number of animals present at the beginning of each time period; these raw numbers are adjusted to yield the standardized numbers listed under lx.

chosen to count the living animals at each interval and thus constructed an lx-based life table. If the counts were accurate, the two life tables would be identical. Figure 2.8 illustrates the graphical plots for the survival curve (lx), the distribution of ages at death (dx), and the age-specific death rate (qx). Comparison of the curve of figure 2.8 to those in figures 2.1 or 2.2 makes clear that this biological population differs markedly from our hypothesized non aging populations in the three life-table characteristics such that (1) The survival curve is more rectangular; that is, very few individuals died early in life; (2) the distribution of ages at death reaches a peak value late in the life span of the population; (3) the further expectation of life decreases with increasing age; and (4) the age-specific death rate increases with age. These characteristics would be expected if the organisms were dying as the result of cumulative, progressive, intrinsic, and deleterious (CPID) changes that resulted in an increased susceptibil-

ity to death—that is, if they were dying of old age. After a certain age the organisms die from proximate causes that would not have killed them in their youth. Their susceptibility has increased. Therefore, the preliminary evidence for the presence of aging and senescence in a population is the presence of a more or less rectangular-shaped survival curve. A variation of this type of survival curve would result if the population were subjected to an initial period of high juvenile mortality followed by a plateau with very few deaths until the onset of senescence, when the age-specific mortality would increase. A survival curve for such a population is shown in figure 2.9. This sort of curve is typical of many populations of large mammals such as impalas, zebras, buffalo, and humans (Spinage 1972; see figure 2.16). Figure 2.9 is based on the analysis of 608 skulls of wild Dall mountain sheep that died at an unknown time. The age of these sheep at death was determined by counting the annual growth rings on the horns.

2.2 Life Tables and Survival Curves

(a)

33

1,000 900 Number surviving

800 700 600 500 400 300 200 100

Figure 2.8 The (a) survival curve, (b) distribution of ages at death, and (c) plot of age-specific death rate for the population of male Drosophila melanogaster whose life table is shown in table 2.3. Note the similarity of these empirical curves to the hypothetical curves in Figure 1.5b and c. (After Lamb 1977.)

0

(c)

Age-specific death rate (qx)

Number dying

(b)

200 150 100 50 0

1.00 0.75 0.50 0.25 0

0

8

16

24

32

40

48

56

64

72

80

Age (days)

There was no way of determining the cause of death of individual animals—whether illness, predation, or natural causes. The corresponding life table (table 2.4) and this survival curve are based on dx data, for the investigator had no other recourse. All other numbers in the life table were generated from this cross-sectional dx data, on the assumption that the population consisted of 1000 individuals and had a constant age structure. Of interest here is the existence of two periods of relatively heavy mortality—very early in life and then again very late in life—with high survival rates in the intermediate years. Again, this type of survival pattern appears to be common in large mammals, including humans.

Another type of theoretically possible survival curve might be expected in populations characterized by an enormously high death rate in early life, followed by a lower rate later in life. Populations of trees or of various fishes, for example, are characterized by high egg and juvenile mortality. Once established, however, the adult organisms have a significantly higher life expectancy than they did as juveniles, presumably because they are now much less susceptible to environmental effects. A similar situation applies to Drosophila; thus, if we assumed that the 1000 adult males alive at the start of the life table in table 2.3 were the survivors of a 90% mortality affecting the egg and juvenile stages (a not unrealistic assumption

34 Chapter 2 Measuring Age-related Changes in Populations

Number surviving (lx)

1,000

800

Figure 2.9 A survival curve for Dall mountain sheep, based on the remains of 608 sheep (see table 2.4) whose age at death was determined from the annual growth rings on the horns. See text for explanation. (After Lamb 1977, based on data given in Deevey 1947.)

600

400

200

0

0 1

2

3 4

5

6 7 8 9 10 11 12 13 14 Age (years)

Table 2.4 A Life Table for Dall Mountain Sheep Based on Estimated Age at Death x (years) 0–0.5 0.5–1 1–2 2–3 3–4 4–5 5–6 6–7 7–8 8–9 9–10 10–11 11–12 12–13 13–14

lx

dx

1000qxa

ex

1000 946 801 789 776 764 734 688 640 571 439 252 96 6 3

54 145 12 13 12 30 46 48 69 132 187 156 90 3 3

54.0 153.0 15.0 16.5 15.5 39.3 62.6 69.9 108.0 231.0 426.0 619.0 937.0 500.0 1000.0

7.06 — 7.7 6.8 5.9 5.0 4.2 3.4 2.6 1.9 1.3 0.9 0.6 1.2 0.7

Source: after Deevey (1947). aMortality rate per 1000 animals alive at beginning of age interval.

for wild populations), then we could construct an L-shaped survival curve for this population, as shown in figure 2.10. An individual that survived the period of high initial mortality would thereafter enter a period in which the further expectation of life would be very long. In some species, the further expectation of life for the survivors might even increase with age; this phenomenon is believed to be true of certain trees. In this case,

the value of lx would only very slowly approach zero. The different types of survival curves we have discussed are summarized in figure 2.11. Curve A represents a population that suffers little from deaths until the onset of senescence, at which time all the members of the cohort die more or less simultaneously. Such a situation might result if one constructed a life table for a species that reproduces only once and dies immediately afterward (e.g., mayflies). The same sort of highly rectangular curve could be generated by a catastrophic environmental event acting on the entire cohort at one particular time. For example, a commercial herd of beef cattle would show such a sharply rectangular curve, with most members of the cohort dying at 2 years of age. It would be difficult to distinguish the two cases based solely on the life table data; additional information would be needed to choose between the options of synchronized senescence or environmental catastrophe. Curve B of figure 2.11 represents a typical survival curve for an aging cohort, such as that shown in figure 2.8a. However, a similar curve was found to apply equally well to inanimate objects such as automobiles (Pearl and Miner 1935; also see figure 2.23), suggesting that any complex system, be it animate or inanimate, can be described by a life table. Curves C, D, and E represent the previously discussed linear, exponential, and L-shaped curves, respectively.

2.2 Life Tables and Survival Curves

Number surviving

10,000

1,000 500 0

0

16

24

32

40

48

64

56

72

Age (days)

Figure 2.10 A survival curve for the Drosophila depicted in figure 2.8a, based on the assumption that the 1000 animals present at the start of adult life are survivors of a 90% rate of larval mortality.

A

Percent surviving

B

C D

E

Age

Figure 2.11 A compilation of the different types of survival curves observed in different populations. (A) Population with few deaths until senescence. (B) Typical survival curve for an aging cohurt. (C) Linear curve. (D) Exponential curve. (E) L-shaped curve. See text for further explanation.

The interest in life tables is prompted by the fact that the relationship between age and mortality they display may tell us whether a population is senescing. Thus we would be most interested in studying populations whose survival curve resembles either curve A or curve B in figure 2.11, but would not likely to focus our attention on populations C–D. In addition, without the survival curve data, we would not be able to deter-

35

mine whether the life expectancy of an aging population could be improved and, if so, which portion of the life cycle should be the target of such interventions. Note that life expectancy can be calculated at any desired age. Although it is often used to denote the expected life span from the time of birth, it can also be used to estimate the life span expectancy given that an individual has lived to a specified age. What empirical basis do we have for implicitly concluding that the decreased probability of survival in these curves reflects an intrinsic biological process? Comfort (1979, p. 25) gave an excellent answer to this question when he showed that “the age distribution of pedestrian deaths in road accidents was similar in contour, excluding early infancy, to the general distribution of human deaths from all causes. This index is highly correlated with vigor, in its biological sense, for it represents a combination of sensory acuity, speed of avoidance, and power of recovery when hit.” A more modern approach that yields similar conclusions comes from the work of several groups trying to construct a “frailty index” for humans. The frailty index is based on individual physiological or health data, and the assumption is that the more disabilities a person has, the frailer the person might be, and the greater the probability he or she has of dying in the near future. An examination of a Canadian population yielded preliminary data showing that the mortality rate does indeed increase as the frailty index increases (Mitnitski et al. 2002), which is of course the point that Comfort made with regard to pedestrian accidents. Another consideration for the experimental gerontologist is the necessity of dealing with survival curves of different species, which may have similar shapes but very different absolute values. Such a comparative approach would be of value in discovering whether the laws of mortality are similar or different in different species. Is there a valid method of easily comparing such diverse data? One early approach to dealing with this problem was put forth by Pearl (1922), who proposed plotting the survivorship in equivalent life spans of different organisms versus percentiles of

36 Chapter 2 Measuring Age-related Changes in Populations 1,000

Figure 2.12 A comparison of the survival curves for three species with different life spans, in which the age of the individuals within each species is expressed as the percentage deviation from the species-specific mean length of life. The mean life span for the Dall sheep is 7.09 years, for the herring gull 2.44 years, and for Floscularia (a sessile rotifer) 4.74 days. (After Deevey 1947.)

Number surviving per 1,000 born

500

Herring gull

100 50

10

Floscularia

Dall sheep

5

1 -100

0

+100

+200

+300

Percent deviation from mean length of life

the life span. He defined equivalent life span as the period between A, the point in the life history of each organism at which the value of qx is at a minimum, and B, the point at which one survivor remains out of the 1000 organisms starting at point A. Pearl then divided the span between these two points into 100 portions, thus measuring age in percentiles of the life span and not in absolute chronological terms. When Pearl used this procedure to compare the survival curves of Drosophila and of humans, he found that they had the same relative shape and thereby concluded that the laws of mortality are fundamentally the same in the two organisms. This comparison also allowed Pearl to suggest that 1 day in the life of a fly was approximately equivalent to 1 year in the life of a person. This procedure might well be the source of the other life span comparisons that we have all heard, such as the idea that 1 year in a dog’s life corresponds to 7 years in a human’s. This method was subject to certain criticisms of a statistical nature. Seeking to improve on it, Pearl and Miner (1935) hit upon the idea of presenting the life table data not in absolute terms, but in terms of percentage deviations from the mean life span (figure 2.12). This different approach led to the same conclusion; namely, that

there are similarities of pattern in these lifehistory characteristics of otherwise different species, suggesting that the number of different biological processes involved in senescence and aging might be of a general nature and, therefore, that their number might be relatively small. The mechanistic data presented later in this text supports this conclusion. This conclusion was more important than was perhaps anticipated at the time, for if every species has a unique mechanism of senescence, it might prove difficult to achieve any sort of general understanding. Nonetheless, the possibility that there might be only a small number of ways in which organisms age would allow us to use a comparative approach to achieve the necessary insight. The similarity of fundamental biological processes has underlain much recent progress in biology. Eakin and Witten (1995a,b) have developed new methods for the uniform comparison of survival curves across species. The recent use of the mortality rate doubling time, a concept derived from the use of the Gompertz plot of agespecific mortality (all of which I will discuss later in this chapter), has simplified the comparison of the life-history characteristics of different populations and species and is now the major method by which such cross-comparisons are made.

2.2 Life Tables and Survival Curves

2.2.1 Constructing Life Tables All of the life tables presented in the previous section were constructed on the basis of observations of a cohort of organisms born at the same time; the age of death of each individual was then recorded. Either the dx curve or the lx curve represents the primary data set; all other values in the life table are derived from this curve. Such cohort life tables can be constructed only for captive or laboratory populations. It is not possible to keep track of all individuals throughout their life in human or most other noncontrolled animal populations. In practice, then, human life tables must be constructed using data that have been obtained by other, indirect methods and are termed period life tables. The differences between cohort and period life tables are essentially the same as those between longitudinal and cross-sectional studies, as discussed in chapter 3. In the period or indirect approach, census data and death certificates for the same population are compared. Comparison of the age distribution of the population with the age distribution of deaths allows the value of the age-specific probability of dying, qx, to be calculated for each age group represented in the original data. All other life table values are then derived from this secondary data set. This manipulation provides us with a cross-sectional sampling of the projected mortality rates for each age group in the population at the same time the data were originally collected. The agespecific mortaltiy rates are those characteristic of the 1 year olds, 2 year olds, 3 year olds, and so on, in that year. The result is not a true historical account of a cohort population. Given a constant environment and a constant age structure of the population, the two types of life tables would tend to approximate one another. These conditions do not apply to human populations. Not only do singular environmental events occur (such as the influenza pandemic of 1918–1919 or the contemporary AIDS epidemic), but in human populations, there are at work long-term cultural trends (such as the introduction of sewer systems or of antibiotics) that reduce the force

37

of mortality. Figure 2.13 illustrates the indicated life table curves for an indirect human life table (table 2.5) based on the population of England for 1960–1962. A comparison of these curves with the corresponding ones derived from a cohort life table of Drosophila (see figure 2.8) suggests that the general shape and pattern of the curves are quite similar. An indication of the long-term trends affecting this population is the alteration in the value of the age-specific death rate, qx, for this population over a 50-year period (see table 2.5). Note that the decrease in the force of mortality is age specific. There has been an 81% drop from 1910 to 1960 in the value of qx at age 0, a 75% decrease at age 40, and only a 12% decline at age 80. These trends suggest that the forces responsible for the increase in life span noted during this half century have a larger impact on younger individuals than on older ones. The independently obtained set of human survival data shown in figure 2.14 verifies this point while raising another issue. A comparison of the combined survival data for males and females throughout most of the 20th century reveals important and different types of changes in mortality over this period of rapid social and environmental change in the United States. In 1900–1902, about 20% of the population died by the age of 10. What makes this particularly appalling is that prepubertal children should be the healthiest subset of any human population, as will be shown by the data and discussion of figure 2.16. By 1919–1921, this infant/child mortality had been cut almost in half, thanks to the introduction of public health measures such as sewer systems. Much of the increase in survivorship during this period was also due to decreases in middle-aged mortality as well. In the next 20 years (up to 1939–1941), the largest gains in survivorship occurred in the middle-aged subset of the population, while most of the mortality gains in the last half of the century (i.e., 1959– 1961 to 1989–1991) were concentrated in the middle-aged and elderly adult subsets of the population. In 1901, the median life span was 58 years; by 1990, it had risen to 79 years. In 1900,

38 Chapter 2 Measuring Age-related Changes in Populations

Number surviving (lx)

100,000

75,000

50,000

25,000

Figure 2.13 Curves illustrating the survival, distribution of ages at death, and age-specific death rate for human males in England from 1960 to 1962, based on the data in table 2.5. See text for explanation and discussion. (After Lamb 1977.)

Number dying per 100,000 (dx)

0

3,000 2,500 2,000 1,500 1,000 500

Age-specific death rate (qx)

0

0.4 0.3 0.2 0.1 0

0

10

20

30

40

50

60

70

80

90 100

Age (years)

1.9% of the population survived to age 90 and 0.03% survived to age 100. In 1990, 80% survived to age 65, 17% survived to age 90, and 1.4% survived to age 100. These numbers document the outstanding accomplishment of the 20th century—namely, the overall improvement of the human environment so that most of us can live out our full life span potential. It is apparent that different environmental changes had different

effects on different parts of the life span. Note, however, that there appears to be only a small and statistically nonsignificant increase in the maximum life span. Does this indicate the existence of some inherent limit to human longevity? We will return to this topic later in this chapter, as well as in chapters 5 and 14. There are two other important conclusions the reader should draw from Figure 2.14. First, this

2.2 Life Tables and Survival Curves

39

Table 2.5 Secular Changes in the Force of Mortality (qx) for the U.S. Population at Various Ages over the 20th Century Age (years) <1 15–24 35–44 55–64 75–84 85+

1901

1921

1941

1961

1981

1998

0.14139 0.00555 0.01029 0.02777 0.12462 0.26083

0.08063 0.00387 0.00676 0.02209 0.11121 0.23914

0.05258 0.00201 0.00500 0.02132 0.10576 0.22796

0.02587 0.00103 0.00293 0.01675 0.08399 0.19634

0.01201 0.00029 0.00222 0.01323 0.06448 0.15432

0.00751 0.00082 0.00200 0.01031 0.05703 0.15112

Source: National Center for Health Statistics Historical Tables (http://www.cdc.gov/nchs/datawh/statab/ unpubd/mortabs/hist290.htm) Note: Data are age-adjusted death rates for both sexes of all races for the indicated age-group during the indicated year.

1900-1902

1919-21

1939-41

1959-61

1979-81

1989-91

100

Percent surviving

80 60 40 20

0 0

10

20

30

40

50

60

70

80

90

100

Age in Years

Figure 2.14 The survival curves for U.S. males and females in 1900–1902, 1919–1921, 1939–1941, 1959–1961, 1979–1981, and 1989–1991. This type of survival pattern is representative of other countries that have made the transition from high to low mortality. (After National Center for Health Statistics 1999.)

family of curves shows that the large increase in human longevity during this century was due entirely to a decrease in premature death and not to any real increase in the maximum life span. More of us are living longer, even though none of our technological inventiveness to date has done anything to alter the biological aging processes. A discussion of the likely outcome on the human life span of combining both environmental and biological interventions is contained in chapter 14. Second, figure 2.14 implies that the 1900 population had many more young people than

middle-aged or older people, while the 1990 population must have a much more equal proportion of young, middle-aged, and older individuals. The age structure of the two populations must be quite different. Evidence to support this supposition is presented in figure 2.15, which specifically illustrates the dramatic shift in the age structure of the two populations as a result of the decrease in the age-specific mortality rates. I return to this topic in chapters 5 and 15. The difficulties in constructing a cohort life table for humans apply equally obtaining accurate

40 Chapter 2 Measuring Age-related Changes in Populations

Population by Age and Sex: 1905 Male

Age

Female

75+ 70-74 65-69 60-64 55-59 50-54 45-49 40-44 35-39 30-34 25-29 20-24 15-19 10-14 5-9 0-4 Population in millions Population by Age and Sex: 2010 Male

Age 90+ 85-89 80-84 75-79 70-74 65-69 60-64 55-59 50-54 45-49 40-44 35-39 30-34 25-29 20-24 15-19 10-14 5-9 0-4 Population in millions

Female

Figure 2.15 The U.S. population sorted by sex and by age in 5-year cohorts. Top panels: The U.S. population in 1905. Bottom panels: The projected U.S. population in 2010. Note the triangular shape of the 1905 population compared to the increasingly squared-off shape of the 2010 population, which largely results from the significant decrease in premature mortality. (From U.S. Bureau of the Census 1965; Day 1993.)

life tables for wild populations of animals. In certain cases, it is possible to ascertain the age of individuals directly—for example, by counting the growth rings in the scales of fishes, in the horns of ungulates, in the trunks of trees, or in the shells of mollusks. In these cases, a large random sampling of the population will allow its age structure (lx) to be determined and an indirect life

table to be constructed. If the indicator structures are stable and are not affected by postmortem changes, the remains of dead animals may be examined and the population’s age at death (dx) determined. The survival curve shown in figure 2.9 (based on the data of table 2.4) is of this type. Several of the more ecologically and statistically reliable life table data sets (such as that in fig-

2.2 Life Tables and Survival Curves

ure 2.9) have been reconstructed and reanalyzed by A. R. Miller (1988). These refined and empirically sound life tables will provide a rigorous test for any theory of aging as it manifests itself at the organismal and population levels.

2.2.2 Special Transformations of Survival Curves The preceding discussion makes clear that, in aging populations, the risk of death increases with age. This idea seems intuitively obvious, yet little could be done with this insight until it was quantitated. The mathematical relationship of this increased risk was first described by Benjamin Gompertz in 1825, and it is an excellent example of the contributions of actuarial necessity to gerontology, for his motive in investigating this relationship was the need to determine the future value of life annuities. This knowledge is crucial to the profitability of insurance companies, which must be able to adjust the premium charged to the risk entailed. Such information is useful in other ways to gerontologists today. Gompertz found that the age-specific death rate increases as an exponential function of age: qx = (q0)ex, where q0 = further expectation of life at the time of birth, or the y intercept; qx = further expectation of life at the beginning of an age interval, x; and x = the slope constant. The equation can also be written in linear form as ln qx = ln q0 + x, which is, of course, a particular form of the general equation for a straight line: y = b + mx. The discussion that follows reveals what is particularly interesting and useful about these equations. The age-specific mortality is defined as the probability of dying at a certain age among those individuals who have already survived to that age. Calculating these values for a given population, and plotting them versus age, produces curves of the types shown in figure 2.16.

41

When plotted in a linear fashion (figure 2.16a), the rapid increase in the age-specific mortality over age 50 or so becomes very evident. The older you are, the greater the probability of death. This rapid increase is exactly what would be expected of an exponentially increasing function. In this kind of group, the rate of increase is so rapid that not much detail can be seen. It is inconvenient to make the scale of the graph much bigger than it is already, but when the same data are plotted semilogarithmically (figure 2.16b), certain aspects of the curve stand out. Ham and Veomett (1980) pointed out five distinct portions (labeled A–E in figure 2.16) in the plot of human age-specific mortality. The very young are poorly adapted for survival and have a high mortality rate (A), due primarily to congenital defects and infectious diseases. Precisely because babies are so sensitive to environmental insults, infant mortality statistics are used as an index of the state of a community’s overall health programs. For each year of life successfully completed, the probability of dying in the following year decreases, up to the age of about 10, when the probability of dying is lower than at any other time in the life span (B). The mortality rate increases among teenagers and young adults (C). This increase is due primarily to accidents, which are the major cause of death in that age range. From about age 30 until advanced old age, there is an almost perfect exponential increase in the age-specific mortality (D). At ages above about 95 years, the rate of increase in mortality for the few people who have survived that long appears to become less (E). This slowing of the mortality rate in life might be artifactual or it might be real. In the first case, a reasonable hypothesis would be that the decreased mortality rates observed in section E of the curve are the result of skewed data arising from a few individuals who, in reporting their age, exaggerated for some reason. Accurate birth records are difficult to obtain for many centenarians, so this scenario is a real possibility (see chapter 6). However, the emphasis recently placed on verification of vital records (Jeune and Vaupel 1999; Perls et al. 1999), coupled with the experimental replication of decreased late-life mortality

42 Chapter 2 Measuring Age-related Changes in Populations

(a)

Age-specific mortality(probability of not surviving an additional year

0.5

Age-specific mortality

1.0

(b)

0.4

0.3

0.2

0.1 0.05 0 0 25 50 75 100

E 0.1 D A 0.01

C 0.001

0.0001 0

Age (years)

B Hypothetical Gompertz mortality at birth (M0) 20

40

60

80

100

Age (years, at beginning of year)

Figure 2.16 Plots of the age-specific mortality in the United States, 1959–1961. The mortality rates are plotted every 5 years up to age 75 and every year thereafter. (a) A linear plot. Note the great increase in mortality at advanced ages. (b) A semilogarithmic plot. The dots indicate the data points; the dashed line is an extrapolation of the data to age 100 and to age 0 in order to yield a hypothetical vulnerability to age-related death at the time of birth. The mortality doubles about every 8 years. A, high mortality rate of very young; B, mortality at age 10; C, mortality among teenagers and young adults; D, mortality at age 30 until old age; E, mortality at age 95. (After Ham and Veomett 1980.)

rates in other organisms and by other investigators, has made this alternative explanation less and less tenable. Something real is happening with mortality rates of the very old. There are two plausible explanations for this observation. First, this slowing of the mortality rate late in life could be due to the existence of a small subpopulation of individuals who age at a slower rate, and thus dominate the small group of survivors left at advanced ages (e.g., the heterogeneity hypothesis). If confirmed, these individuals could represent the differential aging characteristic of the longlived portion of the population and herald the existence of a slower aging, longer lived subset of human beings. An alternative hypothesis (the individual-risk hypothesis) states that this slowing of the mortality rate late in life might simply reflect the late-life plateau in the force of natural selection (see chapter 4), which might occur for physiological, evolutionary, or statistical reasons. If confirmed, this finding would imply that senescence stops in old age, that it does not continue through the entire life span. If so, it might

be useful to understand better what causes senescence to stop and to determine if we can manipulate such processes to make them occur earlier. I discuss these possibilities in “Late-Life Mortality Kinetics” below. Another potentially useful aspect of this agespecific mortality plot is the observation that the largest portion of the human life span is adequately described by that portion of the Gompertz plot indicated in section D of figure 2.16b. Can this region be used for comparisons of mortality rates and/or other interesting demographic parameters between different human populations? Can these mathematical abstracts of the survival data be used for accurate comparisons of the mortality kinetics of different species? Is it possible to make logical comparisons of aging in long-lived and short-lived species? Do different patterns of mortality rates give rise to correspondingly different survival curves? These are interesting questions, each with important implications regarding our ability to understand better the biology of senescence. The next section of this

43

2.2 Life Tables and Survival Curves

chapter is devoted to the search for answers to these questions.

Comparisons of Mortality Kinetics. Given the curve shown in figure 2.16b, one can extrapolate from the most prominent part of the curve, the linear portion (D) of this age-specific mortality, up to age 100 and down to age zero. The result is a straight line that can be described completely with only two components: the y intercept (q0) and the slope (x). This feature represents one of the great values of the Gompertz function, in that it takes all the complex data contained in the life table and implicit in the survival curves and reduces it to just two numbers. The former, q0, is a hypothetical value for vulnerability to death due to age-related causes at birth. Perhaps a better approach is to think of this value as representing the genetically determined vigor of the genotype. The latter value (x) measures the rate of increase in mortality and has been considered to represent the rate of aging. Although some information may be lost in this process of simplification, it makes it easier to compare the longevities of different populations and gain some insight into the processes at work. A change in either value will affect life span. A reduction in the value of q0 would reduce mortality at all ages and result in a longer average survival. A comparison of curves A and B or C and D in figure 2.17 would show that populations B and D have longer survival curves relative to their partners. However, reducing the value of q0 would have no effect on the rate of aging because the slopes of the two lines would be identical. Changing the rate of aging (x) while leaving q0 unaffected would also decrease the agespecific mortality rate and hence increase life span (compare curves A and C or B and D in figure 2.17). Decreasing both values would result in a maximum increase in life span (compare curves A and E in figure 2.17). The actual data tend to support this theoretical interpretation. Gompertz plots of rodent survival data show that the differences in species-specific life spans typically

Age-specific mortality (log scale)

2.2.2.1 The Biological Meaning of Transformed Survival Curves

A

B

C

D

Original Gompertz mortality

Slower increase in mortality

Lower initial vulnerability (M0)

E

Age (linear scale)

Figure 2.17 Effects of different values of the agespecific mortality (q0) and of the rate of aging (x) among population A–E on the Gompertz mortality. These terms and their interactive effects on mortality are described in the text. (After Ham and Veomett 1980.)

involve alterations in both parameters (figure 2.18). Within a species, however, interventions that affect life span typically alter only one or the other of the constants of the Gompertz equation. Detailed analyses of the methods used in these estimates may be found in Carey (1999) and Witten (1995). Subjecting people to highly adverse conditions, such as the malnutrition and confinement associated with war, often results in an increased mortality rate. The Gompertz plots for such populations show that the severe stress does not affect the rate of aging (the slope, or x) but does affect the initial vulnerability (the y intercept, or q0) of the affected population (figure 2.19). It is not known whether the rate of aging would be altered at some longer time after the stressful period was over. Other examples of this phenomenon can be found in the various animal studies. Long-term administration of the anesthetic procaine to rats increases their life span by reducing their initial vulnerability (q0), yet it has no effect on their rate of aging (x) (figure 2.20). Conversely, caloric restriction also increases the life span of rats, but it does so primarily via a decrease in the rate of

44 Chapter 2 Measuring Age-related Changes in Populations

1.0

1 2 3 4 5

▲ ▲ ▲

0.8 Fraction surviving

▲

▲

Sigmodon hispidus Mus musculus Oryzomys palustris Peromyscus leucopus Peromyscus (other)

▲

0.6 ▲

6 5

0.4

0

2

1

0.2

500

0

▲

3 ▲

1,000

1,500

2,000

2,500

3,000

Age (days)

▲

100 Age-specific death rate, per 10,000 per day

Figure 2.18 Survival curves (a) and the Gompertz plots (b) of the mortality data for wild-type populations of rodent species bred and reared in captivity. (After Sacher 1977.)

▲

1 ▲

30

2 3 4

▲

5 10

6 ▲

3

▲

1 2 3 4 5 6

1

0

0

500

1,000

▲

Sigmodon hispidus Mus musculus Oryzomys palustris Peromyscus californicus (inc.) Peromyscus leucopus Peromyscus (pooled, inc.)

1,500

2,000

2,500

Age (days)

aging (x) coupled with a slight increase in the initial vulnerability (q0) (figure 2.21). Presumably, these data tell us that these two different methods of prolonging the life span are bringing about their effects via very different mechanisms, even though they are affecting the same systems.

These mechanistic differences bring about different patterns of mortality, which are then clearly detected by the Gompertz transformations of the survival curve. Detailed analysis of human historical data strongly suggests that the temporal changes in the survival curves of figure 2.14 are

2.2 Life Tables and Survival Curves

(b) A 500 200 B 50

C

20 5 2 0.5 0.2 0

20

40 60 Age (years)

10

Number dying, per 1,000 per year

Number dying, per 1,000 per year

(a)

80

45

Netherlands (males)

1945 1

1946

0.1 0

20

40 60 Age (years)

80

Figure 2.19 Mortality rates as a function of age in human populations subjected to prolonged, dire stress, as analyzed by the Gompertz mortality rate model in which the logarithmic age-specific mortality rate is plotted against age. The rate of mortality (i.e., the slope) remains fairly constant, but the initial mortality rate (i.e., the intercept) changes significantly. (a) Curve A represents Australian prisoners of war held in concentration camps by the Japanese army during 1945; curve B represents civilians in Australia, 1944–1945; curve C represents white females in the U.S. 1980 census. (b) The initial mortality rate shifted without affecting the mortality rate slope in Netherlands male civilians in 1945 versus 1946 during and after World War II. (After Finch 1990, redrawn from Jones 1959.)

10-2

Age-specific death rate per day

Age-specific death rate per day

▲

10-2

10-3

10-4

0

4.0

8.0 12.0 16.0 20.0 24.0 28.0 32.0 Age (months)

Figure 2.20 The effect of procaine on the mortality of rats. The long-term treatment of rats with procaine (gray circles) reduces vulnerability to death (qx) at all ages as compared to controls (black circles) but does not alter the rate of aging (x). (After Sacher 1977.)

Control populations

▲

▲

▲

10-3 Restricted populations 10-4

10-5

10-6 0

200

400

600 800 Age (days)

1,000

1,200

Figure 2.21 The effect of caloric restriction on mortality of rats. Two restricted populations (gray symbols) are compared with three control populations (black symbols). Note that in these populations, caloric restriction significantly reduced the rate of aging (x) but had a mildly adverse effect on the initial vulnerability (q0). (After Sacher 1977.)

46 Chapter 2 Measuring Age-related Changes in Populations accompanied by complex alterations in the q0 and x variables of the Gompertz function (Yashin et al. 2002). Finch et al. (1990) have shown that, with a few modifications, the Gompertz parameters may be used for interspecific comparisons. One modification is to eliminate the neonatal and prepubertal periods (section A of the plot in figure 2.16b) and compute the y intercept value (qx) as if it occurred at puberty (section B), when it is at a minimum. This value is called the initial mortality rate (IMR). Assuming that puberty takes place at approximately 12 years of age, the IMR for figure 2.15b would be about 0.00025 by inspection (alternatively, this value can be calculated from the linear form of the Gompertz equation). The second interesting modification to this equation made by Finch et al. (1990) is the introduction of a new unit, the mortality rate doubling time (MRDT), which is related to the slope (x) of the Gompertz plot by the equation MRDT = ln 2/x = 0.693/x As Finch (1990) makes clear, the MRDT is derived from the slope but is more useful because it varies in the same direction as the maximum life span and is measured in the same units of time. It measures the period of time during which the probability of not living until the next time period doubles. It is more comprehensible to state that the MRDT of humans is about 8 years than to state that the slope of the human age-specific mortality plot in figure 2.16b is 0.087 units. What this MRDT value means is that when you are 36 years old, your chances of dying that day are only half what they will be when you are 44 years old, but twice as high as they were when you were 28 years old, and four times as high as when you were 20 years old. The MRDT quantifies in easily comprehended terms the increased probability of dying with age, which is one definition of aging (see chapter 1). Table 2.6 presents representative values of mortality rate coefficients for a variety of different species with different types of senescence patterns. Consider first the differences between the three human populations listed in that table.

The 1945 prisoners of war (see figure 2.19A) have the highest IMR; the 1997 females have the lowest values. This may be attributed largely to the differences in the three environments. But despite the 100-fold difference in the IMR, the MRDT values seem to oscillate about a value of 8 or so years. Male prisoners and modern women age at the same rate; the former just start from a higher degree of susceptibility (as seen in figure 2.19B). If the slope of the Gompertz curve is indicative of the rate of aging, then altering the environment changes the susceptibility to death of the individuals in the population but does not alter their intrinsic aging rate. It is known that healthy aging in humans (i.e., remaining free of illness and impairment) is largely based on the removal of risk factors present in early and midlife (Reed et al. 1998). The removal of such risk factors lowers the IMR but does not affect the MRDT (figure 2.19). This explanation provides the mechanism by which we can understand why the mortality reductions implicit in figure 2.14 increased the mean life span but had little effect on the maximum life span. Humans have one of the longest MRDT values; the MRDT of a short-lived and rapidly senescing organism, such as the fruit fly, is about 500-fold faster. Note that the difference between humans and horses, or between monkeys and dogs, resides entirely in the MRDT, for their IMRs are equivalent. Thus the MRDT appears to serve as an empirical measure of the rate of senescence. In addition, flying vertebrates (bats and birds) have a longer MRDT than would be expected from their IMR values. Slow senescence appears not to be limited to animals with a large body size, as was once supposed, and the reasons for this are discussed in chapter 5. Finally, an inspection of the limited data available in table 2.6 leads one to the suspicion that, since long-lived organisms with gradual senescence have low IMR and high MRDT values, a similar situation might well occur in those not well-studied but very longlived species with negligible senescence (such as lobsters, etc.; see the website www.agelessanimals .org. for more information on these species). Suppose we consider two species with comparable MRDT values, such as horses and brush

2.2 Life Tables and Survival Curves

47

Table 2.6 Mortality Kinetics of Organisms with Different Senescence Patterns Species Humans Prisoner of war, 1945 U.S. female, 1980 U.S. female, 1997 Baboon females Horse Rhesus monkey Domestic dog White-footed mouse Lab rat Lab mouse Lab gerbil Pipstrelle bat Herring gull Brush turkey Bengal finch Pea fowl Reeves pheasant Japanese quail Broad-tailed hummingbird European robin Starling Andean condor Guppy Lake sturgeon Fruit fly House fly Honeybee worker Winter Summer Soil nematode Rotifer Nonfeeding moth Bamboo Bristlecone pine Tortoise Quahog

IMR/year

MRDT (years)

0.0070 0.0002 0.00004 0.00510 0.0002 0.02 0.02 0.06 0.02 0.03 0.1 0.36 0.004 0.045 0.1 0.06 0.02 0.07 0.25 0.5 0.5 — 0.07 0.013 0.01–0.04 4–12

7.7 8.9 8.0 4.8 4 15 3 1.2 0.3 0.27 0.9 3–8 6 3.3 2.5 2.2 1.6 1.2 — 8 >8 — 0.8 10 0.02–0.04 0.02–0.04

<0.001 0.2 2 6 10 — — — —

0.03 0.02 0.02 0.005 0.005 — — — —

Senescence pattern

Maximum life span (years)

Gradual — — — Gradual Gradual Gradual Gradual Gradual Gradual Gradual Gradual Gradual Gradual Gradual Gradual Gradual Gradual Gradual Gradual Gradual Gradual Gradual Gradual Gradual Rapid Rapid

125 — — —a 33a 46 >35 20 8 5.5 4.5 3.8 >11 49 12.5 9.6 9.2 9.2 5–8b >12b 12 20 70–80 5 >150 0.3 0.3

Rapid Rapid Rapid Rapid Rapid Rapid Negligible Negligible Negligible

0.9 0.2 0.15 0.10 0.03 <120 >5000 >150 >200

Source: compiled from data presented in tables 2.1, 3.1, 3.2, 4.1, appendix 1, and appendix 2 of Finch (1990), unless otherwise noted. Note: IMR, initial mortality rate; MRDT, mortality rate doubling time. aData

from Bronikowski et al. (2002).

bData

from Holmes and Austad (1995).

turkeys (see table 2.6). Is it valid for us to conclude that these two organisms, different as they are from one another, must necessarily undergo the same pathophysiological mechanism of senescence? Probably not. The Gompertz equation and its various derivatives describe an empirical rela-

tionship. The equation is not dependent on a theoretical relationship between life span and some variable. So when we tinker with the equation and get it to fit some empirical data, we know only that the curve fits the data. But because we don’t know why it fits, we can’t logically deduce

48 Chapter 2 Measuring Age-related Changes in Populations the nature of the underlying biological mechanisms. Another argument against overinterpreting these mathematical relationships is illustrated by the fact that inanimate electrical relays display a time-to-failure (survival) curve that is identical in form to that of an aging biological population. It seems obvious that relays and rhinoceroses senesce as a result of very different mechanisms. Thus, the fact that different systems yield similar curves does not mean that similar mechanisms are operating in the two different systems. Another interpretation of the same data is that what we are actually measuring in both animate and inanimate systems is simply the breakdown in connectiveness between compartmentalized but integrated systems (Finch 1990), wherein a subthreshold injury in one compartment increases the probability that another such injury, occurring independently in a connected component, will result in breaking the connection and the subsequent systemic failure of the machine or organism. In this deeper view, then, the important factor is the connectiveness of organisms and the failure that results from its loss. The nature of the particular components—whether copper switches or DNA repair systems—is of secondary consequence (see Weitz and Fraser 2001). Note that this breakdown in connectiveness is the predicted outcome following the fragmentation of gene expression networks, as discussed in chap-

ter 1. I return to this theme in later chapters, particularly chapter 14. 2.2.2.2 Late-Life Mortality Kinetics

The implication of figure 2.16b, that mortality rates may slow down at older ages, has been confirmed by several large-scale animal studies. An examination of the age-specific mortality of more than 1.2 million genetically heterogenous medflies showed that the mortality rate (qx) slowed its rate of increase at about day 29 (about 16% survival), slowly rose to a maximum at day 58 (about 0.2% survival), and then decreased until the last fly died on day 172 (0% survival; figure 2.22; Carey et al. 1992). It follows logically that such deceleration and eventual decrease in mortality rates should be accompanied by an increase in life expectancy, and that is what was found. The life expectancy (ex) of the 1.2 million flies at age 0 was 20.9 days. By age 50, it decreased to a minimum of 6.7 days before increasing to a maximum of 24.8 days at age 86, and decreasing to a final value for the last animal of 6.5 days at age 165 (Carey et al. 1992). The same deceleration of mortality rates was observed in genetically inbred Drosophila strains (Curtsinger et al. 1992; Fukui et al. 1993), in genetically heterogenous (but not genetically inbred) nematodes (Brooks et al. 1994), as well as

0.16

0.12 Mortality rate

Figure 2.22 Mortality rates of a population of 1.2 million medflies maintained in cages of 7200 animals each. The age-specific mortality rates initially rose exponentially with age but then leveled off at about 20 days of age (16% survival), slowly increased to a peak at 58 days of age (0.2 percent survival), and declined thereafter. (Redrawn from data in Carey et al. 1992.)

0.08

0.04

0

0

20

60 40 Age (days)

80

100

2.2 Life Tables and Survival Curves

tality rate for this cohort of Swedish females increased every year to a maximum at age 72. After that point, the rate of change in the mortality rate began to drop and continued to do so through at least age 95. More extensive data of figure 2.24 shows that the mortality rate actually decreases sharply in very old humans, and appears to follow a special logistic form of the Gompertz curve which takes this mortality deceleration into account. If we define aging as being an increased probability of dying with the passage of time, then these data compel us to conclude that human aging begins at about 10 years of age (figure 2.16) and ends at about 110 years of age (figure 2.24), with the inflection point of the change being 72 years (figure 2.23b). What factors might account for these unexpected observations? A recent study attempted to identify the physiological variables that affect the functional capacities of individuals and that thus underlie the late age-specific mortality (Manton

in wasps, yeasts, and automobiles (Vaupel et al. 1998). What about humans? Evidence supporting the suggestion of figure 2.18b can be found in the fact that multiple studies, done in developed countries with good living conditions and good data, show that the mortality rates among old (80+ years) humans have been steadily decreasing at an approximate rate of 1.5% per year since the 1960s (Kannisto et al. 1994; Manton et al. 1994), such that they approximate a high but constant value. A good illustration of this point is shown in figure 2.23. Panel a shows the Gompertz curve for Swedish females during the last third of their life span. Again, the age-specific mortality rate begins to decrease, apparently during the mid-80s. This independent finding (among others) confirms the data of figure 2.16b. In addition, when one plots these data as the year-to-year change in the age-specific mortality rate, then one obtains the plot in figure 2.23b. Note the mor-

a

b

Age-specific death rates for Swedish female cohorts born between 1871 and 1875

Life-table aging rates (LAR) for Swedish female cohorts born between 1871 and 1875

0.12

1.00

0.11

0.50

0.10 LAR (per year)

Death Rate (per year)

49

0.20 0.10 0.05

0.09 0.08 0.07 0.06

0.02

0.05

0.01

0.04 60

70

80 Age

90

100

60

70

80

90

100

Age

Figure 2.23 (a) Age-specific death rates for Swedish female cohorts born between 1871 and 1875. Only the portion of the life span between ages 65 and 100 is represented. Note the mortality deceleration apparently beginning in the mid-80s. (b) The annual change in the age-specific death rates for the same cohorts as the left panel. (The LAR is a measure of the relative mortality increase or decrease with age.) The confidence bars are two times the estimated standard errors of the mean annual change. Note that the age-specific death rates continue to increase until the age of 72, after which time their annual rate of increase decreases, leading to the result shown in the left panel. (After Horiuchi and Wilmoth, 1998.)

50 Chapter 2 Measuring Age-related Changes in Populations

Humans 1

1.0

Death rate

2

3

0.1 80

90

100

110

120

Age (years)

Figure 2.24 Age-specific death rates from ages 80 to 120 for human females. An aggregation of data from 14 countries with reliable data was used to construct the observed curve (heavy black line). Although the data extended to age 122, they are too sparse above 110 to yield reliable conclusions and are omitted here. The line labeled 1 is the Gompertz curve best fitting the data from ages 80–84. The line labeled 2 is the logistic curve that best fits the entire data set. The line labeled 3 is a quadratic curve fit only to the data above age 105. (After Vaupel et al. 1998.)

et al. 1995a). The authors concluded that the deceleration in late-life mortality is brought about in part by the earlier death of frailer individuals (e.g., the heterogeneity hypothesis), and in part because of changes (perhaps socially caused) in the age dependency of the important physiological parameters contributing to an individual’s functional capacity (e.g., the individual-risk hypothesis).This last point was verified by a study that analyzed mortality rates due to specific diseases and found that people older than 75 years showed mortality decelerations (Horiuchi and Wilmoth, 1998). In effect, older people today are healthier and display significantly less morbid-

ity or disabling conditions than did people of the same age a generation ago, and this change has multiple causes. Both demographic and gerontological explanations have some validity. The Manton et al. (1995a) study is most interesting, if only because it is one of the first to endeavor to replace the mere passage of time with documented alterations in the underlying physiological variables. I discuss this study again in chapter 6 and deal with its further implications in chapter 15. The inapplicability of the Gompertz plot to late-life mortality kinetics does not in any way invalidate the use of the Gompertz plot or of values derived from it, such as the MRDT, in the interspecific comparisons we have discussed. This is because the MRDT is based on the linear portion of the Gompertz plot during the time before the deaths of the most long-lived members of the population under consideration. (Note that this linear portion of the curve covers the deaths of about 85% of the popultion.) A recent analysis concluded that the Gompertz model gives a good approximation of the adult age-related mortality and generates a good fit between the expected and observed values of the maximum life spans for many different species (Finch and Pike 1995). Additionally, an analysis of genetically selected long- and short-lived strains of Drosophila showed that the Gompertz model could accurately summarize environmental and genetic alteration of longevity, despite the theoretical expectation of its failure to fit the observed data at very late ages (Nussbaum et al. 1996). 2.2.2.3 Other Considerations

The Gompertz curve assumes that the level of nonsenescent deaths in a population, if there are any, is very low. As the frequency of such accidental deaths (age-independent mortality) increases, the result is a biphasic curve dominated by the constant mortality rate early in life and by the exponentially increasing mortality later in life. The shape of any particular survival curve is dictated by the relative contributions of both types of mortality to the survivorship of the population (figure 2.25). Mathematical derivations of the

(b)

–2 –3 –4 –5 –6 –7 –8

8

7 6 5

4

3

2

Number surviving, per 1,000

(a)

Natural log of mortality rate per day

2.2 Life Tables and Survival Curves

1

–9 –10 –11 –12 0

51

1,000 900 800 700 600 500 400 300 200

1 2 3 4 5 7

100

6

8

0 200

400 600 800 1,000 1,200 1,400 Age (days)

0

200 400 600 800 1,000 1,200 1,400 Age (days)

Figure 2.25 The relationship of survival curves to the Gompertz mortality curves, as depicted by the effect of various amounts of nonsenescent deaths on the shape of mortality curves (a) and on survival curves (b). Lines with the same numbers in both panels represent the same populations. Population 1 represents the effects of only Gompertz mortality, where qx increases exponentially with age. As age-independent mortality increases in populations 2–8, the survival curves progress from rectangular (1) to exponential (8), and the mortality curves approach a constant value. (After Sacher 1977.)

Gompertz equation, as well as conceptually different equations, have been developed to fit the data better. Some sets of population data can be described best by a Gompertz plot (which describes an exponential increase in mortality); others can be best fitted by more complex transformations such as a Weibull function (which describes a power or logistic increase in mortality); and there appears to be no obvious regularity as to which equation best fits any given population (Wilson 1994). However, a more extensive analysis of life tables showed that the Gompertz equation generally displays a better fit to the data than does the Weibull or other functions (Gavrilov and Gavrilova 1991), particularly when a twostage model is constructed to account for the decreased mortality characteristic of very old individuals, as discussed earlier (Carey et al. 1992). A theoretical approach to the problem led Gavrilov and Gavrilova (1991, 2002) to consider both the Gompertz and Weibull equations to be special cases of a more general mathematical treatment called reliability theory, which is the study of the processes underlying failure of complex systems. They suggest that systems in which there is a low level of redundancy and a high level of functionality at the start of the time period will fail according to the Weibull function, while systems in which there is a high level of redundancy and a low

level of functionality will fail according to the Gompertz function. They point out that the Weibull conditions mostly describe mechanical or inanimate systems, whereas the Gompertz conditions mostly describe living systems. If so, then the nature of the failure kinetics is likely telling us something about the initial state of the system under study. Demographic analysis of life table data has greatly assisted us in better phrasing our questions regarding the kinetics and nature of age-specific mortality. In turn, these answers have refined our ability to make accurate demographic predictions. 2.2.2.4 A New Measure of Aging

The conventional measures of aging discussed above count the years since birth, and these metrics allow us to describe the aging of human and animal populations. But as described in chapter 6, during the 20th century the average life span in the developed countries has increased, as has the length of time during which the average individual enjoys sufficiently good health so as to live (and work) independently. The upshot is that the conventional aging measures describe a (chronologically) aging population which is, at the same time, healthier and more independent than might be expected from their age. The conventional age numbers no longer mean what they once did.

52 Chapter 2 Measuring Age-related Changes in Populations What is needed is a measure of aging that allows us to measure age as the average number of years left until death. Such a measure has been proposed: the median age of the population standardized for the expected remaining years of life (Sanderson and Scherbov, 2005). The empirical value of this proposed aging index needs to be established. However, computer projections of the numerical value of this measure in developed societies during the 21st century show that it will decrease, indicating a healthier and longer lived population. Should the retirement age increase in proportion to the mean life span, then the agedependency ratio (see chapter 15) will stay more or less constant, and the financial crisis forecast for aging populations (Petersen, 2002) may be forestalled.

The Gompertz plot may be used to determine the age of sexual maturation when physiological data are nonexistent and when the transition from developmental to adult forms is not clearly demarcated. In species that do show such demarcation, such as Drosophila and other holometabolous insects (insects that undergo complete metamorphosis), then the newly hatched adult represents the beginning of the postdevelopmental stage, and the Gompertz plot should have its lowest initial vulnerability at this point in time. Not only is this definition based on objective quantitative data, but it is also in full agreement with the theoretical relationships posited to exist among reproduction, important life-history features, and evolution (Rose 1991).

2.4 2.3 Distinguishing between Development and Senescence In attempting in chapter 1 to define “aging” and “senescence,” it became apparent that most of the definitions contain no objective criteria that would allow a naive onlooker to unambiguously distinguish developmental processes from the processes of aging and senescence. An inspection of figure 2.16b suggests a simple objective criterion: Inasmuch as development is an adaptive process that enhances the functional capacities of the system, development may logically be considered to have ended when age-specific mortality is at a minimum. This point is selected because it represents the age of maximum functional fitness of the population. The organisms will never be more healthy than they are at this point. Subsequent increases in the age-specific mortality rate of the system may be reasonably attributed to the onset and continuation of senescence processes. Thus, in the human data of figure 2.15b, development ends and senescence begins just before puberty. Note that the MRDT in mammalian populations is usually determined beginning at the age of puberty; that is, the developmental period is omitted. As discussed above, human aging begins to slow down at age 72 and ends at age 110.

Are There Mathematical Limits to Longevity? Both the lay and the professional literature have commonly asserted that there is a species-specific life span limit. The existence of such a limit seems so obvious at first that one might be excused for thinking that such a commonsense conclusion needs no further proof. After all, no human on this planet has yet lived to the age of 123 years. But let’s consider this proposition. It is tantamount to saying that for each species there is a particular age that possesses certain unique properties such that no member of the species can live through that particular time period. Accordingly, if the maximum life span for humans is now 122.5 years, then it is impossible for another human being in the future to live 122.5 years and 1 day. Once reworded, it should be apparent that the statement is not only nonsensical, but it also conflicts with the empirical data already presented that show that the age-specific mortality rate decelerates and even decreases for very old individuals of several species, including humans. If the late-life mortality rate is constant, then the two constraints on the maximum age that any single individual might attain are (1) the size of the initial population and (2) the slope of the Gompertz plot (or the MRDT, as discussed

2.4 Are There Mathematical Limits to Longevity?

earlier). Of these two, the latter is the more important and has by far the larger effect on maximum age (Finch and Pike 1995). If the probability of some rare event (such as living some extraordinary length of time) is constant, then increasing the size of the initial population will increase the number of individuals who will attain that event. Thus, the maximum age attained is a function of population size. Maximum life span is, to a large extent, a probability function. Increasing the MRDT value (i.e., decreasing the slope) will also alter the maximum life span because it means that a larger number of individuals will survive to some arbitrary age. The decreased mortality rate increases the effective size of the population potentially capable of attaining that arbitrary age. Finally, the concept of maximum life span conflicts with the idea that senescence is independent of time. There is no singular age beyond which it is impossible for any individual to survive. Therefore, there is no reason to believe that there is a mathematical upper limit to life span, at least in species in which the late-life mortality rate decelerates and becomes a constant. (A more detailed discussion of this concept may be found in Wilmoth et al. 2000). The evidence against the existence of a limit to longevity includes not only the logical arguments given above but also the historical record. Oeppen and Vaupel (2002) examined 17 published estimates of the theoretical maximum human longevity. Of these, 13 estimates were falsified by actual experience in at least one society within an average of 5 years after publication (and sometimes even before the prediction was actually published). Although this record does not logically rule out the existence of some future limit, it does not inspire a great deal of confidence in the assumptions and calculations underlying these predictions. In addition, the ongoing analysis of current mortality rates yields no hint that we are approaching a hidden limit (e.g., no clustering of maximum cohort ages just below some uncrossable threshold).

53

Yet even as we acknowledge the probabilistic nature of the maximum life span, we must also acknowledge the fact that species do have characteristic, if not maximum, life spans. Flies, mice, cats, horses, people, tortoises, and bristlecone pines represent a real continuum of mean and maximum longevities that we have to accommodate. And we can do so by remembering that these apparently fixed values are outcomes of each species’ particular combination of Gompertz parameters, its initial vulnerability, its MRDT, and its population size. And these parameters are not immutable. An understanding of these late-life mortality kinetics is important because public-policy decisions often turn on demographic predictions, including that of the life expectancy of a particular portion of the population. If there were reason to believe that the human life span had a fixed upper limit, then it would logically follow that the continued increase in the mean life span (see figure 2.14) would one day approach the unalterable maximum life span. This rectangularization of the survival curve (the transition from curve B to curve A in figure 2.11) would compress mortality because people would remain healthy for the greater part of their life, only to succumb to degenerative diseases and die within a relatively short time period (Fries 1980). Such a phenomenon would result in a decrease in the proportion of chronically ill people in the population. This potential decrease has been used to argue for a reduction in the amount of private and public resources spent on treatment of and research into the late-life degenerative diseases. The scientific basis for the argument has been disproven (Schneider and Brody 1983), but this example demonstrates the real-life impact of these supposedly abstract demographic numbers on each of us. There is every reason to believe that demographic projections of longevity will continue to play an important role in ongoing public-policy debates. This example also shows that it is important for all citizens to understand the scientific assumptions underlying such public debate. I return to the issue of aging and public policy in chapter 15.

54 Chapter 3 Measuring Age-related Changes in Individuals

3

Measuring Age-related Changes in Individuals

3.1 Actuarial Analysis of Age-related Changes through Time We may analyze populations to determine whether the individuals within them will survive long enough to have a chance to grow old and age. Populations, however, are composed of many diverse individuals, only some of whom display the expected age-related changes at the expected times. In this sense, therefore, we may conclude that aging is an individual process and must be measured and studied in detail in individuals. The diagnosis of aging may be inferred from the population data, but the study of aging must ultimately refer to its expression in individuals. An excellent review of how to best design experiments to accurately measure aging in individuals can be found in Ingram (1999).

3.1.1 Cross-sectional Studies Almost all the information available regarding age-related changes in animals, human or otherwise, has been drawn from cross-sectional, or “point-of-time” studies. In such studies the variable under investigation is measured for groups of subjects of different ages. The age-related changes are not measured directly; they are inferred from a comparison of the mean values for each cohort. Age-related changes may also be inferred from a regression of the variable on age, made on subjects distributed over the total age

54

span who are measured at about the same time. This experimental design allows us to capture a cross-section of the population values in time— hence its name. Because this procedure is relatively simple and inexpensive, it is a very popular experimental approach. For long-lived species such as humans, this protocol is often the only feasible one. Even when working with a shorterlived laboratory species, such as the rat, a single investigator could hope to do only a dozen or fewer longitudinal studies in his or her lifetime. Thus, longitudinal studies may not be a feasible or desirable approach. However, cross-sectional studies have at least four important drawbacks. First, the cross-sectional approach assumes that the manner in which the average value changes from one age group to the next is an accurate reflection of the change that occurs in one individual with the passage of time. There is no a priori reason this assumption must always be valid. In fact, much of the data to be presented here robustly argue against such uniformity. The second limitation of the cross-sectional approach is that it confounds the effects of environmental changes with the effects of age. For example, starvation affects young children differently from the way in which it affects mature adults. If two such differently aged individuals lived through the same famine, the differences in average values of a particular variable measured after the event might erroneously be ascribed to aging. In this case, the resulting heterogeneity in peak values may really be the result of a general environmental effect, such as an epidemic or a

3.1 Actuarial Analysis of Age-related Changes through Time

55

Height (cm)

3.1.2 Longitudinal Studies 180 (49)

(131)

(215)

(176)

(136) (100)

170

(17)

Weight (kg)

80 (49)

(131)

(215)

(176) (136) (100)

70 (17) 30

40

50 70 60 Mean age (years)

80

Figure 3.1 The regression of height and weight (mean Å standard deviation) in normal males, based on crosssectional data. Numbers in parentheses represent sample sizes. Compare to figure 3.3. (After Shock 1972.)

famine. The age effect shown by the individuals could then simply be a result of the fact that they were at five different ages at the time of the event. In this case there would be no true age effect present, but procedural error would have led us to infer one. The third disadvantage is that the crosssectional approach suffers from the effects of selective mortality. A population of 30-year-old males includes both individuals fated to die at relatively young ages and individuals fated to live a long time, whereas a population of 90-year-old males has been highly selected to include only individuals fated to live a long life. The two populations are not identical in the composition of the individuals within them, and hence comparisons between them may not be valid. The final problem with the cross-sectional approach is that it can provide no evidence regarding the rate of change of a particular variable within an individual, although the planned development of biomarkers (discussed later in this chapter) may alleviate this problem in the future.

The longitudinal method is characterized by repeated measurements of a specific variable(s) on the same subject. Thus the method measures primarily age-related changes in individuals. Such a prospective study makes possible statistical estimates of individual rates of aging for the specified variable(s), once sufficient observations have been collected over a long enough period of time. However, the individual records can be summarized to yield the average difference between groups of subjects of different ages. In other words, the longitudinal data can be reorganized to yield crosssectional data, but the reverse operation is not possible. If the study subjects are of different ages, the cross-sectional data are obtained immediately after this transformation. If the study subjects are all the same age, the cross-sectional data are obtained only when the study runs long enough that data can be collected from the subjects during several different age intervals. Subjects who are members of the same age class are often called a cohort. From a theoretical standpoint, the data from a longitudinal study are more reliable than the data from a cross-sectional study. However, the longitudinal approach is not free of drawbacks. The most obvious disadvantage of this method is the limitation of time and of money. Repeated measurements on a defined group of individuals over a long period of time require long-term commitment by subjects, investigators, institutions, and funding agencies. Such a conjunction rivals the alignment of the planets in its rarity and enhances the value of the studies of this type that have been done (see chapter 8). The use of repeated tests may give rise to a “practice effect,” whereby subjects respond better in later trials than in earlier trials because they have learned the appropriate responses. The practice effect may be more of a problem with psychomotor tests than with more biochemical assays, although the effects of biofeedback on physiological processes cannot be ignored. In addition, the use of institutionalized subjects may render the logistics of a longitudinal study easier, but at the cost of making spurious comparisons

56 Chapter 3 Measuring Age-related Changes in Individuals

Three primary temporal factors may be responsible for chronological changes in a particular variable: age, period, and birth cohort. Crosssectional studies tend to confound age effects with birth cohort effects. Longitudinal studies tend to confuse age effects with period effects. In an attempt to create a study design that would not confound these variables, Schaie (1965) devised a series of three different sequential experimental designs that combined elements of both cross-sectional and longitudinal studies. Combining these three into what he called the “most efficient design” gave rise to a design that did not confound any of these variables. However, its complexity (and cost) soon led to modifications (see Ingram 1999 for review). Today many longitudinal studies take advantage of such modified study designs to more accurately assay their data. Much of the experimental data I discuss in this book does not arise from such complex designs. In the final analysis, then, we use mostly crosssectional and/or longitudinal data and subject them to critical and skeptical inquiry.

3.1.4 Empirical Longitudinal and Cross-sectional Comparisons One advantage of longitudinal studies is that the data may be reconstructed into a cross-sectional format and the validity of the two approaches compared directly for the same set of data. This comparison is very instructive and has been done by Shock (1985) for some of the data obtained from the Baltimore Longitudinal Study of Aging.

Height (cm)

3.1.3 Sources of Confusion and Their Resolution

The following discussion draws from Shock’s observations. Figure 3.1 depicts the cross-sectional regression of height and weight on age in healthy males. The simplest interpretation of these data is that a gradual reduction in height and weight over the ages of 30–85 years constitutes a normal agerelated change. Figure 3.2 presents data from the longitudinal regression of height and weight on age for the same males as in figure 3.1. Individual subjects tend to lose height as they grow older, and individuals younger than 55 years old tend to gain weight, even though the average weight value is falling. For these individuals, the cross-sectional and longitudinal data do not agree. For individuals 55 years and older, however, the two data sets agree. Thus, the cross-sectional interpretation of the height changes is verified by the longitudinal study, but the interpretation of the weight changes is upheld in part and falsified in part. A phenotypic trait such as weight has a large environmental component, and it would not be

180

170

80

Weight (kg)

from an institutionalized population to a healthy normal and mobile population. Even when longitudinal studies have been done on a normal population, it has been recognized that the population is not random but is highly selected with respect to socioeconomic and ethnic properties. The findings of the study should be generalized to the population at large only with caution.

70 (21)

(88)

30

40

(136)

(96)

(93)

(36)

50 60 70 Mean age (years)

80

Figure 3.2 The regression of height and weight (mean ± standard deviation) in normal males during an 8-year period, based on longitudinal data obtained from repeated measurements on the same subjects as those represented in figure 3.1. Numbers in parentheses represent sample sizes. (After Shock 1972.)

3.1 Actuarial Analysis of Age-related Changes through Time

80 Parisians

Weight (kg)

70

60

Kabyles

50

25

35

45 55 Age (years)

65

75

Figure 3.3 A comparison of the effects of environment on age-related weight changes between economically favored Parisians and Kabyles, a North African group leading a primitive life. Note that the Kabyles display only minimal weight gain throughout adult life. (After Bouliere and Parot 1962.)

15 Skin fold thickness (mm)

wise to ignore this factor. How much, if any, of the weight gain observed with age is attributable to environment? Bouliere and Parot (1962) made a cross-sectional comparison between economically affluent Parisians and Kabyles, a North African group with a primitive lifestyle characterized by high energy expenditures and restricted food supply. In the Kabyles, weight changed little throughout maturity (ages 25–55 years; figure 3.3). This lack of change in weight was most likely due to a failure to deposit extra subcutaneous fat during middle age, as demonstrated by the differences in total weight (figure 3.3) and in the thickness of skin folds (figure 3.4). Past the age of 60 years, the lean body weight of human males of both groups appears to decrease markedly (see figures 3.2 and 3.3). Taken as a whole, these data suggest that humans have the ability to gain weight via the deposition of subcutaneous fat, provided that their socioeconomic environment

57

Parisians 10

Kabyles 5

25

35

45 55 Age (years)

65

75

Figure 3.4 A comparison of changes in the iliac skin fold in the same two groups of men as in figure 3.3. Again note the constancy in the Kabyles. (After Shock 1972.)

permits the consumption of extra calories. The catalogue of normal age-related changes must be considered in the context of the environment. This statement amounts to nothing more than the geneticists’ concept that the phenotype is the result of expression of the genotype in a particular environment. The interplay of the genotype with the environment turns out to be of some importance in the study of aging (see chapter 7). In many instances, there is no discrepancy in the results achieved by the two different strategies. Figure 3.5 shows both longitudinal and cross-sectional data for age-related changes in creatine clearance in humans. These data are charted in the same manner as in figure 3.2. An inspection of this graph suggests reasonable agreement between the two sets of data and thus in the conclusions drawn from them. An implicit assumption in the studies described here is that they represent universal traits among the individuals of the species. A more detailed reexamination of the problem led to a different conclusion—namely, that individual variability confounds these assumptions. Consider the case of creatinine clearance (figure 3.6). The later longitudinal study demonstrated that both males and females show similar declines in

58 Chapter 3 Measuring Age-related Changes in Individuals

Creatinine clearance (ml/min/1.73 M2)

150 140 130 120 110 100 90 80

10

20

30

40

50

60

70

80

90

Age (years)

Figure 3.5 A comparison of cross-sectional and longitudinal age changes in creatinine clearance. The dots represent the mean values for each age decade as obtained from cross-sectional data. The short line segments indicate the mean slope of the change in creatine clearance, as based on the longitudinal data for the indicated time spans. Note that the two sets of data agree. (After Rowe et al. 1976.)

creatinine clearance as measured by cross-sectional studies. It further showed a coincidence between the longitudinal and cross-sectional data for males. A reasonable deduction from these data would be that a decline in creatinine clearance should be observed in all humans older than age 35. However, the population is quite heterogeneous. The individual longitudinal displays of serum creatinine clearance show that some individuals

exhibit large and rapid decreases in this trait (figure 3.7a), while others show only small decreases (figure 3.7b) or no change at all (figure 3.7c). An analysis of the entire test population shows that substantial proportions of the test population are significantly different from one another (table 3.1). Only about 58–71% of the study population showed a decrease in creatinine clearance rate during a 10-year period. Between 29 and 42% of the population showed no change. This finding is extraordinary and illustrates how much individual variability may be hidden within normal statistical procedures. Usually, we would have assumed that the agreement of cross -sectional and longitudinal data indicated universality of the trait. We now see that the assumption is not justified and that the only factor we can count on is that most traits will display significant individual variability. These findings in humans are paralleled by findings in other species (see, e.g., Draye and Lints 1995). The next comparison confirms individual variability. Figure 3.8 summarizes cross-sectional data from nine different studies for maximum oxygen (O2) uptake (see Norris and Shock 1974 for details). These data are an indirect measure of the maximum amount of metabolic work that an individual can do. Although the absolute values are different (perhaps as a result of methodological differences), the overall age-associated change observed in both sexes seems to follow the

Figure 3.6 Cross-sectional and longitudinal creatinine excretion values by age and sex in subjects of the Baltimore Longitudinal Study on Aging. Numbers in parentheses represent sample sizes. (From unpublished data of Baker and Frozard, Gerontology Research Center, National Institute on Aging.)

Creatinine excretion (mg)/24 hrs

2,100 1,900

Male longitudinal Male cross-sectional Female cross-sectional

(180) (251)

1,700

(246)

1,500

(192) (162)

1,300 (60) 1,100 900 700

(105)

(49)

(195) (73)

(100)

(54) (101) (24)

500 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 Age cohort (years)

(c)

Creatinine clearance (cc/min)

(b)

Creatinine clearance (cc/min)

(a)

Creatinine clearance (cc/min)

3.1 Actuarial Analysis of Age-related Changes through Time

1

180 120 60 0

4 5

3

6

Figure 3.7 Individual longitudinal displays of serial creatinine clearances plotted against age for 18 representative subjects from the Baltimore Longitudinal Study of Aging. (a) Six subjects followed for 8–14 years. These subjects showed significant decreases in creatinine clearances. (b) Six subjects followed for 11–22 years and showing small but signficant decreases in creatinine clearances. (c) Six subjects followed for 15–21 years and showing no decrease in creatinine clearances. (After Lindeman et al. 1985.)

2

1

60

2

Significant decreases in serial creatinine clearances (negative Bcr > 2 cc/min)

180 120

6 4

3 5 Small but significant decreases in serial creatinine clearances (negative Bcr > 2 cc/min)

0

180 1 2

120 60

59

6

3 4

5 No decrease in serial creatinine clearance (negative Bcr > 2 cc/min)

0

40

45

50

55

60 65 Age (years)

Table 3.1 Percent Change in Creatinine Clearance over 10 Years in 412 Male BLSA Subjects % of individuals showing indicated level of change in creatinine clearance Age cohort

No change

≤ 10%

≤ 20%

≥ 21%

20–29 30–39 40–49 50–59 60–69 70–79

42 39 42 30 29 31

27 29 29 26 22 7

18 19 23 42 37 31

13 13 6 2 12 31

Source: Data courtesy of George T. Baker III and James Frozard, Gerontology Research Center, National Institute on Aging. Note: Mean change from third to eighth decade equals 31%.

same pattern. The low value in childhood increases rapidly to a peak value in the teens and early 20s. This increase is succeeded by a slow decline until the 40s, followed by a more rapid decline until the minimal values of childhood are once more attained. The heterogeneity of the

70

75

80

85

results of several studies suggests that O2 uptake is characterized by a great deal of individual variation. This suspicion is partly confirmed in figure 3.9, which presents longitudinal studies of maximum O2 uptake for two individuals over the age span of 35–87 years. Both individuals exhibit an age-related decrement in this variable, but the patterns are very different. Dr. Robinson displays a gradual, almost constant decrement in both factors throughout his life span; Dr. Dill shows only minimal age-associated changes until after age 65. This pattern would not have been predicted on the basis of cross-sectional data alone. Both men were very interested in exercise and sports, and their physical fitness was reportedly considerably better than that of most of their age-peers. Thus, the differences are not due simply to the effects of physical conditioning and training. Taken together, these data suggest that humans will undergo an age-related decrement in an important physiological factor such as maximum O2 uptake, but that the great latitude in the individual pattern of this decrement suggests that

Maximum O2 uptake (liters/min)

60 Chapter 3 Measuring Age-related Changes in Individuals

▲

▲

▲

2.0

▲ ▲

1.0

125 Maximum ventilation volume (liters/min)

Figure 3.8 A comparison of the agerelated changes in maximum oxygen uptake and in maximum ventilation volume, as observed in nine different studies of both men and women. Note that the several studies, although very heterogeneous, conform to the same general pattern. (After Norris and Shock 1974.)

3.0

100 ▲ ▲

75

▲ ▲

▲

50

25

0

factors other than age significantly affect this variable. One other weakness affects cross-sectional and longitudinal studies equally: the dependence on time as a measure of the aging process. Although the idea may seem odd at first, using time units to measure age may be an imperfect compromise between accuracy and convenience. A good illustration of this is the example shown in figure 3.10a, which depicts the growth rates of individual children between 5 and 18 years. Every individual reaches a peak value at some point, and the shapes of the curves are similar. Yet it is obvious that the timing of the pattern is different in the five individuals depicted. Knowing a child’s age would not allow one to make an accurate statement concerning that child’s growth rate. In this instance, a chronological measure of age conveys very little information.

10

20

30

40 50 Age (years)

60

70

80

A better measure is shown in figure 3.10b, where the curves have been arranged so that their points of maximum growth rate coincide, and the other points for each individual are plotted as deviations in time from this event. This procedure suggests that measuring age by the passage of years may not be as meaningful as measuring age by the passage of certain significant events. We already use this concept in other areas—for example, when we talk about individuals passing through developmental stages that are functionally but not chronologically defined. I return to this concept of the event-dependent nature of aging in the discussion of biomarkers later in this chapter and in chapters 6, 9, 12, and 13. This brief survey of experimental design should leave you with the impression that aging is a highly individual process that requires the intelligent interpretation of both longitudinal and

210 200

Maximum heart rate (liters/min)

190 180 170 160 150 140 130 120 110 100

Figure 3.9 Differences in the longitudinal patterns for maximum heart rate and for maximum oxygen uptake of two individuals, D. B. Dill and S. Robinson. (After Horvath 1981.)

4.0

Maximum O2 uptake (liters/min)

3.5 D. B. Dill S. Robinson

3.0 2.5 2.0 1.5

Bicycle maximum values (treadmill maxima are 20% higher)

1.0 40

35

(a)

45

9

55 60 65 Age (years)

70

Individual curves Mean curve

8

Growth rate (cm/year)

50

75

80

85

90

(b)

7 6 5 4 3 2 Maximum rate

1 6

8

10 12 14 16 Age (years)

18

2 4 6 –6 –4 –2 Time of maximum growth rate

Figure 3.10 The relationship between individual growth curves of adolescents and the mean growth curve. (a) Curves plotted as a function of chronological age. Note that the average curve, which is based on the crosssectional data, does not adequately describe any of the individual growth curves. Compare these empirical data to the hypothetical situation shown in figure 3.1. (b) Individual growth curves plotted as deviations from each person’s age of maximum growth. Note the excellent agreement between the individual and mean curves in this instance. (After Tanner 1955.)

62 Chapter 3 Measuring Age-related Changes in Individuals cross-sectional data in order to draw reasonable and testable conclusions.

3.2 Distinguishing Disease and Environmental Changes from Age-related Changes There are several different and long-standing schools of thought regarding the relationship of diseases to aging (Blumenthal 2002). One group, perhaps expressing what has been the traditional medical model of geriatrics, views aging as the sum of the diseases to which we eventually succumb. In this view, aging is a disease. This point of view may have made sense in the past, when most people died young (see figure 2.14), but it is no longer tenable now that we understand that people age even in the absence of disease. A second point of view, perhaps arising as a reaction to the medical model, is characteristic of what we may call a gerontological viewpoint. This model disclaims any fundamental connection between diabetes or cancer or cardiovascular disease or other such age-related pathological syndromes and the processes of ordinary aging, except to assume that the increased incidence of these pathologies among the older members of the population is probably due to the fact that certain normal, age-related changes are precursors to or precipitate disease, a concept that I will discuss in some detail. Consequently, much energy has been invested in the effort to distinguish “normal” aging from “abnormal” aging, by which is meant aging in the absence or the presence, respectively, of disease. Adoption of this gerontological model in recent decades has allowed us to progress beyond the narrow study of geriatric disease to the identification, characterization, and manipulation of the aging processes that take place in the absence of overt disease. An evolving third point of view suggests that there is a close relationship between aging and disease—that we may consider them two different aspects of the same process. This idea is not a return to the narrow geriatrics view of aging; rather, it is a broadening of the gerontological

model. The study of age-related diseases is intimately linked to our increased knowledge of the aging processes. The linkage lies in the fact that the existence of these age-related diseases empirically delineates aspects of the body’s normal physiology and cell biology that are prone to failure as a consequence of the aging process. A detailed molecular and genetic knowledge of the particular disease syndrome associated with such failures allows us to classify the weak points of our bodily machines, to sort out their modes of failure, and to try to determine which intervention strategies might delay or prevent these failures. Inherent in this proposition is the idea that using clinical tours de force to alleviate the symptoms of disease in older individuals is not as effective as marrying the insights of basic research with the details of clinical knowledge to prevent the onset of disease, and thereby increase certainly the quantity (and probably the quality) of life. The aging process is not the sum of our diseases, nor is it totally divorced from our diseases, but it sets the stage for the possible appearance of particular syndromes of failure. In fact, Fozard et al. (1990, p. 126) end their review of the future of longitudinal studies by concluding that “an adequate description of aging must integrate an account of disease within it.” This third viewpoint of aging has been cogently expressed by Holliday (1995), who views the effects of aging as bringing about a condition of incipient disease. As we age, a variety of deteriorative changes set in. These changes occur with some synchrony, but when deterioration of one organ system becomes more obvious, disease is diagnosed. Such disease is surely the result of aging to some degree, but it also accelerates the deteriorative changes in that organ system and thus contributes to the continued aging of that system and that individual. If an individual died before the diagnosis of the disease, some people would consider her to have died from normal aging; if she died after the diagnosis of the disease, most people would consider her to have died as a consequence of her disease—of abnormal aging. When phrased in this manner, the distinction between the two seems artificial. Aging, then, is not a disease but a cluster of incipient

3.2 Distinguishing Disease and Environmental Changes from Age-related Changes

diseases affecting the functioning of a variety of tissues and organs in more or less predictable ways (R. Holliday, personal communication). The results of a recent conference (see G. M. Martin et al. 1995) suggest that this integrative point of view is becoming more widely accepted by scientists (myself included). I delve further into this point of view in this chapter and elsewhere in the text, particularly in the discussion of human aging in chapter 5 and chapter 9. Despite the intellectually close relationship between disease and aging implicit in the integrated point of view, I do not attempt to give detailed descriptions of age-related diseases here. It is difficult enough to describe the usual progress of aging and senescense, simultaneously accounting for the usual individual heterogeneity in aging, without obscuring the main story with the diversionary tale of disease states. Diseases offer the opportunity to identify potential failure points; once I have identified them, I turn my attention elsewhere. Of course, not all diseases are agerelated, and we have to distinguish between agerelated and time-related diseases. In addition, not every person suffers from the same types or sequences of diseases, and we should be able to account for this phenomenon as well.

3.2.1 Diseases Associated with the Passage of Time The phenomenon of hair graying with age is documented in various mammalian species, including humans. Almost all humans have some gray hair by the time they reach their late 30s or early 40s, and all have gray hair by the time they reach their early 60s (Lamb 1977). No difference has been observed between men and women nor between people of different hair colors. Not all hair is the same; the hair of the head, beard, and pubic area is different from the hair of the eyebrow and the armpit. The former set turns gray; the latter set is much more resistant to graying, especially in women (Kligman et al. 1985; also data of M. Isaki, as presented in Balin 1994b). Histological examination has shown that the loss of pigment is associated with a loss of tyrosinase activity and a

63

production of imperfect melanin granules in the hair shaft (Orentreich and Orentreich 1994). Hair follicles undergo cyclic periods of growth and hair production followed by resting periods of little activity. It is the melanocytes within the hair follicle that provide the pigmentation and their cellular activity and longevity are also under cyclical control. Hair follicles contain a reservoir of melanocyte stem cells which can replace the pool of differentiated pigment cells as they die. There is no human data bearing on the failure of the hair follicles to produce functional pigmentation, and so we have no basis for considering gray hair to fulfill the CPID criteria for an age-related change. It has, in fact, been traditional to view hair graying as a non-harmful loss of function. However, recent findings in the mouse (see Steingrimsson et al. 2005 for review) provide evidence for a mechanism which, if applicable to humans, might alter that traditional view. Various experiments show that at least two regulatory proteins (Pax3 and Mitf) modulate the balance between the stem cell maintenance and differentiation in the mouse. Hair graying is due to a failure of melanocyte stem cell maintenance and may arise due to changes in the follicle microenvironment that indirectly alter the balance between the regulatory proteins, and/or might alter the expression of the signaling proteins (e.g., Kit1) that guide the migration of melanocyte stem cells to the appropriate region of the follicle cell where they may be incorporated into the growing hair. So it is possible to view hair graying as a process which, although harmless in and of itself, is nonetheless indicative of a tissue specific failure of the growth signals that support stem cell amplification and incorporation into our self-renewing tissues. To the extent that such a signaling dysfunction is widespread in the body rather than just being localized, then to that extent we may view hair graying as one of the first visible signs of stem cell dysfuntion and impending loss of function in vital tissues. The fact that not all mouse strains gray with age (Finch, 1990) is evidence that there exist genetically based differences in the ability to support the melanocyte stem cells, and perhaps other types of stem cells as well. It would be interesting to know if there

64 Chapter 3 Measuring Age-related Changes in Individuals is some correlation between stem cell losses in different tissues. Chapters 12 and 13 provide more details on the consequences of failure to fully maintain both differentiated and stem cells in various tissues. A known example of a time-related disease in humans is polycystic kidney disease. In this disease, clinical symptoms are frequently not seen until the patient is in his or her 50s. Polycystic kidney disease has long been viewed as a timedependent disease, but until recently it was not clear in what ways a time-dependent condition is different from an age-dependent syndrome. Transgenic techniques have been used to produce a mutation in the mouse that gives rise to an animal model of this disease (Moyer et al. 1994). In these afflicted rodents, the normal function of this gene appears to be regulation of the cell cycle (see figure 12.1) in the kidney epithelium. Inactivation of the gene in the mutant animals gives rise to epithelial hyperplasia (overgrowth) via either abnormal activation of the cellular proliferative response or abnormal inactivation of cell death (apoptosis; see chapter 11) in the kidney. Either mechanism will, if given enough time, cause enlarged epithelial cysts. This observation provides a plausible explanation for the apparent time dependency of this syndrome (the hyperplasia has no ill effects until a certain threshold is reached after about 50 years of slow overgrowth). An age-related change, in contrast, might be viewed as one in which the hyperplasia would not be constant but would shift from a normal to an abnormal rate after the age-dependent failure of a cellular mechanism that regulated cell proliferation.

3.2.2 Age-related Changes That Might Precipitate Disease It is increasingly difficult to distinguish clearly between common age-related changes and pathological disease states. Normal, deleterious agerelated changes may be necessary preconditions for the development of abnormal pathology. Let’s consider the case of blood pressure and aging, particularly as it illustrates the diversity of age-

related changes in humans (and probably in other species as well). Normally, blood pressure is proportional both to cardiac output and to vascular resistance to blood flow. However, this simple relationship is confounded by the existence of a large number of interacting variables, of which age is prominent (see chapter 9). The major change that often occurs with aging in the arterial wall is a slow, continuous, and symmetrical increase in the thickness of the inner layer of the artery, the intima. The thickening may be a response to a minor injury to the cells of the intima and may even involve the expression of particular oncogenes present in the cell. In any event, this thickening initially begins with a gradual accumulation of smooth-muscle cells and the subsequent proliferation of both these cells and the adjacent connective tissue. The thickening of this layer is coupled with the progressive diffuse accumulation in the layer of cholesterol and other lipids. This process is mediated by dynamics of blood flow, surface geometry, and heart rate; as well as by environment. More recently, it has been shown that certain genetic alterations affecting lipoprotein metabolism can drastically alter the rate of intima thickening and endothelial injury. We are not concerned with these alterations at present (see chapter 5 for discussion). The net result of these normal agerelated changes is a gradually increasing rigidity of the arteries, as suggested by the data of figure 3.11, which shows decreased elasticity of the aorta with age. However, this decreased elasticity may result in an increased systolic blood pressure. In turn, this increased blood pressure is known to be a major risk factor for several vascular disorders, most notably cerebrovascular disease or stroke (Rowe and Minaker 1985). In this manner a normal age-related change increases the risk of serious morbidity and/or mortality and thereby obscures the definitive line once arbitrarily drawn between aging and disease. This scenario of the normal age-related changes in arteries implies that an increase in blood pressure with age is a universal attribute. Is this correct? The cross-sectional data of figure 3.12 suggests that it is. However, analysis of the data of figure 3.13 shows that although most

3.2 Distinguishing Disease and Environmental Changes from Age-related Changes

65

Percent volume increase

200

Figure 3.11 Age-related changes in the ability of the human thoracic aorta to expand when placed under standard pressure in vitro. (After Kohn 1977.)

100

100

50 Age (years)

Blood pressure (mm Hg)

0

210 200 190 180 170

Males

First survey, 1954 Second survey, 1958 Third survey, 1964

Females

160 150 140 130 120 110 100 90 80

Systole

Diastole

70 60

10 20 30 40 50 60 70 80

10 20 30 40 50 60 70 80

Age (years)

Age (years)

Figure 3.12 Age-related changes in the systolic and diastolic blood pressure in a Welsh population. Note how well the three repeated measurements agree. (After Miall and Lovell 1967.)

(about 75%) individuals show increases in both systolic and diastolic blood pressure, a small number (about 12%) show no change, and a comparable number (about 13%) even show a decrease in systolic blood pressure with advancing age. This finding is consistent with the information shown in table 3.1.

One could interpret these data as suggesting that the arterial changes described here are not normal age-related changes because they are not universal within the species. This interpretation would then logically force the conclusion that the differences between the individuals might be due to differences in the individuals’ diet, health

Percent of individual showing change

66 Chapter 3 Measuring Age-related Changes in Individuals

20

Systolic: Mean 0.59 Standard deviation 0.54

15

10

5

0

Percent of individual showing change

30

Figure 3.13 A distribution of changes in blood pressures in individuals of different ages. (After R. E. Harris and Forsythe 1973.)

Diastolic: Mean 0.42 Standard deviation 0.30

25

20

15

10

5

0

–0.4

0

0.4

0.8

1.2

1.6

Change in blood pressure (mm Hg/year)

status, and other characteristics. In other words, the arterial changes might be environmentally induced pathologies. Some comparative cultural studies support this position, such as those by Appel et al. (1997) or McGuire et al. (2004), which demonstrate that people with hypertension can control their blood pressure by making appropriate lifestyle modifications. Alternatively, one could interpret the same data as suggesting that the population is polymorphic for this trait: Most individuals will display this trait of an agerelated increase in blood pressure, and a small proportion of the population is genetically and physiologically different and capable of mobiliz-

ing various mechanisms to compensate for the normal increase in arterial stiffness. Evidence to support this point of view is presented in chapters 5 and 9. This interpretation further suggests that these asclerotic individuals may constitute all or some of the long-lived fraction of the population, as evidenced by part E of the Gompertz curve depicted in figure 2.15. This conclusion is consistent with the evidence showing that not all individuals exhibit what are considered to be characteristic and normal age-related changes (see figure 3.7 and table 3.1). In addition, data now show that centenarians, relative to the general population, have

3.2 Distinguishing Disease and Environmental Changes from Age-related Changes

significantly different frequencies of certain genes that are broadly involved in cardiovascular functioning (see chapter 6); however, the observed genetic differences appear to explain only a subset of the cases of extended longevity. And it is well known that certain pathologies (such as diabetes) and certain environmental conditions (such as unlimited feeding) accelerate the rate of collagen cross-linking in arteries and hence increase blood pressure. Thus, neither of the above interpretations is exclusively correct, and there must be substantial interaction between the two parameters. The classification of arteriosclerosis as a normal or pathological age-related change will, in the final analysis, depend on such difficult interpretation and will always be open to question. It seems much more reasonable to consider increased blood pressure as a common age-related change that takes place in most, but not all, members of the human population and that can, in the presence of contributing genetic and environmental factors, act as a precipitating factor in the etiology of various cardiovascular diseases. But we must bear in mind that manipulating the environmental factors alone can reverse or relieve this age-related change in many individuals (McGuire et al. 2004). A good discussion of the complexity of the mechanisms underlying hypertension appears in Lifton et al. (2001.) Resolving the ambiguities inherent in this interpretation of blood pressure and aging may be of general importance. An evaluation of the data obtained in the Baltimore Longitudinal Study on Aging by Rodeheffer and colleagues (1984) revealed that about half of the generally healthy people enrolled in the study had at least some covert signs of coronary heart disease, perhaps attributable to the mechanisms already discussed. Members of the study population who were free of such pathologies were also capable of maintaining a maximum cardiac output more or less comparable to that of much younger adults (see chapter 5 for a more detailed discussion). This result suggests that the aging process in half the population induced no significant decrement in physiological cardiac function. In this instance,

67

a strict insistence on the principle that age-related changes are identical in every member of the species would require us to eliminate the pathologies of coronary artery disease and its precursor stages from the list of normal age-related changes, or conversely, to conclude erroneously that cardiovascular disease is an integral part of the aging process of every person. Either conclusion is too rigid an application of theory to reality, given the known genetic heterogeneity of the human population and our resulting polymorphic character. Both conclusions run the risk of blinding us to certain fundamental changes of great importance in understanding the biology of aging. The best approach is to recognize that we are a heterogeneous and polymorphic species and that we must discard the notion of universal agerelated changes, particularly when the genetic and phenotypic evidence suggests the existence of a polymorphic trait. If the disease pattern has accurately identified a polymorphic trait, then longitudinal investigations should reveal the existence of particular types or degrees of age-related changes in predisease individuals but not in individuals not fated to display the disease syndrome. Such appears to be the case with arteriosclerosis. The increased rigidity of the arterial wall over time, as described already, appears to be a normal age-related change. These structural changes bring about poorly understood fluid dynamic changes in blood flow. Whether these hemodynamic alterations are conducive to the eventual deposition of plaques appears to depend both on the individual’s diet and exercise regime and on his or her genetic background, which establishes that person’s capability of metabolizing, transporting, and excreting fats such as cholesterol. The phenotype of coronary artery disease therefore depends in part on a sequence of normal age-related changes that tend to put the individual at greater risk, in part on the individual’s environmental situation, and in part on the individual’s genetically determined physiological response mechanisms. A similar situation pertains to age-related changes in the skeletal system; the relevant discussion in chapter 5 is another attempt to come to grips with this difficult problem.

68 Chapter 3 Measuring Age-related Changes in Individuals 3.2.3 Environmental Changes That Modulate Aging We have discussed in general terms the environmental plasticity of longevity (see chapter 1). Given our discussion of mortality kinetics in chapter 2, it should be clear that, for a cohort of long-lived creatures with negligible senescence, the environment determines the mortality of individual organisms. A more detailed account of specific examples is needed to discern the multiplicity of mechanisms by which environmental conditions affect aging and senescence. The bristlecone pine, which has a documented maximum life span in excess of 5000 years, is one of the longest-lived organisms in the world. But such longevities are attained only by trees that live in the harsh, windswept peaks of the White Mountains in California. Individuals of the same species that live in more protected environments have substantially shorter lives, on the order of about 1000 years. This life span is still extraordinarily long by any standard, but the 80% reduction in longevity is striking. The harsh environment is conducive for extended longevity because it results in fewer fungi parasitizing the tree and smaller amounts of inflammable underbrush leading to accidental death (Finch 1990; La Marche 1969). The shortness of the growing season may also play a role, although much evidence suggests that these long-lived perennial trees have an apparently unlimited, or at least a very large, capacity for continuous cell division in their meristematic growth zones (Westing 1964). Perennial trees such as the bristlecone or the beech, or other angiosperms such as the giant saguaro cactus, grow more or less continuously. Their increased size increases their vulnerability to exogenous risk factors, which increase their risk of sustaining vital damage. Such risk factors include the accumulation of underbrush that might harbor insect pests and/or serve as fuel for fires ignited by lightning strikes, the increase in the probability of severe wind and ice damage as the trees grow bigger and offer more resistance to the wind, the cumulative structural damage caused by commensal animals, and so forth. The longer these organisms live and are exposed to

these environmental insults, the greater the probability that the accumulated structural injuries will someday result in a mortal wound. The mechanical senescence suffered by these long-lived plants is environmentally induced (Finch 1990). An environmental effect common in organisms that display gradual senescence is the cumulative exposure of the organism to a ubiquitous environmental component that acts as a toxin for that species. The effect of this exposure often would be to gradually induce physiological dysfunction after a physiological threshold was exceeded. One example of such an effect in humans is the expression in later life of pathologies related to the accumulation of ultraviolet rays, such as the development of a skin cancer or the accelerated aging of the skin as a result of sun exposure. All else being equal, older individuals have a total exposure to the ultraviolet rays of the sun greater than that of younger individuals; thus, older individuals have a greater probability of having accumulated a level of UV-induced damage sufficient to bring about actinic aging or damage to the skin. This effect is so common that dermatologists examine only skin from usually protected areas, such as the buttocks, when they are trying to assess the effects of aging on skin structure. In this context, then, we can regard UV rays as an environmental toxin. The steps taken recently to curtail the production of common chemicals that are capable of destroying the ozone layer of the atmosphere and of increasing the UV flux at earth’s surface can be viewed in part as an anti-aging measure. Social insects such as the honeybee are a wonderful example of how environmental and developmental factors can interact to affect longevity in organisms that display rapid senescence (see Winston 1987 for historical references). Worker bees are sterile females. Those born in the winter have a mean longevity of about 140 days, although some individuals may live as long as 320 days; those born in the summer have a mean longevity of only about 15–38 days. Young worker bees of either seasonal group spend their first few weeks in the hive, attending the queen and nursing the brood. They then shift their activities to the field, where they forage for food. This shift

3.2 Distinguishing Disease and Environmental Changes from Age-related Changes

is accompanied by large increases in juvenile hormone, changes in the brain gene expression patterns, decreases in vitellogenin levels, atrophy of the hypopharyngeal glands, decrease in immunocompetence, and the onset of foraging flights. These changes are linked in the sense that the loss of the major zinc-carrying vitellogenin protein deprives the immune system of a major cofactor, or that changes in brain activity bring about changes in behavior. In winter bees, these physiological changes and the role change from nursing to foraging are delayed until late winter or early spring. The winter bees also have less energetically demanding hive duties than do summer bees, as well as very low titers of juvenile hormone and correspondingly high levels of vitellogenin levels. Thus, senescence and death in the worker bee are not strictly determined by age, but are linked instead to the hormonally induced onset of foraging, with its increased risk of mechanical damage (broken wings, for example), energy depletion, loss of immune function, and subsequent senescence (Amdam et al. 2004; Omholt and Amdam 2004). These dietary and environmental changes constitute an epigenetic regulation of aging, for they are capable of changing the gene expression pattern in the honeybee brain, altering its physiology, and changing its behavior so that the honeybee worker shifts from the category of gradual senescence (winter or hive bees) to that of rapid senescence (summer or forager bees; Whitfield et al. 2003). In fact, Whitfield et al. used experimental manipulations that uncoupled behavior and age in these animals and by doing so, revealed that specific mRNA changes in the brain were primarily associated with behavior. The individual brain mRNA profiles correctly predicted the behavior of 57 out of 60 individuals. The honeybee’s use of neuroendocrine and dietary variables presumably acts as a mechanism that the animal uses to coordinate its internal physiology with its environment. This use of diet, social signals, and the neuroendocrine system to epigenetically regulate the gene expression patterns of key integrative organs and thus modulate the progression from one aging stage to the next foreshadows a general mechanism of aging that is evolutionarily

69

conserved in both invertebrates and vertebrates, which I examine in some detail later (see chapters 6, 7, 9, and 14). Other examples could be offered, but the point is that the life span of an organism cannot be viewed as an intrinsic and unchanging quality. Rather, it is the result of a complex series of interactions between the individual organism and its specific environment. In addition to the parameters already mentioned, such interactions often involve developmental modifications.

3.2.4 Developmental Changes That Accelerate or Retard Aging Development may end at the point of the life cycle at which the age-specific mortality rate of the population under consideration reaches a minimum, as described in chapter 2. However, events that take place during the developmental period may well affect the longevity characteristic of the postdevelopmental, or mature adult, phase of the life cycle. It is thus important to our understanding that we be able to characterize the ways in which developmental events may modulate the aging process. Let’s focus again on the honeybee, a classic example of the fact that the presence or absence of particular hormonal changes during development can cause the same genome to adopt one of two alternative developmental paths, each leading to a morph with a characteristic and different life span. Honeybee females can develop either into sterile worker bees, whose longevity ranges from 1 to 12 months depending on whether they are winter or summer bees, as described in the previous section, or into fertile queens that may live for 5 years or more. The problem is to determine the mechanisms underlying this choice of alternative pathways. The evidence, as presented in Winston (1987), leads to the following conclusions. Both queens and workers develop from fertilized eggs. The larva is multipotent with respect to its possible fates for the first 3–4 days of life. During this period, if the workers feed the larva large volumes of food containing high

70 Chapter 3 Measuring Age-related Changes in Individuals concentrations of sugars, called royal jelly, stretch receptors in the larval gut initiate the secretion of juvenile hormone by the corpora allata glands in the head. This hormone brings about a higher growth rate, production of queen-specific proteins, and other responses that enable the larva to develop into a fertile queen. Failure to feed the larva this royal jelly leads to the development of a sterile worker. Experimentally administered subthreshold levels of food and/or hormonal manipulations lead to the production of worker– queen intermediates. Thus, diet-induced neuroendocrine influences shunt development into a path that gives rise to a large fertile queen with a very long life span. Her longevity is limited primarily by the exhaustion of her sperm stores, which induces the worker bees to kill the queen, after which they bring about the development of a new queen. Old queens display few other signs of loss of physiological function. Queens thus appear to exhibit gradual senescence, whereas worker bees exhibit rapid senescence. We marvel at the long life of the queen, but we must also note the long life of her stored sperm, all 5 million of which are stored in her spermatheca on her nuptial flight and all of which drastically outlive the male drones from which they came. In humans, females significantly outlive males, as described in some detail in chapter 8 (see also D. E. W. Smith 1993). This phenomenon is not generally widespread and is not found in many species (Gavrilov and Gavrilova 1991). However, in species that do display sex-based differences in longevity, we may view the process of sex determination as one that sets the fertilized egg on one of two developmental paths, each leading to the formation of a distinctive morph with a characteristic longevity. Dormancy and diapause are periods of slowed metabolism and growth that may occur during development and/or adult stages in many different types of organisms. They are often induced by specific signals characteristic of adverse environmental conditions. Dormany and diapause are quite common in seeds, and one or the other is also found in worms, insects, fish, frogs, and various marsupial and mammalian orders. From an evolutionary point of view, such processes appear

to have the function of delaying the onset of reproduction until the improvement of environmental conditions increases the probability of sucessfully transmitting one’s genes to the next generation. Diapause and dormancy have two general effects on the life span, depending on the species involved: either none, or an inverse relationship between the time spent in dormancy and the subsequent life span (Finch 1990). The genes responsible for the control of larval diapause in the nematode Caenorhabditis elegans are intimately related to the genes that have been independently shown to be involved in the extension of adult longevity in this organism. (I discuss genes, diapause, and extended longevity of C. elegans in more detail in chapter 7.) Environmental and developmental effects merge in the consideration of the intrauterine environment and its effects on life span and senescence. The literature summarized by Finch (1990) shows that both life span and senescence can be altered by the developmental effects of gonadal steroids or by the sex of the neighboring fetus. In mice, females flanked in utero by males are more agressive as young adults and enter reproductive senescence later than females flanked by other females. In another inbred mouse strain characterized by an early onset of autoimmune disease and a short life span, implants of testosterone into the mother at day 12 during pregancy increased the life span of the unborn pups by 25% relative to untreated controls. Thus, the expression of immune dysfunctions with a postmaturational onset can be significantly altered by the developmental effects of gonadal steroids during specific times of development. Iwase et al. (1995) showed that in the rat, maternal diabetes significantly lowered the birth weight and shortened the life span of male offspring, leading the authors to suggest that the reduced fetal growth induced by the diabetic intrauterine environment may accelerate an age-related degenerative process. The effects of maternal nutrition during pregnancy and lactation were assayed in mice by Ozanne and Hales (2004). The females were fed either a normal (20%) or low (8%) protein diet during pregnancy and/or lactation. The mice were then weaned onto standard laboratory chow

3.2 Distinguishing Disease and Environmental Changes from Age-related Changes

or onto a high calorie obesity-inducing diet for the rest of their lives, and the life spans of the six different groups were measured. As expected, the mice whose mothers were fed a low-protein diet during pregnancy showed about a 27% decrease in life span, regardless of the diet they followed as adults, and the obesity-inducing diet resulted in a 6–9% decrease in life span, depending on the particular group. However, the mice whose mothers were fed a normal diet during pregnancy but a low-protein diet during lactation showed a 6–13% increase in life span. This low-protein lactation diet even protected against the life-shortening effects of a subsequent obesity-inducing diet. It is likely that this nutrition effect operates by altering the activity of the offspring’s insulinlike signaling system, a conserved cell-signaling process that regulates the cell’s economy (see chapter 7). Most striking was that these nutritional differences during development and adulthood resulted in a 57% difference between the shortest and longest lived groups. It is thus possible to have striking congenital or familial effects on longevity that are not primarily genetic but rather arise from changes in maternal nutrition. These results also indicate the large extent to which gene–environment interactions can modulate a phenotype as plastic as that of longevity. If we assume that genes play an important role in longevity determination, an assumption critically discussed in chapters 7–9, then these experimental results mean that maternal nutrition presumably induces changes in gene expression in the offspring. This result represents a possible confounding effect in studies of the genetic determinants of longevity. A similar phenomenon is reviewed in chapter 13 involving the effects of prenatal stress or postnatal handling on the ability of rats to deal with the cortisol stress reaction. These results should also persuade us not to let superficial reviews of the data rush us too quickly into genetic determinism. In humans, the particulars of fetal development also seem to permanently program adult morphology, physiology, and life expectancy. It is well known that maternal malnutrition may adversely affect the developing fetus. What is new is the observation that certain cardiovascular and

71

metabolic disorders may arise from such malnutrition; the specific type of disease observed in the adult depends on the trimester during which the fetus was undernourished (figure 3.14) and/or the subsequent growth rate in early infancy (Robinson and Barker 2002). This situation presumably arises out of the interaction of nutritional factors with the tissue and time-specific patterns of gene expression involved in the development of the cardiovascular and other systems. A disparity between pre- and postnatal nutritional environments can also lead to the later development of certain adult-onset diseases (Gluckman and Hanson 2004). As adult life expectancy can be significantly affected by disease, this observation suggests that environmentally modulated developmental programming plays an important role in the plasticity of longevity in humans. The effects of early growth are not limited to the production of overt disease. For example, growth retardation in late gestation and low weight gain in infancy are believed to give rise to a reduced number of primordial follicles in the ovary, which leads in turn to earlier menopause and a possibly shorter life span (Cresswell et al. 1997). In fact, some familial traits of disease and longevity may be explained in part by such a mechanism. Maternal undernutrition in the second trimester may result in diabetes in the offspring, and these diabetic daughters may then give rise to shorter-lived grandchildren. If identical twins are involved in any of these genealogies, their high concordance might well be due to both a common genotype and a common intrauterine environment. However, current deterministic prejudices might cause many to interpret the data as indicating the effects of only the common genotype, and thereby miss the point. We need a better understanding of the molecular changes that underlie these fetal adaptations and their persistence throughout later life.

3.2.5 Postmaturational Changes That Accelerate or Retard Aging Most of the interventions effective in modulating the rate of aging are postmaturational; that is, they are generally applied to the individual

72 Chapter 3 Measuring Age-related Changes in Individuals

Trimester of pregancy in which fetus was undernourished

Effect on fetus

First

Second

Third

Down-regulation of growth

Disturbed fetoplacental relationship

Brain growth sustained at expense of trunk

Insulin resistance/deficiency

Growth hormone resistance/deficiency

Birth weight

Reduced

Reduced

Normal

Body proportions

Proportionately small

Thin

Short

Weight at 1 year

Reduced

Normal

Reduced

Adult life

Blood pressure

Blood pressure Non-insulindependent diabetes

Blood pressure LDL cholesterol Fibrinogen

Hemorrhagic stroke

Coronary heart disease

Coronary heart disease Thrombotic stroke

Death

Figure 3.14 A schematic representation of the effects of fetal undernutrition, according to trimester of pregancy, on the probable health trajectory of the future adult. (After Barker 1995.)

adult sometime after the developmental period is complete. These interventions can have either positive or negative effects and include smoking, nutrition, exercise, weight control, hormones, and other types of physiological interventions. I discuss these in some detail in chapter 6, in addition to pointing out some apparently ineffective interventions. For now, however, note that there are, even in the adult organism, at least a few means of effectively and significantly modulating the individual’s life span. It appears that aging processes can be modified even while they are taking place, an observation that poses some constraints on the types of mechanisms potentially involved. At a minimum, the existence of postmaturational

modification of longevity suggests that these mechanisms cannot be predetermined by the end of the developmental period but must include some labile component(s) well into adulthood. At a maximum, these interventions argue against the existence of an aging program.

3.3 Individual Rates of Aging and the Use of Biomarkers One of the mandates of the National Institute on Aging is to explore and develop approaches for extending the vigorous and productive years of

3.3 Individual Rates of Aging and the Use of Biomarkers

life. Simply measuring the life span will not yield sufficient information about the efficacy of a particular intervention. Clinicians and researchers alike want to know whether a particular intervention has successfully affected the physiological rate of aging of the system under investigation. This knowledge has practical importance because postponing the onset of clinical dysfunction in a person’s weakest physiological system may result in a substantial increase in both the quality and quantity of life for that individual. Continued progress in increasing the mean life span depends on this strategy. In effect, biomarkers constitute a sort of index, analogous to those indices commonly used to describe other complex systems (e.g., the DowJones numbers, Apgar scores, etc.). And as in those other cases, we are willing to trade off some of the information about the system to simplify and summarize it in a useful form (R. A. Miller 2001a,b). Consequently, the need has arisen to construct a panel of biological markers of aging (“biomarkers”) that can be used to test such segmental interventions both in humans and in laboratory animals. The concept of biomarkers rests on the assumption that the passage of time is only indirectly related to age; an assumption consistent with our definition of aging as being a timeindependent process (chapter 1). If, in a biological sense, the life span of a mouse and the life span of a human are equivalent, then the passage of time is a poor measure of age. What we need is some way to determine the rate of aging in an individual mouse or human, or, to put it in other words, some indicator of the rate at which the individual is transiting from a state of high somatic maintenance and function to a state of lowered somatic maintenance and decreased function. Such indicators are defined as biomarkers, and they are measures that could be obtained during a small portion of the life span and that would accurately predict longevity, either alone or in combination with other variables. In effect, biomarkers of aging are constructs with which scientists are trying to forge a connection between the population-level phenomenon of increased mortality and the individual-level phenomenon of age-related changes in various physiological parameters. The convergence of these two con-

73

structs would likely confer predictive power on the individual biomarkers. The tests composing such a panel of biomarkers ideally should have all of the following characteristics (Baker and Sprott 1988; Ingram et al. 2001; Reff and Schneider 1982): 1. The rate of change with time in a biomarker should reflect the rate of aging. 2. The biomarker should be monitoring a basic and important process. 3. The tests should be nonlethal and preferably noninvasive and cause minimal trauma. 4. The tests should be highly reproducible and should reflect physiological age. Among other things, this implies that the cross-sectional and longitudinal plots should agree with one another (see figures 3.6 and 3.7). 5. The function examined should display significant alterations during a relatively short time period. 6. The functions being measured should be crucial to the effective maintenance of health. 7. The biomarker should have a high cross-species correlation, and the rate of age-related change in the biomarker assayed in multiple species should be proportional to the difference in life span. 8. The biomarkers used should be able to function either as a prospective predictor of life span or as a retrospective marker of aging. Others have pointed out that biomarkers should also meet the following two requirements: 9. Pragmatic: They should be simple and inexpensive to use. 10. Methodological: They should be insensitive to the effects of prior measurements and robust over a wide range of laboratory and experimental conditions; above all, they must measure aging validly and reliably. A concise definition of a biomarker that includes the above points was offered by R. A. Miller (2001a, p. 2):

74 Chapter 3 Measuring Age-related Changes in Individuals To serve as a biomarker, a trait would need to meet three criteria: (i) it should predict the outcome of a wide range of age-sensitive test in multiple physiological domains, in an agecoherent way, and do so better than chronological age; (ii) it should predict remaining longevity at an age when 90% of the population is still alive; and (iii) its measurement should not alter either life expectancy or the outcome of subsequent tests of other agesensitive traits. Given the complexity of our bodies and of the aging process, different panels of biomarkers may be needed so that the prediction of longevity may be based on the rate-limiting variables. Finally, different types of biomarkers may be have a general predictivity, while others may have segmental effects. By segmental, I mean treatments that (1) retard the aging process (and physiological deterioration) in a specific system without significantly affecting the overall survival characteristics, and/ or (2) have this effect only during temporally restricted portions of the life cycle. By retarding the aging process, I mean that the intervention affects morbidity but not mortality (e.g., exercise). A temporally restricted biomarker may imply that the effects of some intervention on morbidity and/or mortality can be modulated only during a certain portion of the life span. The existence of segmental biomarkers suggests that the presumed multiple processes of aging may progress independently of one another (e.g., there may not be a general underlying aging process). For a nonsegmental biomarker to adequately predict either the degree to which one has already aged or else to predict remaining longevity assumes that there is a general covariance of the biomarker with all the various physiological systems of the body (e.g., there might be an underlying common process of aging. The answer to this basic question is not completely clear, but I believe much of the data presented in the rest of this chapter is more consistent with the existence of a common cellbased aging process than with the existence of independent multiple mechanisms of aging. I will return to this debate in later chapters.

All in all, the potential benefits of using biomarkers are matched by the difficulty of constructing them. It is fair to ask what progress has been made toward achieving this goal.

3.3.1 Results from Earlier Studies Costa and McCrae (1980) critically analyzed four earlier studies that attempted to devise a functional age scale for humans. They concluded that none of the studies to that date had yielded any statistically promising results, in part because of the heterogeneity of the variables used, in part because of statistical and methodological problems in measuring and comparing the results, and in part because of problems in concept and definition. Some of the less ambiguous and more widely used variables are listed in table 3.2. These variables are representative of the commonly used biomarkers in humans. Note that they are consistent with requirements 2–7 of the above list of desired properties, as well as being pragmatic, and methologically robust. Even so, note the wide range of correlation coefficients for the same variable across multiple studies. This finding suggests the existence of one or more sources of uncontrollable variation in the test and/or the sample population. In other words, not all variation in a single measure can be ascribed to differences in biological age, because, as we have seen from our discussion on blood pressure, such differences may exist for a variety of reasons not related to aging. This observation should make us pause before placing too much reliance on any single variable. How much correlation with age is desirable? If a particular variable showed a perfect correlation with chronological age, all we would have would be a perfect, and useless, alternative expression of chronological age. The problem here is that, if the intent of the effort is to find an alternative measure to chronological age, any approach that attempts to maximize the correlation of a particular variable to chronological age is logically flawed: A perfect model would merely be predicting the subject’s chronological age.

3.3 Individual Rates of Aging and the Use of Biomarkers

75

Table 3.2 Some of the Physiological Variables Used in Studies on Aging Variable Systolic blood pressure Hearing loss Lung capacity Reaction time Grip strength Diastolic blood pressure Height Visual acuity Forced expiratory volume (1 sec) Accommodation of eye Tapping Weight

No. of studies in which used 9 8 6 5 5 4 4 4 4 3 3 3

Correlation with chronological age 0.16 0.42 –0.77 0.26 –0.52 0.10 –0.68 –0.57 –0.70 0.88 –0.44 0.01

to to to to to to to to to to to to

0.69 0.66 –0.40 0.52 –0.21 0.51 –0.09 –0.42 –0.38 0.57 –0.18 0.56

Source: from Shock (1981).

What we want to do is get rid of age and time, for the reasons put forth in chapter 1. We need to correlate the rate of change of the biomarker with the rate of loss of function of the system being measured, or alternatively with the probability of surviving until some designated age. Can this be done?

3.3.2 Potential Panels of Human Biomarkers 3.3.2.1 Examples of Segmental and Nonsegmental Individual Biomarkers

The Baltimore Longitudinal Study on Aging (BLSA) is a prospective study of human aging initiated in 1958 that has generated much of our knowledge about biomarkers. One finding is that grip strength is a predictor of premature mortality in men who are older than 60 years of age (Metter et al. 2002). However, the data were not predictive for younger men (<60 years). Grip strength is thus an age-stratified segmental biomarker. As shown in figure 3.15, grip strength clearly differentiates the survival prospects of the 25% of men with the greatest grip strength from the 25% of men with the weakest grip strength; it presumably does not reliably separate out the two intermediate groups. Although encouraging, the data could not be used in their present form

to reliably make predictions on individual men, especially those who are not at one extreme or the other of the normal curve of grip strength. Nonetheless, for grip strength to have any statistical relationship with mortality suggests that muscle strength must represent the summed effects of changes in many other bodily systems such as changes in endocrine and/or growth factor secretion, and with the subsequent effects that cascade out of such initial changes. It thus suggests that even this segmental biomarker covaries with aging changes in other important nonmuscle systems, indicating the possible presence of some underlying systemic aging changes. Another long-term (since 1948) longitudinal study of community residents in a Massachusetts town is the Framingham study (table 8.7). This study has yielded some very interesting information regarding a prospective predictor of life span that may serve as an overall indicator of physiological age. This factor, forced vital capacity (FVC), is defined as the maximum amount of air that can be exhaled in a given amount of time (see also chapter 5). The decline in FVC with age is shown in figure 3.16. It yields gender-specific data, primarily because lung capacity is closely associated with height; however, both sexes show approximately the same trends and rate of loss. It is known from other work that smokers show a similar decline that exhibits a greater divergence from nonsmokers with time. The measured level

76 Chapter 3 Measuring Age-related Changes in Individuals

1.0 Men less than 60 years of age

Cumulative Survival

0.8

0.6 Men greater than 60 years of age

0.4

Lowest quartile Second quartile Third quartile Fourth quartile

0.2

0 0

10

20

30

40

Time (years) Figure 3.15 Kaplan-Meier survival plots in 1071 men followed over a 25-year period and sorted at age 60 on the basis of their grip strength (classed in quartiles of <83 kg summed grip, 83–96 kg, 96–108 kg, and >108 kg). Survival is significantly different in older but not in younger men. Note that the major mortality difference in the older men is the enhanced survival of the highest quartile men in their late 70s and early 80s, compared to the generally depressed survival of the lowest quartile men from age 60 on. Also note that there is no alteration in maximum life span. (After Metter et al. 2002.)

Mean forced vital capacity (deciliters)

45

40

Men Women

Cohort

Men Women

Cross-sectional

35

30

Men Men

25

Women 20 Women 15 32

37

42

47

57 52 Age (years)

62

67

72

77

Figure 3.16 The average age trends in forced vital capacity for cross-sectional and longitudinal data obtained from the Framingham study. Note that the cross-sectional and longitudinal results agree with one another. (After Kannel and Hubert 1982.)

3.3 Individual Rates of Aging and the Use of Biomarkers

of the FVC is a statistically significant predictor of premature mortality (see table 3.2 and figure 3.20) and appears to be measuring a function that is more general than impaired pulmonary function. In the words of Kannel and Hubert (1982, pp. 157–159): “In any event, an FVC determination would appear to be an efficient way to identify asymptomatic persons destined for a premature death. . . . FVC is one of the strongest predictors of mortality, second only to age itself. . . . It appears to be a measure of vigor, general musculoskeletal functional capacity and overall health; truly a measure of living capacity.” 3.3.2.2 Age-specific Groups of Biomarkers

Brant et al. (1994) showed that different biomarkers may serve as different age-specific indicators of relative risk. Working with longitudinal data obtained from BLSA, these investigators statistically analyzed several physiological factors collected over a 40-year period and correlated their values with the age at death of the subjects. They found that the predictive value of these biomarkers of mortality was highly age specific (table 3.3). High levels of cholesterol, for example, were predictive of a high risk of mortality at age 40, but not at ages 60 or 80. Body build (body mass index) predicted mortality in differ-

Table 3.3 Biomarkers Significantly Associated with Mortality at Specific Ages Age (years)

Biomarkers

40

White blood cell count Cardiac diagnosis Diastolic blood pressure Total serum cholesterol Serum triglycerides White blood cell count Cardiac diagnosis Systolic blood pressure Forced expiratory volume Visual acuity High body mass index White blood cell count Low body mass index

60

80

Source: adapted from Brant et al. (1994).

77

ent ways; obesity was a risk factor for mortality at age 60, but leanness was a risk factor for mortality at age 80. Independently, a study of Swedish twins led to the conclusion that the heritability of various cardiovascular risk factors had agespecific peaks and were much higher for people under age 65 than for those over age 65 (Row and Kahn 1997). The fact that two different types of studies led to the same conclusion strengthens the case that some biomarkers are age specific. This unexpected situation suggests the need for caution when using age-specific physiological values to construct a numerical measure of biological age. Does this mean that the concept of biomarkers is not useful? No, in fact, it enhances their usefulness by identifying the specific populations that can best benefit from them. Brant et al. (1994) have suggested that the various relative risk values for the traits listed in table 3.3 could be combined to fit the particular profile of any individual and the relative risk of mortality for that individual computed. The example given in figure 3.17 shows that individuals with the higher risk factors have a higher probability of dying within a specified time period than do those with the lower risk factors. Recall that in chapter 2 we defined aging as the increased probability of dying within a certain time interval and quantitated the definition with the Gompertz curve and the mortality rate doubling time (MRDT). By calibrating their data against mortality, not illness, Brant et al. (1994), as well as the authors of the study shown in figure 3.15, have constructed and validated a set of biological indices of aging that can predict the probability of dying within a specified time interval. The indices therefore are measuring different rates of aging in these two populations. Do not overestimate the ability of these indices; they are predicting the differential mortality of two populations. It is not known if these biomarkers would be more or less predictive when tested on populations comprising women or different ethnic or socioeconomic groups. Nor is it known how accurately they might predict a particular individual’s death date in any of these groups. Yet without that information, they are not yet ready for clinical deployment. However, this exercise does prove the

78 Chapter 3 Measuring Age-related Changes in Individuals 100

2.1, a ratio close enough to the predicted difference of 3 based on the ratio of the maximum life spans of the two species (120/40 years). Thus, testing prospective physiological and/or molecular biomarkers for a defined period of time on both humans and shorter lived species to see if they are proportional should allow the identification of presumptive human biomarkers without the necessity of waiting for 75 years for the answer. This innovation is now widespread and is helping identify molecular and cellular biomarkers correlated with longevity. These biomarkers will be necessary for testing human anti-aging interventions, as discussed in chapter 15.

Percent surviving

90

80

70

Low risk High risk

60

0 40

50

60

70

Age (years)

Figure 3.17 Estimated survival probabilities for males measured at age 40 with low or high levels of coronary risk factors, as listed: low (diastolic pressure 75 mm Hg, serum cholesterol 200 mg/dl, serum triglycerides 110 mg/dl) and high (diastolic pressure 95 mm Hg, serum cholesterol 250 mg/dl, serum triglycerides 140 mg/dl). Data are from the Baltimore Longitudinal Study on Aging. The vertical lines show the ages at which both the high- and low-risk groups have an estimated 80% chance of surviving the next year. The difference in the age at which the two groups reach this mortality level is about 5 years; in other words, there is about a 10% difference in survival rates at age 70 between the two groups. (After Brant et al. 1994.)

concept of biomarkers in principle and illustrates the ongoing covergence of the individual and population-level metrics. Many variables may show a strong correlation with age, and they may do so for various reasons that have nothing to do with senescence. How can we empirically determine that the variables are actually measuring the rate of aging and not some other factor? As pointed out in list of biomarker characteristics discussed earlier, the rate of agerelated change in the biomarker assayed should be proportional to the difference in life span in multiple species. Ingram et al. (1999) compared the age-related loss of dihydroepiandrosteronesulfate (DHEA-S) in humans and monkeys. Their data showed that the rate of change of DHEA-S in rhesus monkeys, relative to humans, was about

3.3.2.3 Panels of Physiological Biomarkers

As part of BLSA, Borkan and Norris (1980) assayed the physiological age of 1086 males using the 24 different tests listed in table 3.4. The use of this panel of biomarkers gave rise to some important insights. Burkan and Norris transformed the physiological values into an age-specific score, determined whether that score was skewed in the direction of being younger or older than the mean score for that variable in men of the same age, and then plotted the transformed results of this biomarker panel for each individual, as shown in figure 3.18. This analysis demonstrates clearly that an individual may be older in some parameters than in others, relative to other individuals of the same age, and reinforces the earlier discussion of the existence of substantial individual heterogeneity in aging (e.g., figure 3.7). In addition, however, Borkan and Norris made a visual estimate of each individual’s chronological age, which led to a interesting conclusion. Their analysis of the data revealed that individuals who looked older were indeed biologically older (although note that the data in this example are taken from the two extreme ends of the population, as explained in figure 3.19). A more generalizable aspect of their studies is that those subjects who had died since the start of the study were biologically older at the time the parameters were measured than those of the same age who had survived (figure 3.20). This is important because it means that the biomarker panel has been tested against the gerontological

3.3 Individual Rates of Aging and the Use of Biomarkers

Table 3.4 Mean Correlation with Age for Selected Biomarkers in 1086 Males Studied in the BLSA Biomarker Forced expiratory volume (1 second) Vital capacity Maximum breathing capacity Systolic blood pressure Diastolic blood pressure Hemoglobin levels Serum albumin levels Serum globulin levels Creatinine clearance Plasma glucose levels Auditory threshold (4000 cycles/sec) Visual acuity Visual depth perception Basal metabolic rate Cortical bone percent Creatinine excretion Hand grip strength Maximum work rate Benton visual memory test (errors) Tapping time (medium targets) Tapping time (close targets) Reaction time (simple) Reaction time (choice) Foot reaction time

Correlation coefficient (r) –0.698* –0.606* –0.547* 0.538* 0.368* –0.223* –0.356* 0.092* –0.602† 0.279 0.549 –0.306† –0.232* –0.337* –0.435* –0.538* –0.501* –0.511* 0.502* 0.468* 0.366* 0.287* 0.220* 0.222*

Source: from Borkan and Norris (1980). *p †p

< .01. < .05.

“gold standard” of life span. Note that the deceased men have “older” values of forced expiratory volume and vital capacity, which is consistent with the data presented in figure 3.16, as well as “older” values of neural reaction times. However, the decedents showed “younger” values for grip strength, a finding seemingly at odds with the data of figure 3.15 until one remembers that grip strength is an age-stratified segmental biomarker not likely to yield informative results when summed across decedents <60 years of age. The discrepancy may be an artifact of the statistical procedures used. The most important conclusion to be gathered from this study is that human aging is an individual and mosaic affair, a finding consistent with the data of figure 3.7. In different individ-

79

uals, different biomarkers from the test panel are affected, sometimes in different directions (i.e., younger or older). This likely explains why we should not expect a perfect correlation between any particular biomarker and the rate of loss of function (or alternatively life span). Only a small subset of these variables is statistically significant, yet there are obvious differences in the overall pattern of the biomarker panel in survivors compared to nonsurvivors. The conclusion drawn from the biomarker panel is consistent with the visual estimation of the person’s age done by trained professionals, at least for men at the upper and lower extremes of the age distribution (figure 3.19). These findings are consistent with the idea that there is likely to be some underlying general aging process that secondarily alters these biomarkers in an individual and mosaic manner. The doctors who estimated age in this study judged that larger individuals looked older than their actual age, raising the problem of the subjective interpretation of data, as well as a possible inadvertent confounding of survival characteristics with body size. In fact, chest circumference is probably the best single predictor of FVC; thus one might reasonably expect larger individuals to be longer lived. The issue is not yet clearly understood. 3.3.2.4 Other Mortality Tested Human Biomarkers

The most robust intervention known for significantly slowing aging and extending the healthy portion of the life span is the dietary manipulation known as caloric restriction. Caloric restriction involves the maintenance of good nutrition coupled with a significant reduction in caloric intake. It is most definitely not akin to starvation. I discuss this in some detail in chapter 6 and subsequent chapters. Caloric restriction induces a host of changes in multiple, supposedly independent, physiological systems, and these changes are such as to maintain the body’s maintenance of health and function at a high level for a significantly longer time than is the case with control animals fed a normal diet. The net result of these changes is delayed onset of senescence and highly

80 Chapter 3 Measuring Age-related Changes in Individuals

0.6

Biological age younger older

0.4 0.2 0 –0.2 –0.4

FE

V VC M BP BC -s BP yst. -di as t. Alb Hb u Gl min ob Cr ulin Pl. eat. c g Au luco l. d. s e t Vis hresh .a . Vis cuity .d ep th Co BM r ti R Cr cal % ea t. Gr ex. Ma ip st Vis x. w r. o . Ta mem rk pp o i Ta ng- ry pp ing med . RT clos -ch e RT oic -si e m Fo ple ot RT

–0.6

Figure 3.18 A biological age profile of a single individual, using the parameters given in table 3.4. This profile demonstrates that an individual may be biologically older on some parameters than on others. Biological age is expressed as the two score transformation of the residuals obtained from a regression of the data for each biomarker on age. (After Borkan and Norris 1980.)

0.6

Mean of older-looking subjects Mean of younger-looking subjects

Biological age younger older

0.4 0.2 0 Ð0.2 Ð0.4

FE V VC M BP BC -s BP yst. -di as t. Alb Hb u Gl min ob Cr ulin Pl. eat. c g Au luco l. d. thr se Vis es . a h. Vis cuity .d ep th Co BM r tic R Cr al % ea t. Gr ex. Ma ip st Vis x. w r. o . Ta mem rk pp o Ta ing ry pp -m ing ed . RT clos -ch e RT oic -si e m Fo ple ot RT

Ð0.6

Figure 3.19 Biological age profiles based on subjectively estimated age, using the parameters given in table 3.4. The dotted line represents mean biological age scores of subpopulations that appeared the most old for their age (the top 15% of the distribution). The solid line represents mean biological age scores of subpopulations that appeared the most young for their age (the bottom 15% of the distribution). The starred parameters represent significantly different mean scores by t test (p < .05). Biological age expressed as in figure 3.20. (After Borkan and Norris 1980.)

3.3 Individual Rates of Aging and the Use of Biomarkers

0.6

81

Mean of deceased Mean of survivors

Biological age younger older

0.4 0.2 0 –0.2 –0.4

FE

V VC M BP BC -s BP yst. -di as t. Alb Hb u Gl min ob Cr ulin Pl. eat. c g Au luco l. d. thr se Vis es . a h. Vis cuit .d y ep th Co BM r tic R Cr al % ea t. Gr ex. Ma ip st Vis x. w r. o . Ta mem rk p Ta ping ory pp -m ing ed . RT clos -ch e RT oic -si e m Fo ple ot RT

–0.6

Figure 3.20 A comparison of the biological age profiles of deceased study participants with those of surviving study participants, using the parameters given in table 3.4. The dotted line represents mean biological age scores of 166 men who had died since they were last measured. The solid line represents mean scores for all other (922) participants, who were still alive. The starred parameters represent significantly different mean scores by t test (p < .05). Biological age expressed as in figure 3.20. (After Borkan and Norris 1980.)

significant extension of longevity. Caloric restriction is known prolong life span in all invertebrate and vertebrate species tested to date. Although caloric restriction experiments are now being done on rhesus monkeys, because of their long life (~40-year maximum life span; table 4.8), the experiments are still in progress. Still, sufficient data have been collected to show that biomarker changes known to be induced by caloric restriction in rodents also occur in rhesus monkeys. Specifically, caloric restriction induces decreased body temperature, decreased plasma insulin level, and increased DHEA S blood level in male monkeys (figure 3.21), and these changes are associated with a mortality rate only 60% that of the normally fed controls (Roth et al. 2002; see also chapter 6). As humans are also primates, these data suggested that the same effect might occur in humans. The BLSA contained physiological and mortality data for hundreds of men who died at various ages. The obvious question was whether the longer lived men in the BLSA study showed changes in their biomarkers comparable to those noted in rhesus males on a calorie-restricted diet.

The data of figure 3.21 show this is indeed the case: Longer lived human males showed significant correlations with decreased body temperature, decreased plasma insulin levels, and increased DHEA S blood levels. This occurred even though there was no evidence that BLSA men were voluntarily restricting their caloric intake. Whatever the factors were that made these men live longer, they certainly seem to have altered their biomarkers in the same direction as noted in the calorierestricted rhesus monkeys. Thus, it seems that these observed biomarker changes are most probably associated with greater longevity and that they have some predictive value. Since neither the men nor the monkeys are known to have been subjected to any intervention designed to specifically modify these three different physiological traits, it is reasonable to deduce that there is some underlying general aging process, slowed by caloric restriction in the monkeys and by unknown factors in the men, that secondarily alters these (and other) biomarkers. A separate study showed that caloric restriction had a powerful protective effect against atherosclerosis in men (Fontana et al. 2004), raising

Baltimore Longitudinal Study of Aging Male Humans 1.0

D Lower

0.8

Higher 0

5

10

15

20

0.6

0

10

15

20

0.6

Lower 0

5

10

15

20

Survival time (years)

37

20

B

15 10 5 0 150

Higher

F

38 37.5

25

1.0 0.8

38.5

Insulin Level

Higher 5

A

39

25

Lower

0.8

NIA Primate Aging Study Male Rhesus Monkeys

39.5

25

1.0

E

40

% of initial DHEAS value

Cumulative Survival

0.6

Temperature (C )

82 Chapter 3 Measuring Age-related Changes in Individuals

C

100

50 0

25 Control

CR

Figure 3.21 Caloric restriction (CR)-treated monkeys and longer-lived men have similar values in important physiological variables. Values for male rhesus monkeys are means ± standard deviations for 20–30 animals in each group, after 3–5 years of 30% CR. Numbers of men in the temperature, insulin, and DHEAS studies of BLSA were 324, 199, and 192, respectively. All subjects were in good health, and their age ranged from 19 to 95 years. All analyses are significant to at least p < .05 by t test or proportional hazard model, corrected for initial subject age. (After Roth et al. 2002.)

the possibility that the long-lived men in the BLSA study may have been moderate eaters. The fact that the rate of aging can be significantly slowed by applying an environmental or nutritional intervention such as caloric restriction shows that significant changes in the aging process can be brought about by lifestyle or other nongenetic changes. We will not have to get our genes changed in order to change our gene expression patterns. The data presented in chapter 7 show that our genes do not function in a vacuum but rather are kept informed by the neuroendocrine system about the caloric quality of our environment, and they alter their expression patterns accordingly. This interpretation has an obvious relationship to future anti-aging interventions. I return to this topic elsewhere in the text, particularly in chapter 15. As a result of these and other data, the National Institute on Aging has funded a 7-year study

on the possible benefits of caloric restriction in humans. The study has only recently begun, but the interested reader may follow its progress at http://calerie.pbrc.edu. In addition, a list of about two dozen ordinary clinical tests that are thought to constitute a useful panel of biomarkers relevant to the effects of caloric restriction on health status may be obtained at http:// calorierestriction.org/book/print/224. 3.3.2.5 Gender-specific Biomarkers

A longitudinal study of human females has shown that age at natural menopause is inversely associated with longevity (Snowden et al. 1989). Women who reported an early onset of natural menopause (less than 44 years) had a significantly elevated mortality relative to those who experienced the onset of menopause at ages 50– 54 years. This observation suggests a correlation

3.3 Individual Rates of Aging and the Use of Biomarkers

3.3.2.6 Is There a Single Indicator

of Biological Age? If there is some sort of general aging process, then perhaps it would be useful to measure it and construct some numerical measure of biological age to serve as a functional analogue to chronological age. The literature records many attempts to construct a single numerical indicator of biological age (see Balin 1994a). Although many of these studies present interesting and useful physiological data, none of them can escape one criticism or another, often of a statistical nature. Does this mean that the attempt to quantify human aging is intrinsically impossible? Not at all. One can apply mathematical principles other than regression and correlation to chronological age in an effort to identify and quantify the aging process. Nakamura (1994) used the technique of principal component analysis, which is designed to discern the underlying structure of interrelationships among the test scores measured in a complex system while retaining as much of the variation present in the data set as possible. His results were most interesting. Rather than constructing a single index of biological age, Nakamura identified 13 factors organized into 3 classes. The first factor (F1), a general aging factor, accounted for 13.5% of the total variance in test scores. The second set of 11 system-specific aging factors accounted for 57.5% of the variance. The third set, termed uniqueness (the individual variance inherent in every variable but not necessarily related to any aging process), accounted for 29.0% of the variance. These results suggest that, while there may be a fundamental and unitary aging process, the system-specific expression of

this process is what is most easily detected and what accounts for most of the variance. Building on this result, Nakamura (1994) incorporated the values of the 11 physiological variables into a complex equation, constructed a biological score of the age-related changes in these physiological functions, and manipulated this score to yield a numerical biological age. When he plotted this biological age for 462 healthy men against their chronological age, he found that the former closely tracked the latter (figure 3.22). When he performed the same procedure on diabetic or hypertensive adults, he found that both of these ill groups had physiological values that translated into biological ages about 4 years older than their chronological age (figure 3.23). This significant discrepancy between biological and chronological ages was interpreted as due to a more rapid rate of aging in these ill individuals. An analogous strategy was adopted by Hochschild (1989, 1994), who developed a suite of 12 noninvasive and automatically administered physiological tests, which he administered to 2462 office workers in the United States. Hochschild used data transformation techniques to avoid the confounding of biological and chronological age already discussed, presented his data as a deviation

90

CA = BA

80 Biological age (years)

between ovarian aging and the aging of other tissues. Other data presented in chapter 8 are consistent with this point of view. The mechanisms involved are not clear, but age at menopause might serve as a prospective biomarker of aging for females. One implication of this observation is that there is some sort of general or systemic aging process that regulates all body systems. I deal with this general concept in some detail in chapters 7 and 14.

83

70 60 50 40 30 Mean biological age

20 20

30

40 50 60 70 Chronological age (years)

80

90

Figure 3.22 The relationship between biological age (BA) and chronological age (CA) in 462 healthy men. (Data from Nakamura et al. 1988.)

84 Chapter 3 Measuring Age-related Changes in Individuals 90

Biological age (years)

80 70 60 50 40 30

Hypertension subjects Mean BA = 57.25 4.17 yr Mean CA = 53.08

20

Diabetic subjects Mean BA = 55.57 4.57 yr Mean CA = 51.00 20

30

40 50 60 70 Chronological age (years)

80

90

Figure 3.23 The relationship between biological age (BA) and chronological age (CA) in 62 hypertensive and 65 diabetic men. Note that in both groups of subjects, biological age is more than 4 years greater than chronological age. (Data from Nakamura et al. 1988.)

about the mean in a manner analogous to that shown in figures 3.18–3.20, and used these results to construct a unitary numerical index of biological age. Hochschild then proved the validity of his tests by demonstrating that they yielded a higher biological age for subjects who were known to have risk factors, such as smoking, for chronic disease. Both Nakamura (1994) and Hochschild (1994) validated their protocols against people with chronic disease or with risk factors for chronic disease. Both protocols may be accurate predictors of disease, but that fact does not validate them as predictors of aging. If either procedure is to fulfill the eighth criterion on the list of desired biomarker characteristics (i.e., that the biomarker should function as a prospective measure of life expectancy or as a retrospective measure of aging), then it must be verified by longitudinal studies that follow each individual until the time of death, as was done in the several BLSA studies discussed above. If the construct is valid, individuals with a biological age score higher than their chronological age generally should die earlier than individuals with a biological age score lower than their chronological

age. In the absence of such information, we do not know exactly what we are measuring, and thus we cannot reliably interpret what the observed relationship between aging and disease means. Nakamura and Miyao (2003) have recently repeated and reevaluated their study, using only healthy men. They again identified nine physiological variables that changed with age in this group (forced vital capacity, forced expiratory volune in 1 second [FEV], systolic blood pressure, red blood cell count, hemoglobin, hematocrit, albumin, albumin/globulin ratio, and blood urea nitrogen). A similar study done with healthy women identified only five age-related physiological variables (FEV, systolic blood pressure, hemoglobin, glucose and albumin/globulin ratio; Ueno et al. 2003). No reason was identified for the gender difference. Using principal components analysis, these investigators again concluded that aging consists of a time-dependent complex integration of a primary, or general, aging process and least three secondary, or system-specific, aging processes. Thus they independently verified their main conclusion from their earlier study (Nakamura 1994). Except for the inclusion of time, this is not dissimilar from the conclusion presented in chapter 14 that senescence can be viewed as alterations in a hierarchial gene interaction network. Ueno et al. (2003) determined biological age scores for the women and concluded that biological age changes relatively slowly in women less than 65 years of age, but then shows a rapid increase after that age. It would seem that Ueno et al.’s biological age construct is measuring the rate of senescence among these chosen variables. All of the above studies are consistent with the view that a general aging process may express itself in specific ways in different systems (see chapter 14). This conclusion is consistent with data obtained from laboratory animals ranging from flies to mice, as discussed below. Nonetheless, compacting the data of figure 3.23 into a single number is probably an error. We have seen that some biomarkers are age or gender specific, and it is not clear how their segmental nature could be incorporated into a single number. In addition, we have seen that the ac-

3.3 Individual Rates of Aging and the Use of Biomarkers

tual changes in biomarker values seems to yield different individual mosaics of alterations, not all of which are statistically significant but many of which covary with mortality. At our present state of knowledge, it is not clear how different individual mosaic patterns can be subsumed within a single number. What would be most useful would be a set of human biomarkers that could be measured early in adult life (i.e., before age 30) and that would reliably predict the survival prospects of the individual. Those people with shorter-life predictors might then be taught and encouraged in the use of appropriate behavioral and/or physiological interventions that would act to buttress their weak points and help them to live a healthier life. The existence of such biomarkers would also serve as a target of intervention research. In this context, the work showing that body weight, hormones, and immune status in young mice serve as predictors of longevity should serve as a model for future human research (Harper et al. 2004). Indeed, a suite of multiple biomarkers of physical and mental functioning has been tested on a large population (~1000) of middle-aged and older adults and revealed the existence of statistically significant associations between certain of these biomarkers and the individual’s functional state (Seplaki et al. 2004). It is interesting that women had a more heterogeneous mix of physical, psychological, and cognitive impairments than did men. The longitudinal continuation of this study should allow for the future testing of the multiple biomarker approach. 3.3.2.7 Biomarkers of Mortality and Disease

All the biomarkers discussed up to this point have been designed to measure the rate of aging as evidenced by the probable longevity. They are not meant to predict the onset of disease or of death. However, I pointed out in chapter 1 that diseases are simply the systemic failures that highlight the weak points of the organism’s evolved design and allow us to identify and investigate them. It would be useful to be able to characterize the predisease state of the individual and to do this characterization at the molecular, physiological, and

85

biomarker levels. Such characterization would allow both a prediction of future probabilities as well as the development of an appropriate intervention. There are gene expression differences between a state of high systemic function or minimal aging and a disease state. What are the characteristics of the transition state between the two? Is it reversible? At the very least, having a clinically useful biomarker of incipient disease will signal that effective clinical interventions are necessary. Delaying the onset of disease will allow for an increase in an individual’s longevity (and certainly their “health span”), even though their rate of aging would not be affected. The data of figure 2.14 show that large societal changes can flow from this attention to biomarkers of disease. The following two examples illustrate such biomarkers. Remember that mortality and disease biomarkers likely measure a decline in function preceding a systemic failure, whereas biomarkers of aging likely measure the extended maintenance of function and longevity. The Frailty Syndrome. An example of an agerelated group of biomarkers are those involved in the “frailty phenotype,” often found in elderly adults (Fried et al. 2001). Frailty has been defined as a clinical syndrome in which three or more of the following criteria are present: unintentional weight loss, self-reported exhaustion, low grip strength, slow walking speed, and low physical activity. The syndrome likely represents a loss of the individual’s physiological reserve capacity and thus the presence of subclinical loss of function in various systems (i.e., the presence of incipient disease; see Newman et al. 2001). Frailty syndrome is not concordant with the presence of overt diseases or disabilities but is predictive of their future occurrence. This suggestion is buttressed by the theoretical work of Lipsitz (2002), which is based on the idea that normal physiological function requires the integration of complex networks of feedback signals, control systems, and other regulatory phenomenon. These dynamic interactions can be measured. Healthy individuals show highly irregular, complex interaction dynamics that degrade with age to yield more regular and less complex patterns.

86 Chapter 3 Measuring Age-related Changes in Individuals The lack of complexity is interpreted as indicative of degradation of the regulatory links between various systems, with the consequence that the individual displays a decreasing ability to mount a focused and integrated response of multiple organ systems when challenged by some stress. This interpretation suggests that aging may represent the breakdown of intercellular regulatory mechanisms, and I deal in more detail with this concept in chapter 14. The clinical presence of this frailty phenotype (i.e., presence of three of the five criteria) is independently predictive for an approximately doubled risk of morbidity and death, whereas the presence of an intermediate frailty phenotype (i.e., presence of one or two criteria) indicates a doubled risk of progression to the full phenotype, relative to age-matched nonfrail (i.e., absence of all five criteria) control adults. This syndrome, although most common in elderly adults, may be better understood as a physiological stage-specific set of biomarkers, rather than as a strictly agespecific biomarker panel, which is predictive of a terminal decline in reserve capacity leading to mortality. The onset of systemic body failure is only correlated with chronological age but is likely directly related to physiological status, a view that may be supported by the emergence of additional biological markers characterizing the frailty phenotype at a molecular level (Ferrucci et al. 2002). Behavioral Biomarkers. All the other biomarkers discussed in this chapter are molecular and/or physiological variables correlated with longevity. To the best of my knowledge, there is currently no formal list of human behavioral biomarkers as such. Yet there is a widespread knowledge and clinical appreciation of the role that various lifestyle behaviors can have in adversely affecting the health and functioning of older adults (Damush et al. 1999; Willett 2002). A person’s pattern of nutrition, physical activity, body mass index, smoking, level of social engagement, and psychological characteristics significantly affect the functioning of the individual, independent of the existence of any chronic medical conditions (Bronson et al. 1999; Drewnoski and Evans 2001; Seeman and Chen

2002). Given the alarming increases in the prevalence of chronic conditions such as diabetes, coronary heart disease, hypertension, and cancer, it is likely that benefits from our increased knowledge of aging mechanisms will be canceled out by the predicted surge in chronically ill, shortlived people. There is good reason to use behavioral biomarkers, or risk factors, both as predictors of these chronic disease states and as indicators that a change in specific behaviors is indicated. It is estimated that the incidence of the chronic conditions mentioned above can be potentially reduced by 70–90% by lifestyle changes (Willett 2002). Such changes not only prevent the negative effects of future disease but can also bring about beneficial states that have a positive effect on longevity. For example, habitual physical activity in both men and women results in better functioning than might be predicted based on chronological age (Nakamura et al. 1996, 1998). Comparable findings are available for the other variables (see figures 6.6–6.8). Childhood mental activity test (IQ) scores may be a modest predictor of adult longevity. A cohort of Scottish children who had taken an IQ test at age 11 were later traced and their survival at 76 years of age ascertained (Whalley and Deary 2001; Hart et al. 2005). For both sexes, there was a small but significant difference in the IQ scores of survivors (102.0 ± 14.2) versus nonsurvivors (97.7 ± 15.4). Note that the difference of 4.3 points falls within the standard deviation of either mean. The survival difference may have resulted from both individual and societal preferences which allowed higher scoring individuals a better chance of entering safer occupations, practice a healthier lifestyle, and otherwise engage in behaviors that do not shorten life. Animal studies strongly suggest that a behavioral change such as habitual physical activity prevents age-related alterations in gene expression in the mouse heart (Bronikowksi et al. 2003), and other data suggest that such behavior may prevent or delay the onset of the human gene expression pattern indicative of congestive heart failure (Tan et al. 2002). Thus, these behavioral biomarkers can be viewed not only as empirically derived risk factors but also as surrogate indica-

3.3 Individual Rates of Aging and the Use of Biomarkers

tors of basic biological states. In this sense they fulfill the biomarker definition given above, and they give us a means to characterize and quantitate the effects of environmental factors on longevity and senescence. At this time, behavioral biomarkers potentially indicate future systemic failure, which can allow individuals to deliberately choose to alter their behaviors and thus provide the physiological and genetic conditions necessary to minimize premature morbidity and mortality. On a longer time scale, it is likely that a significant extension of longevity by whatever means will be unlikely in individuals who have a lifestyle that promotes premature mortality. Triage being what it is, the presence of one or more of these behavioral biomarkers of mortality and disease (smoking, lack of exercise, body mass index>25, etc.) in an individual may be sufficient reason to prevent that person being prescribed some scientifically based anti-aging intervention in the future (and not a distant one either; see chapter 15). A Potential Panel of Primate Biomarkers. People are not mice. Interventions with the potential to alter the human life span must be tested experimentally before being considered for widespread use. This requirement means that they must be tested in nonhuman primates. Macaques, also known as rhesus monkeys, are the most commonly studied genus of nonhuman primates. Since they have a maximum life span of at least 40 years, it is not practical to use an extension of maximum life span as the only assay for the effectiveness of such interventions. Assaying the effectiveness of these interventions requires the development of a panel of primate biomarkers. It would not be surprising if such a primate panel were also useful in assaying human aging. A good start on sorting out the useful variables has been made in a cross-sectional analysis of biomarkers in the rhesus monkey. The goal of the experiment was to construct a battery of biomarkers for each age–sex class that would allow the assessment of the biological rate of aging in these monkeys in significantly less than 40 years. From an initial list of about 290 candi-

87

date variables, Short et al. (1987) empirically selected 72 variables for further study across young, middle-aged, and old animals. Of these, 20 were significantly influenced by age and were therefore selected for inclusion in an initial longitudinal study (Bowden et al. 1994). Only eight physiological and three behavioral variables, listed in figure 3.24, each showed a significant rate of change with time and were used to construct a biological index of aging (Short et al. 1994). In addition, the investigators determined the rate of change of certain antioxidant levels, found that they varied in the expected direction (see chapter 10), and included them in the model as protective mechanisms that might modulate the animal’s intrinsic rate of aging. The strategy of these investigators illustrates how one may move ahead and produce what appear to be useful biomarker panels even though the biomarkers cannot be definitively validated, in terms of prospectively predicting maximum life span, within the professional careers of the people involved. A similar time-saving strategy may have to be adopted for translating this biomarker panel from macaques to humans. Much of the ongoing work on caloric restriction in rhesus monkeys has allowed the characterization of those diet-induced physiological effects that may predict an increased life span (Mattison et al. 2003). Such effects include decreased body weight and fat mass, improved glucoregulatory function, decreased blood pressure, decreased blood lipids, decreased body temperature, and an attenuated decline in both DHEA-S and melatonin. If their predictive effects on longevity are borne out, then these parameters may provide the basis for future human biomarkers. Note that three of these factors have already been shown to be predictive of enhanced longevity in humans (figure 3.20); thus the use of calorically restricted primates represents a valid approach to detecting and testing presumptive human biomarkers. It would be most interesting if these physiological variables were shown to affect a molecular mechanism such as the level of intrinsic oxidative stress, known from the studies summarized in figure 3.24 to be intimately involved in the organism’s loss of function (see chapter 9; Golden et al. 2002).

88 Chapter 3 Measuring Age-related Changes in Individuals

Chronological age

Etc.

SEX

MCH

LYM

CRE

TP

IGA

CL

NA

FNG

Etc.

ACT MNT PRS

Rate of biological aging

Etc.

VITC

VITE

CROT

Etc.

Aging-rate modulator

Figure 3.24 A generalized model for biological aging in the pig-tailed macaque monkey, a nonhuman primate. Eight factors showed significant age-related within-subject change: fingernail growth (FNG), serum sodium (NA), serum chloride (CL), immunoglobulin A (IGA), serum total protein (TP), serum creatinine (CRE), lymphocytes in white blood cells (LYM), and mean corpuscular hemoglobin (MCH). These factors were used as biological measures of aging. In addition, three measures of sexual behavior, activity level (ACT), mounting of female by male (MNT), and presentation of female to male (PRS), were used to quantify sexual activity (SEX), which was then integrated into the model. Because more age-related factors may be found in the future, “etc.” allows for an open-ended model. High levels of various antioxidants—carotenoids (CROT), vitamin E (VITE), and vitamin C (VITC)—were found to be associated with low values of biological aging, and vice versa. Therefore, these factors are thought to act as modulators of the age-related changes in the indicated biomarkers. (After Short et al. 1994.)

A Potential Panel of Rodent Biomarkers. Recognizing the need to validate at least the shortlived rodent biomarker panels against longevity measures, the National Institute on Aging embarked on a large-scale, long-term, multiple longitudinal study of biomarker characterization and analysis using four mouse and three rat inbred strains (or their hybrids). Some of the animals were subjected to caloric restriction; much of the subsequent analysis focused on the effects of that intervention and is discussed in chapter 6. The

results of this large study have been published and were summarized by Sprott (1999). Maintaining normal functioning of the immune system is essential for continued survival and longevity of mammals. The T-cells (i.e., thymusderived white blood cells; see chapter 12) of an organism comprise several different subsets, each one distinguished by particular molecules on their cell surface. The relative levels of each of these subsets varies with age in both mice and humans. The question arose as to whether the relative levels

3.3 Individual Rates of Aging and the Use of Biomarkers

89

cohorts, as above. Four different genetic loci were statistically associated with these differences in body weight when young, indicating genetic control of this phenotype. In addition, there are suggestions of the involvement of different hormone levels also being involved. Combining the T-cell subset data at 8 months, the body weight data at 3 months, and the animal’s thyroxine (T4) levels at 3 months led to a tripartite model which is a more robust predictor of adult longevity than is any one of these early-life factors (Harper et al. 2004). Sorting the animals by these three factors led to the identification of groups that differ by 18% in mean life span and 16% in maximum life span. These effects are about half those obtained with caloric restriction (chapter 6) or genetic interventions (chapter 7), but were brought about without any experimental intervention. These life span differences likely represent the natural intrapopulation variation between geneotypes for longevity. The data suggest that the pathways influencing immune maturation, hormone levels, and growth trajectory are good targets for intervention studies. They also suggest the utility of using a panel of biomarkers rather than trying to derive a single number to represent biological age.

of these different T-cell populations might serve as a biomarker of longevity. Using the HET-NIA genetically heterogeneous mice described in figure 7.29, the peripheral blood of each of the 559 F2 offspring of this strain were assayed at 8 and 18 months for their levels of several different subsets of T-cells (R. A. Miller 2001). The results, summarized in table 3.5, show that certain T-cell subsets can significantly affect the longevity of their bearers, and assaying the level of these subsets in young or middle-aged mice allows one to correctly predict their longevity. Note that not all T-cell subsets are predictive, and those that are constitute both gender- and age-specific biomarkers. If one mathematically sorts the entire cohort of mice into two clusters based on their predictive T-cell scores, the population is simultaneously divided into two groups that significantly differ in their mean longevities. These data are conceptually similar to the human data displayed in figures 3.17 and 3.21 in that one can use them to sort a population into two different longevity cohorts. Assaying immunological biomarkers in the ongoing monkey and human longitudinal studies might yield useful information regarding their use as a primate biomarker. Using HET-NIA mice, R. A. Miller et al. (2002) showed that body weight in young mice (2–4 months) was a significant predictor of longevity, with lighter animals living longer than heavier ones. Again, the body weight data can be used to sort the population into two different longevity

Panels of Invertebrate Biomarkers. Biomarkers have proven useful in deciphering the aging processes of some model organisms. In yeasts, the

Table 3.5 Ability of Different T-cell Subsets to Serve as a Predictor of Longevity When Measured in Young and Middle-aged Mice Predicts longevity at 8 months in T-cell subset

Description of subset

CD4 CD8 CD4M CD8M CD4V CD4P CD8P

Class II restricted helper cells Class I restriced cytotoxic cells Memory CD4 cells Memory CD8 cells Virgin CD4 cells P-glycoprotein-positive CD4 cells P-glycoprotein-positive CD8 cells

Source: from Miller (2001).

Mated Ɋ

Virgin Ɋ

Virgin ɉ

Predicts longevity at 18 months in Mated Ɋ

Yes

Yes

Yes Yes

Yes Yes Yes

Virgin Ɋ

Virgin ɉ

Yes

Yes

Yes

90 Chapter 3 Measuring Age-related Changes in Individuals late-life increase in generation time can be viewed as a prospective biomarker of mortality (see chapter 4). In the nematode Caenorhabditis elegans, the level of general motor activity is a good statistical predictor of mean and maximum life span (T. E. Johnson 1987). The age of onset of structural alterations to muscle nuclei and the subsequent sarcopenia (age-related muscle loss) is significantly delayed in long-lived mutants, thus providing a possible mechanistic explanation as to why this particular behavioral biomarker serves as an indicator of aging rate in C. elegans (Herndon et al. 2002). In Drosophila, the age at which a fly lost its ability to respond to light or gravity, as well as to generally move about, was indicative of its rate of aging (Leffelaar and Grigliatti 1984). Long-lived animals aged slower because they lost their biomarkers at later chronological times than did the normal-lived controls (Arking and Wells 1990). This biomarker-based observation also focused attention on the fact that the mechanisms responsible for this slower rate of loss of function must begin to operate earlier in life rather than later. In all these cases, the use of biomarkers not only allowed researchers to measure the rate of aging in two or more differently aging strains, but it allowed them to independently confirm their survival curves: Longer lived animals were really aging slower as judged by their rate of biomarker loss. The use of biomarkers was essential to the scientific conclusion.

3.4 Criticisms of the Biomarker Concept There are some long-standing criticisms of the biomarker concept. First is the valid criticism that the construction of biomarkers should not be based on cross-sectional data, for all the reasons presented at the beginning of this chapter. Most of the human longevity biomarker data presented above, as well as all the primate, rodent, and invertebrate biomarker data, are based on longitudinal studies. Thus this criticism was valid some decades ago but has been outmoded by the increasing sophistication of contemporary studies.

Another criticism has to do with mixing biomarkers of disease, or risk factors, with the concept of biomarkers of aging. This is a valid criticism and accounts for the separation of our discussion of human biomarkers of disease from those of longevity. A third criticism is based on the denial that biomarkers of aging can ever exist. The reasoning is that aging is really a parallel progression of multiple, independent, degenerative changes, none of which is driven by some theoretical general aging process. This criticism amounts to circular reasoning, for the assumption of independent mechanisms is used as proof of its existence; it is clearly an illogical statement. In fact, one can predict that if this hypothesis is correct, then there should be no general covariance in rates of aging between different systems nor any correlation between such variables and longevity. But much of the human and animal data presented above demonstrates the apparent existence of such covariance. The facts I review in chapter 7 show that there are systemically acting longevity regulating mechanisms. The hypothesis is not consistent with the known data and must be dismissed. Finally, there is a school of thought, perhaps best exemplified by Masoro (1988a), that questions the whole concept of biomarkers by suggesting that we are in no position to use and interpret biomarkers at this time because we have no good idea as to what constitutes the aging process(es). Although this objection sounds reasonable, a strict acceptance of this position appears to require that we not experiment with biomarkers until we fully understand the aging process. This restriction may be logical, but it is not practical. The evidence linking high blood sugar levels to increased hemoglobin glycation rates or to more extensive cross-linking of collagen, for example, seems to be both persuasive and useful, even if we must admit that we do not fully understand each of the mechanisms involved. As R. A. Miller (2001b) noted, clinicians can use one easily obtained hemoglobin glycation data point as a substitute for the laborious long-term accumulation of blood sugar levels; the glycated hemoglobin data serve as an index of the individual’s prior history of blood sugar levels. If we waited until

3.4 Criticisms of the Biomarker Concept

we understood every aspect of protein crosslinking before we used the glycated hemoglobin data in a clinical manner, then real harm would be done to many individuals. Moreover, this harm would have come about as the result of a conscious decision to withhold the test’s use, and this might well be unethical. Biomarkers then are designed to fulfill the same task for the assessment of aging as the glycated hemoglobin does for the assessment of diabetes. Wilson (1988) has pointed out that there are real risks in not proceeding with the development of effective biomarkers. We need biomarkers not only to guide us in identifying interventions worthy of further study, but also to help keep at bay the charlatans with ineffective treatments (Fisher and Morley 2002). So, biomarkers of aging exist. How then shall we use them? The review above focused on two general types of studies. In one type of study, the goal was to arrive at a unitary numerical index of biological age, to be used in a manner analogous to chronological age. In the other study type, the goal was to arrive at a multiple-factor panel of biomarkers, with no single index of biological age, but with the ability to predict longevity. The former choice gives rise to some conceptual problems, as voiced by knowledgeable commentators (Ingram et al. 1995, p. 707): Although biologic age and biomarkers of aging are related concepts, a distinction should be made between them. The concept of biologic age emphasizes the construction of a single index derived from test results reflecting biologic function. These individual tests are referred to as biomarkers of aging. Research in biomarkers of aging, however, does not imply the need to compile different tests into a single index. Rather, biomarkers research can involve the examination of multiple aging processes to recognize the multidimensional nature of aging and the possibility of an intervention affecting specific aspects of aging (segmental effects), rather than general processes. The comments of McClearn (1997, p. 87) independently yield the same conclusions: “[O]ur comprehension of aging will evolve iteratively

91

from application of a diversity of biomarker variables. Each of these will have strengths and shortcomings from methodological and measurement points of view. The siren song that a ‘gold standard’ index of aging can be found should be ignored.” A second conceptual problem arises from the statistical procedures used to increase the validity with which various variables measure aging by regressing them against chronological age. As pointed out earlier in this chapter, even if the variable showed a perfect correlation with chronological age, all one would have would be a useless alternative expression of chronological age capable only of predicting the subject’s chronological age. However, some alternative mathematical methods do not fall into the same logical trap (Furukawa 1994; Hochschild 1994; Nakamura 1994) and appear to be reliable. But is the variable valid? Is it really measuring changes in aging and the implicit changes in mortality associated with aging? Only the data from BLSA have been formally subjected to the test of serving as a retrospective measure of aging (see figures 3.21 and 3.22). Remember that the biomarker panel used in those reports (Borkan and Norris 1980) had predictive validity only for the individuals at the extremes (either alive or dead, or in the highest or lowest 15% of the test population). Restricting predictive validity to the tails of the normal curve severely limits the utility of such panels. One possible reason for the limited utility may lie in the observation that many biological biomarkers tested for their ability to predict future mortality behave differently with respect to age and/or gender and/or the organism’s environment (see table 3.5). These observations raise a third conceptual problem. The age specificity of such generally accepted biomarkers of aging means that the formulation of biological age is much more complex than simply gathering a group of markers that are highly correlated across the entire age span. This does not mean that biomarker panels are not useful, for the data in figures 3.17 and 3.20 illustrate their potential predictive uses. It does mean, however, that the panel of biomarkers used to assess life expectancy in 40-year-old individuals may be quite

92 Chapter 3 Measuring Age-related Changes in Individuals different from the panel used to assess life expectancy in individuals of other age groups or in individuals of the same age group but subjected to different environmental influences. In turn, this age/gender/environment specificity strongly suggests that the construction of an acrossthe-board unitary numerical index of biological

age is conceptually flawed and that measuring changes in a number of different age-related physiological variables may be useful. Forecasting biological age may well involve monitoring changing suites of physiological responses throughout the adult life cycle. It may be useful, but it does not promise to be simple.

Generation (cell cycle) Virgin cell

Part II Daughter 1

Daughter 2

Daughter 3

Daughter n

Dead cell

1st

Aging

Why Do We Age?

This page intentionally left blank

4

Evolutionary and Comparative Aspects of Longevity and Senescence

4.1 Why Have Long Life and Aging Evolved? Philosophical speculation about the origin of senescence has been with us since at least the dawn of recorded history. Scientific inquiry into this question began in the middle of the 19th century. A complete scientific explanation should have two major components: an accounting of the origin of senescence and a characterization of the cellular and organismal mechanisms involved in the expression of senescence. According to Mayr (1961), the first component addresses the nature of the ultimate processes (the why of aging), and the second component addresses the details of the proximate mechanisms (the how of aging). In this chapter, I discuss only the first component: Why should we age? During the past 150 years, numerous theories dealing with why organisms age have been put forth. Much of the early gerontological writing lacked systematic critical analysis and/or empirical testing of a theory’s predictions. In addition, these essays confused Darwinian selection acting at the level of the individual with group selection (Rose 1991). As Comfort (1964) points out in his classic text, most of these theories assumed that aging and senescence arose as a result of either a particular general property inherent to life (such as cellular wear and tear) or as the outcome of a specific process found in only some forms (such as toxins produced by intestinal bacteria). The first group of hypotheses could have been disproven by evidence of life

forms that did not show senescence; the second set of hypotheses could have been falsified by evidence of senescence in life forms that did not possess the particular trait in question. But few people were concerned with critical testing, so many inadequate theories lived on. The beginnings of our modern understanding have their roots in the work of August Weissman, who in 1891 first drew the attention of biologists to the distinction between somatic line and germ line and explicitly identified senescence as a property of only the somatic cells (Weissman 1891a). The germ line is potentially immortal in the sense that one could trace an unbroken genetic continuum across generations only through the germ line cells. The somatic cells are derived from the germ cells and are fated to age and die. There is no cross-generational continuity between somatic cells. However, the germ cells are not totally resistant to aging processes, since older parents as a group often have a higher frequency of certain morphogenetic abnormalities. But these changes, whatever they might be, seem to affect no fundamental aging traits, for life expectancy is not diminished for the normal offspring of older parents, and this seems to be by far the most common observation. Thus the germ cells seem to be free of the effects of age (but see the results of Gavrilov and Gavrilova [1997], as discussed in chapter 6, which suggest that there may be significant exceptions to this statement). Both Medvedev (1981) and Bernstein and Bernstein (1991) have concluded that the germ cells’ capability of meiotic recombination and repair is what enables them to rejuvenate themselves

95

96 Chapter 4 Evolutionary and Comparative Aspects of Longevity and Senescence continuously at the molecular and cellular levels and thereby evade the deleterious age- and timerelated injuries that eventually impede the functioning of the somatic cells. We now know that the meiotic cell contains various DNA repair functions that either are not present in somatic cells or are present at a greatly reduced level. Avise (1993) points out that all gametes are structurally and functionally autonomous (cellular free-agents as it were) and are thus diametrically opposed to the somatic cells from which they sprang, which are “trapped in a web of interdependencies” (p. 1299). This observation prompted Avise to raise a question originally posed by James Crow: “Is passing through a single cell stage itself important? . . . Starting with a single cell, sexual or asexual, permits each generation to begin with a tabula rasa largely unencumbered by the somatic mutations for previous generations” (Avise 1993, p. 1299). The requirements for independent functioning, even for a short time, may have provided the selective pressure necessary for the germ line’s high level of DNA repair ability. Avise (1993) also extended the original domain of the concept to include the mitochondrial and chloroplast DNAs. These molecules are essential for organismal function, yet appear to have only limited DNA repair systems; this is not so much a paradox as a description of the central role that mitochondria are now known to play in the aging process (see chapter 11). In unicellular organisms, there is no distinction between soma and germ line at the cellular level, although forms such as Paramecium maintain a distinction between somatic and reproductive nuclei within the same cell (as we’ll see later in this chapter). The distinction between soma and germ line is also blurred when regeneration creates an entire new organism from a smaller donor piece. In this form of asexual reproduction, the genetic identity of the original founder organism may be retained for thousands of years. Regeneration is particularly widespread in plants; many popular horticultural and fruit varieties are propagated by grafting or from shoots. The mitotically reproducing somatic cells of plants can often give rise to meristematic cells, which in turn give rise to all other differentiated cells, includ-

ing reproductive cells. Portions of this temporal and spatial clone may senesce and die, yet this process no more affects the identity of the original founding individual than cell turnover in the skin affects our identity. In many cases, however, asexual reproduction does not involve a blurring of soma and germ line, but rather refers to organisms that have a clear distinction between the two yet are capable of parthenogenetically producing fertile eggs. It was once thought that asexual organisms would not senesce. However, it has been shown that senescence does occur in asexually reproducing organisms (Martinez and Leviton 1992). This observation suggests that the formation of somatic cells with a restricted potency is what leads to senescence, and not necessarily the sequestration of totipotent meiotically derived germ cells (Bell 1985). Our interest focuses on the processes that permit the onset of senescent changes in the somatic cells regardless of the mode of reproduction. Modern gerontology may be said to have begun with the acceptance of the need for quantitative analysis and empirical validation. The mid-20th century was chosen by Comfort (1964) as the beginning of this period, perhaps because of the publication of Medawar’s seminal essay on the topic in 1952. Yet we cannot overlook the early contributions of those few pioneers in the field who first glimpsed its future direction. Both Weismann and Wallace in 1891 and Bidder in 1932 had independently searched for and found evidence showing that not all life forms senesce. From this fact, they independently arrived at an evolutionary explanation for the occurrence of senescence in living organisms. In addition to his work on the issue of soma versus germ line, Weismann (1891b) noted that unicellular forms show no signs of aging, and from this he deduced that senescence is inherent to metazoans (both views have now been disproved). Weismann (1891b) postulated that senescence had evolved via the mechanism of natural selection, arising by chance but perpetuated as a positively beneficial adaptation because the “unlimited duration of life of the individual would be a senseless luxury. . . . Worn-out individuals are not only valueless to the species but

4.2 Modern Evolutionary Models

they are even harmful, for they take the place of those which are sound” (p. 24). Weismann built on the much earlier argument of Alfred Russel Wallace (the codiscoverer of natural selection), which was written in about 1865 but published only as a footnote in Weismann’s book (1891b; reprinted as appendix 3 in Finch 1990). Weismann’s argument may be evolutionary, but it is both circular and flawed, for it assumes what it ostensibly tries to explain (that is, the decrease in the probability of survival with the increasing age of individuals) and then denies its own premise (by suggesting that older and less fit “worn-out” individuals can outcompete younger and fitter “sound” individuals). But this hypothesis does have the honor of being the first explicitly evolutionary explanation of senescence to be put forth, and it recognizes the existence of a link between reproduction and aging. Rose (1991) suggests that both Wallace and Weismann had the kernel of the central idea, that natural selection would favor the sacrifice of immortality in exchange for increased reproduction at the level of the individual. Bidder (1925) noted that certain fish exhibit no signs of senescence or loss of vigor with age. These observations led him to postulate in 1932 a logical statement of the evolution of senescence: that the loss of function characteristic of senescence would have no effect on the organism’s reproductive effort if the loss started after most of the individual’s reproduction was completed. The loss of function would be invisible to the processes of natural selection. In Bidder’s view, senescence is not the outcome of a particular process that positively selected for it, but of the organism’s physiological processes operating in the absence of selective pressures. Although this argument was neither quantitatively presented nor empirically proven (Rose 1991), it is a correct encapsulation of the modern view that senescence occurs because the force of natural selection rapidly dwindles away when reproduction is over. We asked, Why must we age? Perhaps we expected a more philosophical answer, so it may strike some as disconcerting to realize that the answer is, Why not? However, this is a liberating

97

answer, for it implies that if a reason can be supplied, then senescence can be delayed. One of the goals of biogerontologists is to supply such reasons, and much of this book is devoted to the description and critique of those efforts.

4.2 Modern Evolutionary Models “Nothing in biology makes sense except in the light of evolution.” This famous quote by the well-known geneticist Theodosius Dobzhansky (1973) sets out the twin tasks before us: to make sense of the fact that organisms age and senesce, and to understand why closely related species may have very different characteristic longevities. Our current understanding is based on the fact that most organisms live in a hazardous natural environment—one where predators, disease, and starvation impose a heavy toll on the population. Because the individual members of any sexually reproducing population are genetically diverse, the environmental hazards are likely to impose a differential survival rate among the different genotypes present. As a consequence, genetic variants that have an enhanced survival rate will have the opportunity to leave more offspring in the next generation. Conversely, alleles that reduce the reproductive success of the individuals carrying them will occur in fewer individuals in the next generation. Over time, some genetic variants and the phenotypes associated with them will be favored, while others are not. In short, natural selection is operating. This situation is complicated by the fact that most known populations are structured by age; that is, the population is composed of individuals of different age classes, each of which represents a different proportion of the population. The high mortality rates common among wild populations, particularly those characterized by no parental care of offspring, mean that not many individuals live long enough to show signs of senescence and aging (see chapter 2). As a result, in any population there are usually many more young breeding adults than old ones (see figure 2.15A). One consequence of this age structure is that

98 Chapter 4 Evolutionary and Comparative Aspects of Longevity and Senescence deleterious genetic variants that act late in life will not be selected against because their carriers likely will have died from environmental hazards before then. Bidder (1932), Medawar (1946, 1952), and Williams (1957) independently achieved this insight. Medawar proposed that new mutations were always being generated at a low rate in any population. Most of these mutations are harmful and would be stringently selected against. However, deleterious mutations that act late in life would not be selected against because most of the organisms that carried them would have died at an earlier age from one or another of the environmental hazards. Consequently, natural selection could not act to decrease the frequency of such genes. They would accumulate and, if and when the high mortality was reduced by environmental manipulation, the larger numbers of older organisms would exhibit the effects of these late-acting deleterious mutations—senescence. This view is called the mutation accumulation hypothesis. Williams (1957) proposed another proximal mechanism through which this ultimate process could be played out. He hypothesized the existence of genes that have beneficial effects early in life but deleterious effects late in life. These genes’ effects are not constant but change as a function of age and/or environment. Such pleiotropic genes, as they are called, would accumulate because they would be positively selected for on the basis of their early beneficial effects, while their late deleterious effects would escape the scrutiny of natural selection due to the premature death of their carriers. One possible example of this category might be the inflammatory response (Van Den Biggelaar et al. 2004). The proinflammatory response is thought to increase early survival by resisting infections in the young, but in the elderly the same response might lead to increased cardiac mortality. Again, the outcome of this scenario would be the appearance of senescence in older, postreproductive organisms. This interpretation is called the antagonistic pleiotropy theory. The population geneticists, beginning with Haldane (1927), Fisher (1930), and Norton (1928), and continuing to the present day (Rose 1991;

Charlesworth 1994a,b), have quantitatively described these models and verified their predictions. Both of these theories appear to be operative to varying degrees in different organisms. Although each of these theories is conceptually powerful, the way in which they would apply to any particular species might vary if some of their underlying assumptions applied differently in that case. Such derived responses have been characterized in the wild and include a case in which guppies in high-predation environments have longer life and reproductive spans than expected (Reznick et al. 2004). Unique environmental circumstances thus may evoke a unique type of life history. Notwithstanding this caveat, the standard theories apply robustly to the model organisms dealt with in this text. Thus, we now have a coherent conceptual framework that allows us not only to understand how natural selection might operate to bring about senescence, but also to empirically test the specific predictions of these evolutionary models. Medawar and Williams have completed Weissman’s task of mechanistically explaining the existence of senescence and, in so doing, have tied the study of aging irrevocably to the central theory of modern biology (see Rose 1991, or Gavrilov and Gavrilova 2002, for a more detailed review of the theories).

4.3 Fecundity and Longevity: The Relationship between Reproduction and Life Span MacArthur and Wilson (1967) were perhaps the first to write about the role fecundity plays in the life history of organisms. If we view evolution as a game in which the goal is to maximize one’s genetic contribution to the next generation, then there are winning and losing strategies for the players in the game. Organisms may be viewed as adopting strategies that result in large numbers of offspring coupled with high mortality (prodigal, or, in the jargon of the field, r-selected, where r represents the rate of population increase in equations describing population growth) or small numbers of offspring coupled with lower mortal-

4.3 Fecundity and Longevity: The Relationship between Reproduction and Life Span

ity (prudent, or K-selected, where K represents the carrying capacity of the environment in equations dealing with population growth). Table 4.1 lists the attributes of each strategy. These strategies should be viewed more as the polar ends of an intergrading spectrum of strategies rather than as two discrete choices. High fecundity is a necessary characteristic of r-selected organisms, but a long life span is not. A good example of an r-selected life history is that of the common meadow vole (Microtus pennsylvanicus). This small rodent matures at less than 1 month of age and produces litters of five young every 3–4 weeks throughout the year. In captivity, one vole produced 17 litters in 1 year! Although some individuals have lived as long as 28 months in the laboratory, in the wild only about 0.1% of the animals reach the age of 10 months (Solomon and Vandenbergh 1994). The average longevity of these voles in nature has been reported as about 1.2 months, a value consistent with the very high death rate in the wild. At the one extreme, the r strategy also describes those organisms that reproduce once in large numbers and then die. The young must fend for themselves with no parental care. Once the next generation has reached the age of reproductive maturity, there is no adaptive advantage for them to live any longer than is necessary to reproduce. The Pacific salmon (Oncorhynchus spp.) is perhaps the best-known example of this strategy. Another similarly extraordinary case is that of the marsupial “mouse” of Australia (Antechinus and Phascogale spp.). In A. stuartii and P. tapoatafa, the males live no longer than 11.5 months. Near the end of their life, they stop eating and

Table 4.1 Traits Characteristic of Alternative Life-History Strategies Prodigal (r-selected)

Prudent (K-selected)

Many young Small young Rapid maturation Little or no parental care Reproduction once

Few young Large young Slow maturation Intensive parental care Reproduction many times

Source: from Curtis (1983).

99

engage in a competitive, brief, and frantic mating period. All males die shortly thereafter, probably as a result of hormonally induced stress. For some period of time, the only males in the entire population are the male embryos being carried in utero by their mothers. The mothers live long enough to suckle their young and wean them. Very few females live long enough to breed a second time. At the other extreme, the K strategy encompasses the behavior of mammals such as humans that have a low reproductive rate coupled with a higher degree of parental care. In this case, there is clearly an advantage for the adult to live a substantial length of time past the age of reproduction, for the offspring will not survive in the absence of long term parental care (see chapter 14). A long life span is a characteristic of many K-selected organisms. The distinction between the r and K strategies, then, revolves about the pattern of population growth. As illustrated in table 4.2, r-selected organisms produce large numbers of individual offspring, each with a low individual probability of survival. If all the offspring survived, the population would grow exponentially and would shortly suffer the effects of overpopulation. Even with a high mortality rate, there are more than enough young to ensure that sufficient numbers will survive to adulthood to maintain the population. Adults of species that adopt the r strategy tend to be short-lived organisms whose major

Table 4.2 Reproductive Capacity of the Housefly (Musca domestica) Generation

No. if all survive

1 2 3 4 5 6 7

120 7,200 432,000 25,920,000 1,555,200,000 93,312,000,000 5,598,720,000,000

Source: adapted from Kormondy (1969). Note: In 1 year, about seven generations are produced. The numbers are based on each female laying 120 eggs per generation, each fly surviving just one generation, and half of these being females.

100 Chapter 4 Evolutionary and Comparative Aspects of Longevity and Senescence

Population density

3

4

2

In evolution, the name of the game is to play again. Individuals are competing with others of their species to maximize the number of copies of their genes that are represented in the next generation. The strategy used by the members of any given species is the outcome of their specific developmental and ecological characteristics; lifehistory analysis suggests that the most important characteristics are those related to reproductive practices. Empirical evidence of the close link between reproductive schedules and longevity, at least for mammals, is shown in figures 4.2 and 4.3. In figure 4.2, the age at first reproduction is strongly correlated with life expectancy at birth. Figure 4.3 shows that there is an inverse relationship between an organism’s reproductive potential and its longevity. Taken together, these figures present the quantitative data underlying the concept of r and K life-history strategies. Two other examples will suffice to show that organisms have some room to manipulate life history in view of changing environmental conditions. First, when parasitic wasps are raised

0.8

Relative age at first reproduction

adult function is to reproduce. K-selected organisms, in contrast, combine a lower reproductive potential with a higher probability of individual survival. They utilize parental care to maintain their population numbers at the carrying capacity of their environment (figure 4.1). Adults of species that adopt this strategy tend to be longerlived individuals for which reproduction per se constitutes only one of their adult functions. Carey (2003) has pointed out that K-selected organisms tend to use two different longevity strategies. One involves the need for some species to wait for intermittent or patchy environmental conditions before they can reproduce; the need to wait for these occasional but absolutely necessary circumstances may constitute a selective pressure in favor of long life. Marine tortoises or deep sea fish are examples of this category. The other strategy involves social organisms in which overlapping generations live together; the feedback effects of certain social interactions may constitute a selective pressure in favor of long life. Humans are a good example of this category. These two K-type strategies are not mutually exclusive, and some species may use both. A more detailed discussion of these strategies and their role in understanding the life history of longevity is presented in chapter 14.

0.0

–1.2 –1.2

1

0.0 Relative life expectancy

0.8

Time

Figure 4.1 The growth curve of a K-selected organism. Biological growth curves are usually sigmoidal, or S-shaped. As with exponential growth, there is an establishment phase (1) and a phase of rapid acceleration (2). Then, as the population approaches environmental limits, the growth rate slows down (3 and 4) and finally stabilizes, although minor fluctuations around K (the mathematical symbol representing the carrying capacity of the environment) may continue. (After Curtis 1983.)

Figure 4.2 The relationship between relative age at first reproduction and relative life expectancy at birth for natural populations of mammals. Relative values refer to deviations from logarithmic regression lines of age at females’ first breeding, or expectation of life at birth, on adult female body size. The correlation coefficient (r) = of 0.98 is only slightly decreased to 0.89 (p < .001) by removal of the effects of body size through partial correlation. The symbols refer to different mammalian genera. (After Holliday 1995.)

4.3 Fecundity and Longevity: The Relationship between Reproduction and Life Span

150

Potential offspring (logarithmic scale)

100

50

20 15 10

0

0

10

20

30 40 50 60 Longevity (years)

70

80

Figure 4.3 The relationship between reproductive potential and longevity of 47 genera of eutherian mammals. Reproductive potential is the maximum number of offspring that might be produced under ideal conditions. Longevity is based primarily on the maximum life span of limited numbers of individuals in captivity. (After Holliday 1995.)

under specific environmental conditions indicative of a decreased life expectancy (such as decreasing barometric pressure or photoperiod), they alter their reproductive behavior by laying many eggs early in life, even in suboptimal environments. In the absence of these specific cues, and thus faced with the environmental conditions associated with a longer life span, the wasps lay fewer eggs early in life but deposit them in optimal environments. The simplest interpretation of these observations is that the wasps are sensitive to conditions that affect their potential longevity and alter their behavior to maximize their lifetime reproductive fitness (Roitberg et al. 1993). This is not unique to wasps; certain fish, when faced with deleterious conditions, will lay larger numbers of smaller eggs (possibly shortening their offspring’s life spans) in an effort to maximize the mother’s own fitness (Elnum and Fleming 2000). Second, it is now well known that calorierestricted animals live significantly longer than animals fed ad libitum, an observation discussed in more detail in chapters 6 and 7. The im-

101

portant point in this context is that calorierestricted animals become partly or fully sterile but can breed again once food becomes more plentiful, and they can do so at a later age than can animals fed ad libitum throughout their lives. Thus, one can interpret these observations as showing that animals can reallocate their available energy to maximize somatic maintenance and individual longevity when their energy resources are so low as to jeopardize the probability of successful reproduction. The life-history strategy an organism adopts reflects a particular balance between somatic maintenance and reproduction, but this balance is not static. The organism apparently has the ability to modulate this balance significantly when environmental conditions change, and hence temporarily to adopt a more suitable life-history strategy. The fact that such longevity regulating abilities already exist means that (1) the life span is plastic such that there are environment-dependent longevity phenotypes (Arking et al. 2002c; McClearn 2004), and (2) it might be possible to use these existing abilities as a springboard for the development of effective longevity-extending interventions (see chapter 15). Studies done on small mammals suggest that some populations do in fact show signs of aging and senescence even in the wild (Gaillard et al. 1994; Slade 1995). Does the presence of older individuals in wild populations mean that aging was selected for by natural selection? Is there a direct selection for old animals? Such a conclusion would directly contradict the tenets of the modern evolutionary theories of aging as discussed above. Fortunately, there is another explanation for the presence of older animals in wild populations that does not require such a conclusion and is consistent with both evolutionary theory and fact. An organism of postreproductive age has a great selective advantage if it is able to maintain its physiological vigor for as long as the parents bear even some responsibility for the survival of the young (and thus for the transmission of copies of their own genes into the next generation). Reproduction does not end with the birth of the babies but involves some variable period of postnatal care. Parents that die before

102 Chapter 4 Evolutionary and Comparative Aspects of Longevity and Senescence portance is a theoretical analysis by Kirkwood (1985, 1987) that points out that the organism must channel and apportion its energies into reproductive activities, as well as into the maintenance and repair of its soma. Although the energy cost of making eggs or sperm probably stays more or less constant over time and is therefore the same for both young and old, this is not the only energy cost incurred in reproduction. The energy costs of courtship, as well as of pregnancy and child rearing, are high and represent a significant investment of energy by the organism. In addition, some energy must be devoted to the repair and maintenance of the soma if the organism is to survive. It is reasonable to assume that even a well-fed organism has only a certain amount of energy available to it. Thus the problem facing the organism is how best to allocate its finite metabolic energy to maximize both reproduction and survival. Kirkwood (1985, 1987) mathematically compared the success of different allocation strategies of investment in somatic repair and maintenance on reproduction and survival. As figure 4.4 shows, increasing the amount of effort expended on somatic repair results in an increased survivorship but a decreased fecundity. One can use these results to calculate the joint effect of changing survivorship and fecundity on the evolutionary fitness of a genotype. The theoretical and empirical results, shown in figure 4.5A, lead to several conclusions. First, there is a clearly defined range

this task is complete place their offspring at high risk and accordingly lower their own reproductive fitness. Under these circumstances, there should be a strong selective pressure to increase parental longevity so as to successfully complete reproduction. This is one example of the social-interaction K-type strategy discussed above. There is reason to believe that the evolutionary trend toward increased postreproductive longevity seen in many groups of organisms is not separable from the processes that operate before the end of reproduction, but rather can be viewed as being part of the same life-history strategy designed to ensure the survival of an organism’s genes into the next generation (Holliday 1996a). In addition, there are other reasons for the occurrence of senescent animals in the wild, and I investigate them at some length in chapters 5 and 14. What, then, keeps K-selected organisms from living forever? Why doesn’t a long-lived animal continue to reproduce and to live indefinitely? First, some parts of an organism, such as the heart, are absolutely essential for life, have no redundant or backup organs to take on their functions if they should fail, and have no significant repair or regenerative mechanisms. Given enough time, all such organs will eventually fail. They thus constitute an inherent weak link, the failure of which will lead to death for those members of the cohort that do not die as a result of random accidents. Given such accidents and system failures, immortality is not possible. Of more general im-

(b)

Fecundity

Survival (lx)

(a)

Age

Age

Figure 4.4 The effects of increasing the energy investment in somatic repair and maintenance on (a) survivorship (l x) and (b) fecundity. The three curves in each graph represent three different levels of energy investment. Survivorship increases and fecundity decreases as more energy is invested in somatic maintenance. (After Kirkwood 1987.)

4.3 Fecundity and Longevity: The Relationship between Reproduction and Life Span

of investment in somatic repair at which reproduction will be at a maximum. Second, this point of maximum reproductive effect will be at a level that is lower than the minimum energy level needed for indefinite somatic repair (immortality). Reproduction requires less energy than repair. Third, allocating energy to maximize somatic repair will decrease the organism’s Darwinian fitness. The empirical data in figure 4.5B show that the penalty for overinvesting in somatic maintenance is a very severe reduction in one’s Darwinian fitness. This strong and real constraint on the animal’s ability to maximize somatic maintenance

103

is one reason the world is not overrun with longlived creatures. A dual solution is required in which the animal can combine maximal reproduction with some optimal longevity. In most cases, decreased fecundity over a longer life span yields fewer copies of an individual’s genes in the next generation than does a higher fecundity over a shorter lifetime. Thus, fitness should be maximized at a repair level lower than that required for indefinite somatic repair. Hence we die. The theory seems plausible. Is it correct? Experimental data on fecundity, energy metabolism, and longevity in a particular Drosophila

Investment in somatic maintenance

A .4

Rate of natural increase Nonaging

.2 Aging

0 -.2 -.4 -.6

0

S'

S*

1

Investment in somatic maintenance

B 1500 RS

RS = 1150

1000

500

Wm

0 0

20

40

60

80

100

Figure 4.5 (A) The theoretical relationship between the intrinsic rate of natural increase and the level of investment in somatic repair and maintenance. The optimal proportion of investment in maintenance, S*, is less than S', the amount of energy investment required for indefinite somatic maintenance. Note that the theory predicts that investing energy in somatic maintenance will increase reproductive fitness until some optimal level (S') is attained, after which more investing more energy results in a decreased fitness. (After Kirkwood 1990.) (B) The actual relationship between the predicted lifetime reproductive success (RS) of the Wayne State Drosophila strain as a function of investment (ml O2/day) in maintenance, Wm. Zero investment in Wm yields zero RS. Increased investment in Wm yields increased RS until the optimal point of 1150 eggs/ lifetime is reached. This corresponds to Wm = 86 ml O2/day, which corresponds exactly to the independently determined experimental value. (After Novoseltsev et al. 2002.)

104 Chapter 4 Evolutionary and Comparative Aspects of Longevity and Senescence strain has been used to calculate the optimal combination of fecundity and longevity that would allow the fly to attain the highest Darwinian fitness value (Novoseltsev et al. 2002). The calculated values are consistent with the flies’ observed patterns. A mismatch between the available energy and the power demands made on the fly due to reproductive or maintenance processes leads to death of the organism (Novoseltsev et al. 2003). Taken together, these data suggest the importance of energy allocations to longevity, and they provide a mechanism that links reproduction to aging. Evidence showing that reproduction and longevity are tightly linked has been found in organisms ranging from bacteria to humans and is presented at various points throughout the book. Although more testing of the theory’s predictions need to be done, the data available support the correctness of the “disposable soma” theory. This process should not seem alien or far fetched, for it is nothing more than the cost– benefit analysis that most of us have made when faced with the decision of whether to continue investing our hard-earned money in repairs to the old car or to invest our money in a new car. At some point, the cost of repairs exceeds the cost of purchase. Hence, old cars are discarded (or traded in for newer ones). The important point is that this analysis implies that life span is modulated not by genes whose primary action is to shorten or extend life, but by the longevityassurance genes that function in the processes involved in somatic maintenance and that alter life span as a pleiotropic result of this primary action (Sacher and Hart 1978; Williams 1957). This analysis of ecological and population genetics has summarized the reasons that lead us to believe that patterns of longevity are inseparable from, and flow out of, the life-history strategy of an organism. Ecology and evolutionary studies also suggest that each organism makes its living in its own distinctive manner. Yet beneath this obvious diversity lies a fundamental identity that links together all living creatures. This fundamental similarity shows itself in the fact that diverse spe-

cies accomplish biological tasks in one of a small number of ways. For example, all eukaryotic organisms (and most prokaryotic ones as well) digest food molecules and extract energy from food using almost identical metabolic processes. The cells of a yeast and the cells of a human undergo the same process of cell division, controlled by almost identical genes. Neural circuits work in the same way in snails and insects and humans. This similarity does not extend across all levels, however. The evolutionary similarities are most obvious at the cellular level and less obvious at the population level. For example, even closely related species, such as chimpanzees and humans, use a similar brain to organize widely divergent but adaptive behavioral and social responses. Because life-history strategies are based on those adaptive behavioral and social responses, it is not surprising that phylogenetic relationships should be less obvious at these higher order levels than at cellular and molecular levels. In other words, closely related species may have widely divergent life expectancies if they follow different lifehistory strategies. The virtues of a comparative approach to the study of aging are threefold: • It should allow us to survey the diversity of living creatures and determine how common and widespread aging is. • It should allow us to identify the biological mechanisms that appear to play a causal role in regulating the rate of aging in different species. Uncovering novel methods of regulating fundamental processes may provide valuable knowledge. • It should allow us to overcome the ethical, legal, financial, and temporal constraints associated with doing experiments with humans or any of the ordinary laboratory rodents. The comparative study of aging has both theoretical and practical importance. In our comparative survey of aging, I try to make clear how our investigations have fulfilled, at least in part, each of these three goals.

4.4 Strengths and Weaknesses of the Evolutionary Viewpoint

4.4 Strengths and Weaknesses of the Evolutionary Viewpoint The overwhelming strength of the evolutionary viewpoint is that it has integrated the study of senescence into the mainstream of biological thought. It has allowed us to study longevity as just another phenotype and to separate its scientific analysis from our emotional reaction to it. The study of disease is important in its own right, but to conceive of aging and senescence only in the context of the medical model is to view these processes as an incoherent assortment of symptomatic bodily failures, with no obvious connection to the evolutionary principles implicit in genetics, development, physiology, neuroscience, and in all other aspects of biology. Even diseases have been examined through an evolutionary lens (Nesse and Williams 1994). A strict adherence to the medical model implicitly relegated senescence to the sidelines, but viewing senescence through an evolutionary prism allows us to understand the Darwinian connection between aging and reproduction. This connection is not merely one of professional pride. The evolutionary model suggests testable and broad theories, something that the medical model has not done. It also makes clear the distinction between ultimate and proximal causes of senescence, to borrow the phrasing of Mayr (1961), and lets us work out the underlying mechanisms. The empirical data collected as a result of inquiries into ultimate causes demonstrate that the life span of experimental animals can be significantly modulated as a result of alterations in reproductive behaviors, thus linking senescence to reproductive fitness. The data also provide us with a coherent idea of the interplay between repair and reproductive functions. Longevity is not a phylogenetically consistent trait, but its connection with species-specific life-history strategies and the need of an individual to pass copies of its genes to the next generation enables us to understand why longevity isn’t consistent. The evolutionary theory is based on the analysis of gene frequencies in aclonal or sexually re-

105

producing organisms. It is not surprising that the current concepts are somewhat less successful in explaining the existence of senescence in clonal forms. These genetically invariant organisms are potentially immortal, but at least some such organisms senesce (Martinez and Leviton 1992). One way to explain the paradox is to assume that deleterious mutations are inevitable in such forms because the repair processes present in germ cells are absent in these clonal somatic cells (Medvedev 1981b). One can show mathematically that the mean physiological and reproductive fitness of a finite population will decline because there is always a chance that the class of individuals with the fewest deleterious mutations will be accidently lost. Without the benefit of sexual recombination, there is no way to recover this class. This decline in fitness has been called “Muller’s ratchet” (H. J. Muller 1964) and has been postulated to lead to the evolution of senescence (Bell 1988; Maynard Smith 1988; Partridge and Barton 1993). However, this situation also implies that the mean duration of existence of clonal species should be significantly less than that of aclonal species because they should succumb to this mutational load. The available data suggest that this prediction does not hold for a variety of different organisms. For example, the evolutionary persistence of the clonal species of corals is equivalent to that of aclonal species, both of which last about 9 million years (Finch 1990). Certain asexual freshwater crustaceans have a fossil record that extends back at least 268 million years (Butlin and Griffiths 1993). If mutations are accumulating, they do not seem to have a discernible evolutionary effect. One way to escape this paradox is to assume that asexual species are composed of large populations, a condition that may allow them to escape Muller’s ratchet (Bell 1988). However, large populations for asexual species seem unlikely in at least some cases (Butlin and Griffiths 1993). A recent extension of the basic evolutionary theory suggests that the population growth rate is important in determining whether senescence will evolve in an asexual population (Orive 1995). Although clonal reproduction does retard the

106 Chapter 4 Evolutionary and Comparative Aspects of Longevity and Senescence evolution of senescence, it does not preclude it. Senescence in clonal species is not well understood, but it appears to be amenable to continued analysis. Nor is it clear how programmed aging, indicated by a synchronous, nonrandom mortality among different branches of the colony, could have arisen in asexually reproducing colonial forms (Rinkevitch et al. 1992). All of the data and interpretations are complicated by the fact that many clonal species retain the option to reproduce sexually at times. In sexually reproducing clonal species, reproductive fitness increases as the size of the colony increases; the fact that they become sexually mature when they reach the extrinsic limits of colony growth shows that the life history in these forms is still driven by their reproductive strategy (Harvell and Grossberg 1988). The loss of sexual reproduction may obscure but does not obliterate the evolutionary connections between reproduction and longevity because the major determinants of reproductive fitness in clonal organisms are colony size and organization. Current evolutionary theory deals in general solutions and says little regarding the details of longevity and senescence in specific taxa. One specific deduction from the theory that some have made is that there should be great diversity in the details of senescence in different life forms precisely because of the decreased force of natural selection in the postreproductive stages. However, a survey of mammals reveals what appears to be a common pattern of senescence over diverse species with a 30-fold difference in life spans (Finch 1990). In general, the force of natural selection on survival begins to decrease at the age of sexual maturity and drops to zero after the reproductive period. The result is an age-related increase in mortality rate (qx) as discussed in chapter 2, which gives rise to the commonly observed mortality pattern in most mammals. The age-related increase in mortality suggests three new possibilities. First, the common senescence pattern is a primitive mammalian trait that has persisted unchanged for at least 50 million years. Second, it may represent convergent evolution occurring in the absence of natural selection. Third, it may reflect the inevitable outcome in individuals of certain fundamental develop-

mental constraints common to all mammals. These could include finite supplies of stem cells in the immune and other systems, the absence of repair and regeneration mechanisms in the heart, and so forth. I believe the first and third possibilities are the most likely. Senescence patterns may be constrained by the involvement of highly conserved genes regulating the balance between fecundity and senescence in all forms studied (chapter 6). They may also be constrained by developmental patterns that give rise to the body plan and to the distribution of rate-limiting molecules and cells (chapter 5). One human weakness of some practitioners of the evolutionary approach is to assume that all creatures behave just as their favorite laboratory animal does. This may not be the case, particularly if one is talking about different taxa and different environments. Another weakness is to assume that one’s favorite theoretical genetic mechanism must be true, despite what someone else’s data might suggest, especially if obtained with a different species. This problem is obvious in the technical discussions regarding which theory better explains the results obtained in experiments with laboratory animals—the mutation accumulation theory or the antagonistic pleiotropy theory (Charlesworth 1994b; A. G. Clark 1994; Schnebel and Grossfield 1988)—when there is some evidence to suggest that both sorts of mechanisms may be playing a role (Service et al. 1988). Another human weakness in applying the evolutionary approach was the reluctance to consider that a process as complex as aging and senescence could be brought about by single genes with a large effect. However, recent work has shown that this is indeed the case, and the identification and characterization of such genes are described in chapter 7. The new facts have forced the improvement but certainly not the abandonment of the evolutionary models. The continuing development of quantitative models, such as that of Novoseltsev et al. (2002) discussed above, which take into account both the phenotypic trade-offs involved within the context of both the genetic architecture in the cell and the epigenetic interactions between the genomes of different cells, promise to better reflect the developmen-

4.5 Comparative Aspects of Aging

tal genetic history of the organism and to better predict the magnitude and direction of evolutionary change in the expression of senescence. We have focused our attention only on the evolutionary explanation for aging. But there are a larger group of questions having to do with lifehistory theory and the evolution of aging and longevity. The evolutionary theories of aging are the intellectual core of life-history theory, but they do not address all aspects of the latter (Carey 2003; Gavrilov and Gavrilova 2002). I address these questions in chapter 14. The faults I have listed are those commonly observed when theory is being challenged robustly; these faults are not inherent in the evolutionary explanation of aging and senescence. The fact that the details of the theoretical mechanisms are still being worked out does not invalidate the entire theory. All in all, modern biogerontology is based on the concept that senescence makes sense only in the light of evolution.

4.5 Comparative Aspects of Aging

Prokaryotes

Eukaryotes

At present, most biologists classify all living organisms as belonging to one of five different major kingdoms. These kingdoms are distinguished from one another by their possession of certain structures, the presence or absence of which irrevocably casts their members into certain ways of life. The absence of a nucleus, the discrete membrane-enclosed storage compartment for the genetic information, distinguishes the kingdom Monera (bacteria; also called

Plants (synthesize their own food)

Fungi (digest food externally)

prokaryotes) from all the other forms of life (eukaryotes). The eukaryotes are divided into large groups depending on whether they consist of unicellular forms (kingdom Protista) or multicellular forms (kingdoms Plantae, Animalia, and Fungi). The multicellular forms are distinguished from each other by their method of obtaining food: Plants synthesize their own food via photosynthesis, animals feed on other organic foodstuffs and digest them internally, and fungi feed on other organic foodstuffs but digest them externally. These relationships are shown in figure 4.6.

4.5.1 Unicellular Organisms: Senescence Arising from Nucleocytoplasmic Interactions The single-celled organisms are extraordinarily diverse, belonging to both the kingdoms Monera (bacteria) and Protista (unicellular algae and protozoans), and thereby comprising some species that are prokaryotes (cells without a nucleus) and others that are eukaryotes (cells with a nucleus and other organelles). Evidence of aging and senescence has been found both in bacteria and in protists. The ubiquitous appearance of aging in all known kingdoms supports the view that aging is a fundamental aspect of living organisms. 4.5.1.1 Bacteria

It was once thought that bacteria did not age. We now know this is not true; bacteria have been aging for a long time, but we have just learned

Animals (digest food internally)

Protists Algae

Protozoans

Monerans Bacteria

107

Viruses

Figure 4.6 The five-kingdom classification scheme of living organisms.

108 Chapter 4 Evolutionary and Comparative Aspects of Longevity and Senescence about it. It was long known that bacteria continuously divide when raised under optimal conditions. It was also well known that bacteria could mobilize various sorts of stress responses when the external environment changed in an adverse manner. Only in recent years were these phenomena viewed as representing two different phases that constitute an aging response. Bacteria subjected to nutrient depletion enter a nonproliferating phase and gradually lose their ability to recover and reproduce. This process is referred to as conditional senescence induced by growth arrest, and it has several interesting similarities to aging processes characteristic of eukaryotes (Nystrom 2002). One obvious and important similarity is the trade-off between reproduction and maintenance. Growth-arrested cells are highly resistant to a variety of environmental stresses. This resistance is conferred on the cells when they activate a number of different genes coding for resistance to various stresses. Two things are of interest here. First, the function of these genes and their protein products are similar to those genes induced in long-lived worms and flies in that they protect the cell against the same sort of stressors (e.g., heat, cold, oxidative stress). This similarity reflects the fact that all living systems are susceptible to the same kinds of environmental factors, and so all need to erect defenses against them. I discuss stress resistance thoughout much of the book and make clear its relationship to longevity and senescence in chapters 7 and 9. Second, the induction of these 50 or so genes relies predominantly on a single regulator: the sigma factor (s). Thus the whole range of bacterial stress resistance is brought about mostly by one regulatory molecule. The sigma factor is a particular protein subunit of the enzyme RNA polymerase, and it comes in two closely related forms, sS and s70, the genes for which differ from each other mostly in the region of the subunit that codes for a particular promoter binding site (Hengge-Aronis 2002). Thus the two subunits will confer a different specificity for target genes on the resulting RNA polymerase molecule. Either one or the other of these subunits, but not both, can bind to the same enzyme molecule. What determines which one will do so?

The binding mechanism is diagrammed in figure 4.7. When the cell is exposed to stressful conditions (carbon starvation, caloric restriction, amino acid restriction, stress), high levels of a particular molecule, guanosine tetraphosphate, or ppGpp, are generated. The high concentration of ppGpp molecules allows them to bind readily to the forming RNA polymerase molecule. Such ppGpp–RNA polymerase molecules preferentially bind with the s70 subunit. The resulting completed enzyme molecules preferentially activates the stress-resistance genes mentioned above, but they will not activate the genes needed for growth and proliferation. Thus, these bacteria will cease to grow and will transit into the conditional senescence stage, where they are highly resistant to many environmental insults. However, if the environmental conditions outside the cell are conducive to growth, then only low levels of the ppGpp molecule are produced. Under these conditions, no ppGpp molecules are available to bind to the forming RNA polymerase, and so the enzyme preferentially binds only the s70 factor. Such enzyme molecules preferentially activate the growth and proliferation genes but not the stress-resistance genes. Thus, these bacteria will continue to grow and proliferate but are very sensitive to various stressors (Jishage et al. 2002). In effect, the bacterial cell uses the concentration of the ppGpp molecules as an approximate indicator of the external environmental conditions, and the presence or absence of this indicator sets off two (more or less) mutually exclusive molecular cascades that culminate in the activation and repression of two different gene sets. The molecular structure of this cascade forces the bacterial cell to make a binary choice: It can grow rapidly and be sensitive to stress, or it can cease growing and be stress resistant. The binary choice forces the cell to trade off one attribute for the other; it cannot simultaneously express both (Kurland and Mikkola 1993). As Nystrom (2002) points out, this situation is formally equivalent to the disposable soma model discussed earlier (figure 4.5) in that it allows the cell to allocate its available energy to either reproduction or somatic maintenance. The bacte-

4.5 Comparative Aspects of Aging

Optimal Conditions

109

Carbon Starvation, CR, Stress

ppGpp

ppGpp Low

ppGpp High

RNA polymerase

RNA polymerase

S

Growth & Reproduction Genes

70

Maintenance & Survival Genes

Figure 4.7 A model for the trade-off between reproduction and survival in bacteria. The model is based on the argument that RNA polymerase (RNAP) is limiting for transcription and that sigma transcription factors, such as sS and s70, compete for binding to RNAP. This competition is regulated by the nucleotide ppGpp, which accumulates during deleterious conditions, enhancing the binding of s70. The resulting RNAP complex primarily activates the genes involved in maintenance and survival, while repressing the genes involved in growth and proliferation. This causes growth arrest of the bacterial cell. However, in optimal conditions, there are only low ppGpp concentrations, little or no binding of the nucleotide to the RNAP occurs, and the RNAP preferentially activates the growth and proliferation genes while repressing the genes involved in maintenance and survival. (Redrawn after Nystrom 2002.)

ria are conforming to the same fecundity/longevity trade-off as are other organisms. This mechanism, in which extracellular signals are picked up by membrane receptors and changed into intracellular signals, which then set off a molecular cascade, eventually resulting in changes in gene expression, explains the process by which the evolutionary trade-offs are made in bacteria. It is essentially similar to the process of cell signaling found in eukaryotes, known as signal transduction, and its similarity suggests the long-term conservation of this general mechanism. If the growth-arrested bacterial cells are stressresistant, then why should they eventually lose their ability to recover and reproduce (i.e., undergo senescence)? It turns out that their loss of function is only indirectly due to oxidative stress, which is one of the major proximal senescence

mechanism in eukaryotes. Aging bacteria initially show a decline in ribosome fidelity, and so they begin to incorporate the wrong amino acids into polypeptides. Such abnormal polypeptides show a high rate of misfolding and give rise to malformed proteins. These abnormal proteins are apparently susceptible to oxidative damage and modification and begin to accumulate in the cell, where they constitute a target for secondary oxidative damage (Dukan et al. 2000). The evidence indicates, however, that it is the presence of the abnormal proteins and not their oxidative damage that is responsible for the cell’s progressive loss of function and senescence. But this is not the only way bacteria may undergo senescence. It was recently shown that the bacterium Caulobacter crescentus also ages, but by means of an apparently different process

110 Chapter 4 Evolutionary and Comparative Aspects of Longevity and Senescence (Ackermann et al. 2003). C. crescentus lives as a stalked cell attached to a substrate. It undergoes an asymmetric mitotic division, which leads to the formation of a free-swimming swarmer cell and the stalked mother cell. This stalk does not appear to be replaced during the entire life span of the cell. The reproductive output of the stalked mother cells decreases with age, which is a good operational definition of aging (see chapter 1). The systemic distribution of old cell components to only the mother cell might underlie the aging process in this organism. A similar process is thought to be involved in certain aspects of reproductive aging in yeast (see below). Aging and senescence occur in bacteria. The evolutionary trade-offs between reproduction and somatic maintenance, as well as the generic nature of the signal transduction mechanisms involved, appear similar to the processes seen in eukaryotes. The senescent mechanisms seem to differ somewhat from those characteristic of eukaryotes, apparently being dependent on an increased error rate in the ribosomes, something not seen in eukaryotes. It will be interesting to determine to what extent bacterial senescence is related to eukaryotic mechanisms and to what extent the bacteria have novel mechanisms not known in other kingdoms. 4.5.1.2 Protists

Protists, single-celled eukaryotic organisms, have been called “nature’s experiments,” encompassing as they do an almost bewildering variety of genetic organizations and mating behaviors. They can thus be viewed as a rich resource for deciphering how different genetic systems interact with the environment to produce species-characteristic life spans. The life spans of protists range from about 50 days for Paramecium tetraurelia to immortality for Tetrahymena thermophila (SmithSonneborn 1985). These, of course, are clonal life spans. A clone comprises the daughter cells that are derived from a single cell and that are genetically identical to their progenitor. These organisms can reproduce either sexually or asexually. Clonal offspring, which are genetically identical to one another, are produced only by asexual

(mitotic) cell divisions. In most species, in the absence of sexual fertilization, the probability that a cell will give rise to viable progeny at the next cell division decreases as the time increases since the last sexual (meiotic) fertilization. The clone deteriorates, and eventually all members of the clone die. Thus, the clonal life span value is the sum of the life spans of all the mitotically produced, genetically identical but physically distinct cells descended from the single progenitor cell. This clonal life span concept is used in other aspects of studies on cell aging (this is the “Hayflick limit” that I discuss in chapter 12); it is the concept you would use if someone were to ask you the age of your skin. As discussed in some detail in chapter 5, the individual cells that make up our skin each have a life span of about several weeks. Yet our skin persists from before birth until death, despite the turnover of most of its component cells, much as a candle flame persists despite the rapid appearance and disappearance of the molecules undergoing combustion. The age of the skin is the summed age of all its component cells. Let’s briefly examine the aging process as seen in three of the better-known protists: the common amoeba and the graceful ciliated paramecium and Tetrahymena. I also examine the patterns of aging of another interesting organism, Volvox, which appears to lie just over the boundary between unicellular and multicellular organisms. Perhaps you have encountered these organisms in an introductory biology laboratory. If so, it is probably reasonable to assume that, whatever you thought of them at the time, you never considered them as undergoing aging. Yet some do. From these simple unicellular organisms we might be able to detect some of the mechanisms operating within our own cells. Amoebae. Amoebae appear to be immortal, provided it is maintained in an exponential growth phase. When its food supply is restricted for a substantial period of time (3–5 weeks), the organisms are induced to switch over to a finite life span ranging from 4–30 weeks. These mortal cells show two types of behavior. On cell division, type A cells produce one viable daughter cell and

4.5 Comparative Aspects of Aging

one inviable daughter cell. Type B cells produce two daughter cells, both of which will live until all cells in the clone die. Experiments in which nuclei and/or cytoplasm from one cell type were transplanted to another led Muggleton-Harris (1979) to suggest that this odd behavior is the result of a complex nucleocytoplasmic interaction. These experiments also suggest that the shift away from immortality is accompanied by the appearance of type B factors in the cytoplasm or by the alteration of the nucleus from type B to type A. The molecular basis of these alterations is not known. The importance of these observations lies in the suggestion that the aging process in this simple asexually reproducing cell is under the control of certain molecular signals, that the presence or absence of such signals depends on the environment, and that these signals involve an alteration of the nucleus via the cytoplasm. This information suggests that the aging process in higher organisms also has a molecular genetic basis. Paramecia. Paramecia, unlike amoebae, but like many other ciliates, contains two kinds of nuclei. The micronucleus is the germline nucleus and shows little transcriptional activity other than at fertilization. The macronucleus controls somatic activities and is not only very active, but it is often extensively restructured during somatic development and maturation. Transplantation of young or old macronuclei into short-lived hosts suggested that the macronucleus is capable of “remembering” its age (Aufderheide 1987), which suggests that the aging process in paramecia is accompanied by permanent functional changes in the nucleus. The macronucleus is destroyed during the next round of fertilization. Thus this little “animalcule,” to borrow Antoni van Leeuwenhoek’s term, has evolved an operational separation of germline and somatic functions in a manner analogous to that seen in multicellular organisms. Both structure and function are altered with increasing age in paramecia (see Smith-Sonneborn 1985, 1990). The life span of paramecia is under some form of genetic control because at least one mutant has been isolated that significantly reduces

111

the organism’s clonal life span (Takagi et al. 1989). Other experiments have revealed that the cytoplasm from aged parents is more likely to result in abnormal progeny and will eventually become incapable of supporting the existence of a normal nucleus. Occasionally, such aged cytoplasm may be rejuvenated by a young nucleus. It is more common, however, to find evidence of young cytoplasm rejuvenating aged micronuclei. As the clone ages, the micronucleus appears to be progressively affected by the increasingly toxic effects of the aging cytoplasm. There is some reason to believe that this increased nuclear damage reflects the age-dependent loss in DNA repair capacity. What is clear is that we once again find evidence of a complex nucleocytoplasmic interaction, the molecular basis for which is slowly emerging (see Smith-Sonneborn 1990). In addition, however, the study of paramecia first introduces us to the phenomenon of organelle damage. When a paramecium divides asexually, the anterior daughter cell retains the original (parental) oral apparatus, while the posterior cell receives the new oral apparatus. If the two newly formed daughter cells are separated from each other and their clonal life spans measured, the posterior cell line usually appears to be more viable than the anterior cell line. Both cells have received essentially identical nuclei and cytoplasm; thus it is unlikely that either of these are the important factors in this differential mortality. It has been suggested that the increased vigor of the posterior cell line is due to its ability to repair any preexisting accidental damage to the gullet when it replicated during cell division. This repair ability appears to reside in the micronucleus but can be transferred to the macronucleus in mutant strains that do not possess micronuclei (Smith-Sonneborn 1990). The anterior cell’s inability to repair such damage, since it inherited the original or parental gullet, would result in decreased viability. Thus, the longevity of paramecia is determined both by nucleocytoplasmic interactions and by organelle repair processes. As we’ll see, this strategy appears to be common and may operate in humans as well.

112 Chapter 4 Evolutionary and Comparative Aspects of Longevity and Senescence Tetrahymena. Tetrahymena does not normally exhibit an age-related increase in somatic line abnormalities or in germline abnormalities. It appears to be immortal. This does not mean that the organism does not make “mistakes.” On the contrary, even the immortal line produces defective (flawed and mortal) daughter lines, but at a constant rate. There is no increase in the frequency of defective daughter lines with increasing age of the parent cell. The lack of such age-dependent decrements in function is one sign that the cell is not senescing. Even within the defective lines observed in highly inbred strains, the defects can be eliminated via successive inbreeding and stringent selection to eventually produce viable progeny. There is no solid explanation for this immortality. Smith-Sonneborn (1985) has made the interesting suggestion that Tetrahymena, although diploid, has retained a haploid genome subunit organization. This organization would subject each of the haploid genome subunits to immediate and stringent selection. If so, then this might eliminate the accumulation of deleterious mutants responsible for senescence. It might also explain how Tetrahymena can repair organelles, such as its gullet, at any time during its life cycle. In this respect Tetrahymena stands in stark contrast to paramecia, which can repair organelles only at fertilization. The molecular structure of any ciliate macronucleus is not yet known; therefore, this suggestion has not yet been tested. However, certain otherwise anomalous genetic behaviors of Tetrahymena make sense if viewed in the light of this idea. The genome of Tetrahymena undergoes a controlled and complex process of molecular reorganization (Brunk 1986). Examining this reorganization may also lead us to insights applicable to other organisms and other systems. Tetrahymena is already known to contain odd but generally interesting molecules. Ribozymes, RNA molecules that can act like an enzyme (Zaug and Cech 1986), were first found in Tetrahymena, as was the hybrid RNA–protein enzyme telomerase—both of which are present and important in other organisms generally. Telomerase is now known to be an important enzyme in the cell’s longevity, and it is discussed in more detail in chapter 12. It will be interest-

ing to see if this ciliate yields any other interesting molecules that can help us better understand the molecular biology of senescence in this and/ or other organisms. 4.5.1.3 Yeasts

We are perhaps most appreciative of the important role that yeasts play in our life when we contemplate the fruits of their labors quietly aging in a wine rack. But yeasts are also useful for more mundane purposes, such as genetic investigations into the aging process. These unicellular eukaryotic fungi have been extensively investigated for many decades, in part because of their commercial importance. A large variety of mutants affecting various different basic cell processes have been collected over the years. These collections of genetic mutants have enabled yeasts to serve as excellent organisms for the genetic analysis of these processes. Because of this extensive genetic knowledge, the yeast Saccharomyces cerevisiae was selected as one of the genetic model organisms of the Human Genome Project. The complete genome of this yeast has been mapped, and the complete sequence of the genome was released in 1996. Many specialized mutants and strains are available. To use yeasts for studies on aging, it was first necessary to establish a method of measuring life span in these organisms. It was then necessary to show that their viability decreases in an agedependent manner. There are two different ways of measuring life span in yeast (Gershon and Gershon 2000). In the “budding life span” method, individual cells are grown on an agar substrate which both immobilizes them and provides a nutrient source (Mortimer and Johnson 1959). The immobilized mother cell periodically reproduces by means of forming a new and smaller daughter cell by mitosis, which eventually grows to the same size as the mother cell. The daughter cell is physiologically distinct from the mother cell but stays attached to it for some time before dropping off or being removed by the investigator. Using the number of times that a given yeast cell buds during its life span as a measure of aging has the advantage of allowing the life span to be

4.5 Comparative Aspects of Aging

ure 2.25) and yields a linear Gompertz plot (figure 4.8; Jazwinski et al. 1989; Kaerberlein et al. 2001; Kennedy et al. 1995; Pohley 1987). Because these two traits are the identifying characteristics of an aging population, the exercise demonstrates the fundamental utility of the population measurements discussed in chapter 2; in their absence, it might have been difficult or even impossible to agree about whether yeasts age. It was further found that the mean and maximum life span values are characteristic for any given strain of S. cerevisiae but that they vary widely from one strain to another. Such evidence suggests the existence of strain-specific genetic mechanisms regulating longevity. Using the stationary phase method, Longo and colleagues (1996) determined that the mean survival time of yeast was about 7–8 days and is characterized by a very steep drop off in viability, which can, however, be significantly extended by various genetic interventions. The obvious effect of gene mutants on longevity indicates the important role of genes in determining life span in yeast. I discuss this topic in more detail in chapter 7. Every method of culturing yeast and measuring aging has its strong and weak points (Gershon and Gershon 2000), but the fact that the replicative life span, rather than the chronological life

measured in physiological rather than in chronological terms; thus this method gives us the replicative life span. Removing time from the analysis of aging is a great advantage, as pointed out in chapter 1. Alternatively, one may use the “stationary phase” method (Longo et al. 1996), in which a population of yeast cells is maintained in liquid culture until the cell number reaches a plateau. Although all cells stop dividing due to the depletion of the glucose in the media, it is likely that the individual cells have each budded some variable number of times and are thus at different replicative ages. The quiescent cells can then be maintained for some time on either the expired media or on distilled water, and their viability can be determined by periodically measuring the ability of aliquots of the cell populaton to grow when removed from the liquid media and grown on agar plates. This method thus gives us the chronological life span, which is the total amount of time during which the quiescent cells can retain their ability to grow on a rich media. Using the budding life span method, several laboratories have independently demonstrated not only that a specific cell of S. cerevisiae has a finite life span, but that the survival curve of a synchronous cohort of such individual cells is rectangular (similar to that of curve 2 in fig-

(a)

(b)

50

0

1.000

Age-specific mortality (qx)

Percent live cells

100

0

10 30 20 Age (generations)

40

113

0.100

0.010

0.001

0

20 10 30 Age (generations)

40

Figure 4.8 Survival characteristics of Saccharomyces cerevisiae, strain X2180-1A. Life spans were determined for 43 individual cells. Virgin cells that had never budded were used to start the experiment. Every time the cell budded, the daughter was removed and the mother was scored as being one generation older. (a) Results presented in the form of a survival curve. (b) Age-specific mortality rate presented in a Gompertz plot. (After Jazwinski et al. 1989.)

114 Chapter 4 Evolutionary and Comparative Aspects of Longevity and Senescence span, is characteristic for a given yeast strain (Muller et al. 1980) is a strong argument in favor of the budding yeast method. A detailed list of the morphological and physiological changes that occur during yeast aging was presented by Jazwinski (1993, 2000b). Some of these changes can be used as biomarkers of aging. These useful indicators include (1) increase in cell size, (2) increase in the number of bud scars, (3) increase in the chitin component of the cell wall, (4) loss of cell polarity as evidenced by a random budding pattern, and (5) increase in the generation time. The increase in the generation time is the most interesting of these changes and is perhaps the single best diagnostic biomarker of aging in this organism. The generation time of individual yeast cells increases with replicative age, accelerates as the cells enter the phase of exponential increase in mortality, and becomes acute two to three generations before the cessation of mitosis by the mother cell (Jazwinski 1993). Long-lived strains display a shorter generation time and a delayed increase in generation time when compared to strains that have shorter life spans (Egilmez and Jazwinski 1989).

These data on yeast are encompassed in the cell spiral model of yeast aging (figure 4.9). Note that the budding cells undergo an asymmetric cell division such that the daughter cell is smaller than the mother cell. This can only come about via an asymmetric placement of the mitotic spindle, an event under genetic control. The asymmetry is not limited to cell organelles but includes damaged proteins as well. Carbonylated proteins are the irreversible damage products of oxidative damage, and the cell will eventually tag and destroy them. Until that time, the dividing yeast mother cell retains the damaged proteins during mitosis and does not pass them on to the budding daughter cell (Aguilaniu et al. 2003). Cells lacking a functional SIR2 gene fail to retain the damaged proteins and distribute them evenly to both mother and daughter cell. Such cells also showed an abnormal distribution of actin, suggesting the involvement of the actin cytoskeleton in this asymmetric division. It is possible that the extra load of oxidatively damaged proteins in the young daughter cell could have some adverse effect on its viability and longevity, especially under stress conditions. The SIR2 gene is an important lon-

Generation (cell cycle) Virgin cell 1st

Figure 4.9 The cell spiral model of yeast aging, showing the relationship between successive cell division cycles and the aging process. A virgin cell enters the spiral, grows, and buds, producing a daughter cell. The mother cell continues in the spiral, proceeding through successive cell cycles. The number of generations is limited, and the mother cell ultimately dies. Daughter cells enter the spiral at the top. (After Jazwinski et al. 1989.)

Daughter 1 2nd

Daughter 2

Daughter 3

Daughter n

Dead cell

3rd

nth

Aging

4.5 Comparative Aspects of Aging

gevity determinant gene in yeast, and I discuss it in greater length in chapter 7. There are other examples of the involvement of asymmetric cell divisions in aging (e.g., Volvox, stem cells, etc.). An inspection of figure 4.8 suggests that the yeast life span may be divided into a reproductive phase and a postreproductive phase. Only the reproductive phase can currently be measured in physiological terms; the postreproductive phase still is measured in terms of time. The reproductive phase is usually strain specific, suggesting a strong genetic component. In fact, genetic analysis has yielded at least 17 genes that play a role in determining the replicative life span of this organism. The stationary phase is strongly influenced by the environmental conditions and is thus quite variable in length, although a number of genes have been found that significantly alter the cell’s ability to survive different environmental stresses. Genes affecting three broad physiological processes that determine yeast life span have been identified: metabolic control, stress resistance, and genetic instability. These topics are summarized here and discussed in more depth in chapter 7. Metabolic control refers both to the effect of nutrition on the cell’s metabolism and longevity as well as to an intracellular signaling system that connects the mitochondrion to the nucleus. In the first case, it involves the processes underlying the longevity-extending effects of caloric restriction (see chapter 6). In the second case, it involves the mitochondrion, which is the site of oxidative respiration and the source of almost all the energy we use to power our bodies. The cell’s various organelles need to be kept in synchrony with one another, in part because the metabolic states of the organelles must be consistent with each other and in part because most of the organelles’ molecules are coded for by the nucleus. When the signaling system is active, a mitochondrial signal activates a transcription factor in the cytoplasm, which then moves into the nucleus and there activates a number of genes encoding various metabolic enzymes, which are then translocated into the cytoplasm, the mitochondria, or the peroxisome. The net result of this signaling is to bring about changes in the metabolic state of the cell,

115

and this is believed to shift the organism into a low-damage pattern associated with extended longevity in other organisms. Metabolic control involves interorganelle communication and integration of their responses with the external environment. Stress resistance refers to the cell’s ability to resist various stressors, the most important of which is oxidative stress (see chapter 10). The cell has a large number of genes whose protein products can capture and/or inactivate the reactive oxygen species of molecules that cause oxidative stress in the organism. Other gene products help the cell defend itself against stressors such as heat shock. Some of these various stressresistance genes are localized in chromosome regions that are initially activated by the histone deacyltase enzymes mentioned below. Stress resistance is often correlated with alterations in metabolic control. Genetic instability may occur in one of two general ways. First, chromatin-dependent gene regulation refers to the mechanisms that distinguish transcriptionally active chromosome regions from transcriptionally inactive (or silenced) regions. The inactive regions are silenced by being wrapped quite tightly around the nucleosomes (histone proteins), which are an integral part of the chromatin. The DNA in such a wrapped region is not accessible to the polymerases and other molecules necessary to activate and transcribe it. What determines whether the DNA will wrap tightly or not about the nucleosome is whether or not the histone proteins have an attached acetyl (or other) side group. Presence of the acetyl group allows the DNA to be loosely wrapped and thus make it available for transcription. There are a number of genes whose protein products function as histone deacetylases and thus play an important role in altering the state of the chromatin, permitting the transcription of genes in that region (Guarente 1999). The second process, gene dysregulation, refers to the inappropriate activation of chromosomal regions by means other than those listed above, giving rise to the dysfunctional expression of various genes. This may come about via mechanisms that include but are not limited to the chromatin-dependent mechanisms

116 Chapter 4 Evolutionary and Comparative Aspects of Longevity and Senescence discussed above. Thus, mutational damage and alterations in signal transduction processes may also be involved. Gene dysregulation might also occur as a result of transcriptional or conformational changes in other gene products that interact with the gene in question. However it comes about, the inappropriate or heterogeneous collection of gene products functions in the cell much like an out-of-tune choir: much noise but little melody. Gene dysregulation should be reflected in a shift of the gene expression pattern from a state of high functioning to a (noisy) state of lower functioning. As pointed out in chapter 1, definitions and measurements that may be correct and valuable at one organizational level may be meaningless at another level of complexity. Yeast and other unicellular organisms are a good example of this difficulty: Organismal age must mean one thing for a yeast cell and quite another for an individual human because cell age and organismal age are coincident for the yeast but not for the human. Given that conceptual chasm, how does one translate findings obtained in these organisms to more complex metazoans such as mammals without simultaneously transferring erroneous concepts? In a thoughtful review, Gershon and Gershon (2000) identified and described a variety of conceptual and technical problems in translating unicellular data to higher metazoans. It is clear that yeast show the same mortality kinetics that we have established as diagnostic for the presence of aging (figure 4.9). The fact that orthologues of human genes are found in yeast (Steinmetz et al. 2002) and the complementary fact that some human genes can function in yeast and some yeast genes can function in multicellular organisms such as the nematode C. elegans strongly suggest that valuable information about aging mechanisms at the cellular level can be obtained from the study of yeast aging. If the appropriate precautions are taken, then the ease of working with yeast should allow us to identify both species-specific as well as highly conserved aging mechanisms. Knowledge of the latter might guide us to identify or better understand their homologues in humans.

4.5.2 Volvox and Other Simple Multicellular Organisms The evolution of multicellular organisms was a major advance that made possible the division of labor and the differentiation of cells for particular tasks. There are several evolutionary paths by which such multicellular forms are thought to have arisen. One of them appears to have involved the coalescence, over evolutionary time, of individual cells into a colonial organism, with the subsequent specialization of the cells into somatic and reproductive cells. This process can be seen best in the primitive green alga Volvox carteri, which consists of a hollow sphere of about 2000 small, flagellated somatic cells (each of which resembles an existing, more primitive, freeliving unicellular [protist] species of alga). This sphere surrounds and encloses about 16 larger reproductive cells. The somatic cells can move but cannot divide; the reproductive cells can divide but cannot move. They are each specialized for different tasks. The reproductive cells divide while inside the mature adult and give rise to juveniles, which are then released from the parents. The juveniles mature and give rise to another generation of adults, while the somatic cells of the old parent die (Kirk 1988). The somatic cells of the parent senesce and die shortly after reproduction is complete (Hagen and Kochert 1980). The terminally differentiated somatic cells show a characteristic pattern of morphological and biochemical changes (such as a loss of chlorophyll or a decrease in protein synthesis) that decreases their probability of survival and can thus be called senescent changes. These changes are inherent in the somatic cell itself; early surgical removal of the reproductive cells does not rescue the somatic cells from their fate. They can be rescued from death only by certain mutants that convert the somatic cells into reproductive cells, but then there is no protection for the mass of reproductive cells. A somatic cell undergoes a series of specializations and differentiations that allow it to perform a highly specialized function but that inevitably result in the death of the cell. These observations lead us to

4.5 Comparative Aspects of Aging

the conclusion that aging in this species is an intrinsic part of a program of gene-controlled development. Whether an embryonic cell will differentiate as a somatic cell or a reproductive cell appears to depend on the differential expression of at least three genes, as shown in figure 4.10. During development, genetically controlled asymmetric mitosis gives rise to daughter cells of unequal size, and this initial difference leads to different patterns of gene expression. This, in turn, leads to the functional differentiation of the two cell types; the smaller cells becoming somatic cells and the larger cells giving rise to reproductive cells (Kirk 2001). The glsA protein is bound to the mitotic spindle in dividing cells and has substantial sequence similarity to a human mitotic protein (S. M. Miller and Kirk 1999). Another chaperone-type protein (hsp70) colocalizes with

117

the glsA protein to the spindle and, with the probable assistance of other unidentified proteins, shifts the mitotic spindle from the center to one side of the cell. This action somehow sets up a differential expression of the regA gene, which has all the properties of an active transcriptional repressor so that it is expressed in the small cells but not in the large cells (Kirk et al. 1999; figure 4.10). The regA gene acts either directly or indirectly to shut down a number of essential chloroplast genes in the small cells. These plant cells depend on photosynthesis to provide energy for growth. Without active chloroplasts, these cells cannot grow. They have enough existing chloroplasts to provide sufficient energy for cellular maintenance but not enough for growth nor for long-term cell maintenance. In addition, genes involved in reproductive processes are repressed. As a result, these small cells become terminally

Mature Gonidium 32 cell embryo

gls ON

Large Cells

Small Cells

regA ON

lag OFF

regA OFF

lag ON

Repress Gonidial Genes Transcribe Somatic Genes

Transcribe Gonidial Genes Repress Somatic Genes

Somatic Cell Differentiation

Gonidia Formation, Rapid Growth

Cell Death

Reproduction

The Volvox gene cascade for aging

Figure 4.10 Diagram of a working hypothesis regarding the basic genetic program for germ-soma differentiation in Volvox. At the 32-cell stage, the products of the gls genes act to promote the asymmetric divisions that produce large and small cells. Then the regA gene product acts to repress gonidial genes in the small cells, while the lag gene products act in the large cells to repress somatic genes. (Redrawn after Kirk 2001.)

118 Chapter 4 Evolutionary and Comparative Aspects of Longevity and Senescence differentiated, nondividing somatic cells with a limited source of energy. They eventually senesce and die. The reverse situation occurs in the reproductive cell such that the regA gene is repressed and the lag gene is expressed. If one expresses the lag gene in early embryos before the asymmetric cell division, then no somatic cells are formed, and all the cells are potentially immortal reproductives (Kirk et al. 1999). Sensecence disappears. The functions of the lag gene are not yet known in detail, but they probably repress somatic cell genes from being expressed in the reproductive cells. Note that the above mechanism forces the embryonic Volvox cell to make a genetically based binary choice: it can opt to grow or to differentiate, but it cannot do both simultaneously. In effect, this amounts to an allocation of energy by the organism such that minimal energy is expended on the protective somatic cells and much energy is expended on the reproductive cells. In this context, the genetically controlled fates of the Volvox embryonic cells are consistent with the disposable soma theory. Aging, senescence, and death seem to be built into the multicellular way of life even from its very beginnings.

4.5.3 Fungi: The Role of Mitochondria in Senescence Many more species of fungi exist than those few we usually see in supermarkets. The two species most important to our survey of aging are Neurospora crassa and Podospora anserina, which are closely related ascomycetes. The ascomycetes constitute a very common group of fungi, comprising some 30,000 free-living species and including the yeasts, many of the common black and bluegreen molds, and the morels and truffles prized by gourmets. Neurospora and Podospora are wellestablished laboratory tools for investigating biochemical genetics, and their analysis has allowed us to discern the roles that mitochondrial DNA (mtDNA) and antioxidant defense systems play in the extension of longevity in these species. I discuss antioxidant defense systems in more detail in chapter 6; mitochondrial DNA is what concerns us here.

Neurospora and Podospora have provided us with an interesting insight into an aging process that might have been overlooked if we had confined our attention to the traditional laboratory rodents (see Griffiths 1992 for review). The body plan of fungi is fundamentally different from that of most other organisms. Fungi usually do not have discrete cells; rather, they contain mitotically derived nuclei within a common cytoplasm. They usually reproduce vegetatively via mycelial or hyphal growth, and they can be widely dispersed via asexual cellular spores. Life span in these organisms can thus be measured either as the ability of the colony to continue growing or as the ability of the spores to survive for some time before germinating. The growth of Podospora ceases at a time and a size characteristic of each geographic race. The nongrowing portions of a clone fuse, undergo meiosis, reproduce sexually, and produce ascospores. This process restores the growth potential of the clone. The old hyphae (somatic tissues) eventually die. Senescence is not transmitted to ascospores. Thus life span in this organism refers to the continued mitotic growth and vitality of the somatic cells constituting one clone or individual. Genetic crosses have shown that the characteristic life span of any given race is inherited as a cytoplasmic factor localized to the mitochondria, cellular organelles that are essential to energy metabolism in all eukaryotic cells. All mitochondria contain their own DNA, but for proper functioning they rely on their own genome, as well as that of the nucleus. Cellular senescence in this fungal genotype is associated with the alteration of its mitochondrial genome into covalently closed, circular, head-to-tail multimers of a portion of the mtDNA. These elements are known as senescent DNA, or senDNA. Different strains and even different cultures of the same strain may show different types of senDNA (Griffiths 1992). Regardless of their precise origin within the mitochondrial genome, these different senDNAs appear to act in much the same way. As senDNAs accumulate in aging cultures, mtDNA molecules change radically, in that complete wild-type molecules disappear. This transition is associated with a reduction of specific cytochromes, resulting in

4.5 Comparative Aspects of Aging

a disruption of energy production in the mitochondrion (Kuck et al. 1985). These senDNAs appear to be discrete mtDNA sequences that can autonomously excise and amplify themselves such that they may be found in high concentrations in the mitochondria of senescent cells. The concentration of these senDNA molecules in the young hyphal cell appears to be positively correlated with the race-specific life span: very low in long-lived strains and higher in shortlived strains (Wright and Cummings 1983). However, certain nuclear mutations confer immortality on their host. The genetically based suppression of the formation of senDNA suggests that senescence may depend on a certain sort of interaction between the nuclear and mitochondrial genomes, and a model to this effect has been proposed (Esser 1985; Kuck et al. 1985). More recent work has shown that genetically controlled alterations of the mitochondrial respiratory complex are necessary if Podospora is to express a long life (Krause et al. 2004; Sellen et al. 2004). This work demonstrating the existence of an interaction between the nucleus and the mitochondria foreshadowed the demonstration of these interactions in other laboratory models, as described in chapter 7. Neurospora shows a senescence pattern similar to that of Podospora both in its general progression and in its involvement of mtDNA (see Griffiths 1992 for full review). Extragenomic elements are progressively inserted into the mtDNA as the culture ages, causing cytochrome abnormalities and loss of growth potential, until at death intact mtDNA is barely detectable. The source of the extragenomic DNA appears to be one of several different strain-specific mitochondrial plasmids. Removal of the plasmids from a strain removes the senescent phenotype. Griffiths (1992) has suggested that the involvement of these plasmids in the senescent phenotype is a mistake, in that both the mitochondrial and plasmid genomes are probably sharing the same molecular machinery to carry out their own replication, and so occasionally the two systems become entangled with one another, thereby setting off senescence. This phenomenon may appear to be an odd evolutionary mechanism of senescence restricted to a few species of fungi and of no real consequence

119

to us. Yet we must recognize that evidence suggests that bioenergetic decline and the accumulation of mtDNA damage are associated with the degenerative effects of aging and age-related diseases in humans and other organisms (Wallace et al. 1995). Such damage is particularly noticeable in postmitotic tissues and may be related to oxidative damage (Martin et al. 1995). I explore this topic further in chapters 7 and 10. Whether such a mechanism plays a causal role in organismal aging is unknown, but it is now under active investigation. We cannot afford to overlook this clue revealed by the comparative approach.

4.5.4 Invertebrates: General Patterns of Aging The varieties of senescence in invertebrates are exceeded only by the number of their species. Comfort (1979) presented data on the maximum life spans of 282 invertebrate species; as table 4.3 shows, the numbers span the gamut from 28 days to 90 years. In addition, Comfort (1979) presented descriptions of senescence patterns for about 20 different supraspecific groups. It is impossible to summarize the patterns of aging and senescence in such a diverse group of organisms. It is far more useful to describe in some depth the patterns of aging and senescence observed in a few experimentally important species. Provocative insights may be achieved perhaps more easily from a detailed but limited review than from an inclusive but superficial commentary. The following discussion of invertebrate aging is limited to rotifers, nematodes, and insects because enough intensive work has been done on each of these groups to merit our study of the data. 4.5.4.1 Rotifers: The Power of Theory

The 2000 species of the phylum Rotifera are very small aquatic pseudocoelomates. They are sometimes called “wheel animalcules” because the beating of a crown of cilia around the mouth causes them to spin like tiny wheels. These small animals have a muscular pharynx with hard jaws; they are omnivorous. Most species reproduce

120 Chapter 4 Evolutionary and Comparative Aspects of Longevity and Senescence Table 4.3 Maximum Recorded Longevities for Some Invertebrates Phylum and class or order Porifera Demospnogiae Coelenterata Anthozoa Platyhelminthes Cestoda Turbellaria Ascheiminthes Nematoda Rotifera Annelida Polychaeta Oligochaeta Anthropoda Arachnida Crustacea Subclass Cirripedia Subclass Malacostraca Insecta Ephemeroptera Diptera Isoptera Lepidoptera Coleoptera Hymenoptera Echinodermata Echinoidea Asteroida Mollusca Amphineura Gastropoda Subclass Prosobranchia Subclass Opisthobranchia Subclass Pulmonata Bivalvia Cephalopoda

Species

Maximum life span

Suberites carnosus

Source of dataa

15 years

c

85–90 years

c

Taeniarhynchus saginatus Dugesia tigrina

>35 years 6–7 years

h c

Wucheria bancrofti Callidina sp.

17 years 5 months

h c

Sabella pavonina Lumbricus terrestris

>10 years 5–6 years

c c

11 years

c

Cereus pedunculatus

Filistata insidiatrix Balanus balanoides Astacus

>5 years 15–25 years

w ?

4 weeks 9 weeks >25 years 44 days >10 years >19 years

c c w c c c

Echinus esculentus Marthasterias glacialis

>8 years >7 years

w c

Chiton tuberculatus

12 years

w

Cloëon dipterum Drosophila melanogaster Neotermes castaneus Maniola jurtina Blaps gigas Lasius niger

Patella vulgata Haminea hydatis Rumina decollata Margaritana margaritifera Loligo pealii

15 4 12 70–80 3–4

years years years years years

wg w c wg w

Source: based on data from Comfort (1964) and Lamb (1977). ac

= kept in captivity; w = in wild conditions; g = age estimate based on growth; h = host case history.

parthenogenetically, but many also reproduce sexually. The maximum life span is very specific, ranging from 12 days to 2 months. The nuclear number is fixed in rotifers; there is no cell division, and therefore they grow only via an increase in cell size. After a period of adult vigor, the aging animal enters a period of senescence, in which it becomes sluggish in its behavior, the tissues shrink and become opaque, pigment is de-

posited in the gut and associated organs, cells degenerate, and death ensues. (Note the similarity to the age-related changes observed in other invertebrates, such as Drosophila; see table 4.3.) In many species, this senescent process begins while the female is still reproductively active. The conjunction of these two processes often results in a diminished rate of egg laying coupled with the production of apparently abnormal eggs.

4.5 Comparative Aspects of Aging

The correlation of abnormal reproduction with the onset of senescence is by no means limited to rotifers. There is a considerable amount of individual variation in the length of the senescent period. An endogenous process of senescent degeneration appears to affect all parts of the animal’s body at more or less the same time. There doesn’t appear to be a pacemaker organ. As Comfort (1979) has pointed out, this pattern is consistent with the idea that some or all of the somatic cells have a fixed survival time. This constitutes one of the theoretical attractions of rotifers to the experimental gerontologist, for their cells undergo a synchronous death. The other main attraction of rotifers lies in their alleged possession of deleterious parental age effects, and a review of this aspect of scientific history may teach us to be wary of being blinded by theory. Lansing (1947, 1954) worked with two parthenogenetic species that reproduced via ameiotic divisions of the female germline. What resulted were clones of genetically identical offspring with which Lansing could do his experiments. He found that eggs laid by old mothers had shorter life spans than did eggs laid by young mothers. Furthermore, if the strain was continuously propagated from older mothers for several generations, then the mean life span would decrease progressively, and the strain would invariably become extinct. Given the genetic identity of the animals, these results suggested that death of Lansing’s clonal lines was due to the existence of an extragenic, transmissible, and cumulative age factor. This hypothetical cytoplasmic factor, passed from generation to generation via the egg, was thought to influence longevity by accelerating the onset of senescence in later generations. Such “Lansing effects” have since been claimed to occur in many other organisms (see Lints 1978) and have been viewed by many as a general aspect of aging. From studies on experimental animals, offspring of older parents are known to be quantitatively different from offspring of younger parents. But until Lansing’s work, there was no reason to suggest that these effects were cumulatively transmissible over multiple generations.

121

However, there may be a problem of overinterpretation here. First, this accelerated aging of subsequent generations is not found in other cases in which cytoplasmic factors are believed to be involved, such as Paramecium or Podospora. Second, Lints (1978) pointed out what many readers of Lansing’s papers overlooked—namely, that all of the cloned lines eventually died out, regardless of the age of reproduction. Third, Lints also showed that both young and old mothers gave rise to offspring with shortened life spans. There appears to be a problem with oogenesis both in young and in old mothers, but not in middle-aged mothers. In other words, there is an optimal age for reproduction. Reproducing at suboptimal ages would allow the population to die out simply because of the reduced fecundity, not because of any effects of aging as such. Even though the major experimental basis for the Lansing effects has been weakened by this analysis, the Lansing effects have been claimed to exist in other organisms. Lansing effects were assumed to involve age-dependent alterations in the cytoplasmic factors that the mother contributes to the egg. Further, these factors are presumed to alter the patterns of gene expression in succeeding generations. The mechanisms involved are obscure, and there is some disagreement over the interpretation and the importance of the data. The topic was reviewed by Lints (1988). It is possible that a maternal age effect is operating in the rotifers. An even more probable explanation is that the existence in the literature of descriptions of the Lansing effects may have persuaded other gerontologists to conclude that artificial selection experiments for long-lived organisms were bound to fail because one would have to breed continually from old mothers. Because the experiments were bound to fail, then what was the point of doing them? It is the theory which decides what we can observe. If the theory is in error, what does that do to our observations and assumptions? Consider that none of the vast amount of material summarized in chapter 7 would have been obtained had everyone listened to the naysayers, data or no data. The warning is pertinent whether working with rotifers or with humans.

122 Chapter 4 Evolutionary and Comparative Aspects of Longevity and Senescence

4.5.4.2 Nematodes: Genes and Altered Proteins

Nematodes are cylindrical, unsegmented pseudocoelomate worms, most of which are free-living microscopic forms. There are 12,000 described species; however, given their small size and inconspicuous habitats in the soil and elsewhere, it is not surprising to learn that some authorities believe that as many as 500,000 nematode species exist. Humans are hosts to about 50 parasitic nematodes, which cause diseases such as pinworm, hookworm, intestinal roundworm, and filariasis. These few diseases probably account for the bad reputation of nematodes as a group. Several different species of free-living nematodes have been used in research on aging. In 1974, Sidney Brenner published a persuasive paper describing the genetic system of Caenorhabditis elegans and convinced many of this nematode’s experimental virtues (Hodgkin 1989). The Nobel Prize for Physiology and Medicine for 2002 was awarded to Brenner and two of his students (John Horvitz and John Sulston) for their pioneering work in making this organism a valuable laboratory animal. Most of the work being done today focuses on this species and, to a lesser extent, on Turbatrix aceti. C. elegans is easily cultured in dishes containing agar as a semisolid supporting medium and a layer of bacteria as a food source. Under these conditions, C. elegans lives to a maximum of 20 days. It is particularly suitable to genetic studies, and the heritability of the life span of this species is estimated at 20–50% (T. E. Johnson and Wood 1982). Several different mutants that confer long life have been isolated and described; several of them are intimately involved with regulating the aging process. The available data suggest that these mutants lengthen the life span by increasing the animal’s resistance to oxidative stress. A more detailed discussion of their genetic control of aging appears in chapter 7. Senescence in nematodes is signaled by a progressive increase in mortality, a failure to respond to environmental stimuli, the accumulation of lipofuscin (the so-called aging pigment), and deterioration of the internal anatomy. Senescence is also accompanied by a striking biochemical

phenomenon in which structurally altered enzyme molecules accumulate in older animals. This phenomenon was first observed in 1970 by Gershon and Gershon in Turbatrix aceti. They found that a purified enzyme, isocitrate lyase, obtained from old animals had only 60% of the specific activity found in the pure enzyme obtained from young animals. The implication is that the enzymes from the older animals were altered in some way. This alteration might be due to inefficient protein synthesis in old animals or to some subtle physical changes taking place in old molecules. It is probably some sort of wearand-tear phenomenon that affects old molecules because it has been shown that these old molecules can be made as good as new simply by chemically unfolding them and allowing them to spontaneously refold in the test tube (Yuh and Gafni 1987). Even though the processes that cause this structural change are still not clear, this initial observation has sparked an interest in determining whether other enzymes in the nematode are altered in the same way with aging and whether other species display such changes as they age. Perhaps the deterioration in our own physiological functions arises from altered, and hence less efficient, enzyme molecules. Five enzymes in T. aceti are known to be altered with age (table 4.4). All of them demonstrate substantial differences (about 50%) in the age-dependent specific activities of these enzymes. This is a real process. However, the other fact not stated but implicit in table 4.4 is that very few enzymes (only 10 out of the several 1000 present) demonstrate this agespecific alteration. Thus, either this process has little to do with aging, or the enzymes involved are rare limiting enzymes that regulate the activity of important metabolic pathways dependent on them. It will be an important task of future research to determine whether the facts support this alternative interpretation. Whatever the nature of the mechanism involved, this phenomenon is not limited to nematodes but also appears in various different enzymes in both rats and mice (table 4.5). Some of these structurally altered vertebrate enzymes, such as superoxide dismutase, are believed to be inti-

4.5 Comparative Aspects of Aging

123

Table 4.4 Properties of Altered Enzymes from T. aceti

Isocitrate lyase Enolase Fructose-1,6-diphosphate aldolase Phosphoglycerate kinase Elongation factor 1

Altered specific activity

Altered Km

Electrophoresis change

Antigenic difference

Immunological cross-reaction

Altered heat stability

Yes Yes

No Yes?

— No

No Yes

Yes Yes

Yes Yes

Yes Yes Yes

No? Yes? —

No No —

No — —

Yes — Yes

Yes No —

Source: after Russell and Jacobson (1985). Note: Change in relative quantities of isozymes. All isozymes have altered specific activity.

Table 4.5 Distribution of Altered Enzymes in Several Species Enzymes Isocitrate lyase Phosphoglycerate kinase Enolase Elongation factor 1 Aldolase Tyrosine aminotransferase Ornithine decarboxylase Phosphorylase Glucose-6-phosphate dehydrogenase Superoxide dismutase Lactic dehydrogenase

T. aceti X X X X X

Mouse

Rat X

X X X X

X

X X X

Source: after Rothstein (1982).

mately involved in the protective aspects of the aging process; others, such as ornithine decarboxylase or glucose-6-phosphate dehydrogenase, could be regulatory enzymes. The existence of thermodynamically nonrandom change that would alter the specific activity of certain key enzymes and hence initiate senescence is an interesting concept that should be examined further. We now know that the transcription of an important stress protein, hsp70, shows an agerelated decline due to structurally altered transcription factors. Thus, the concept needs to be broadened beyond enzymes to include all regulatory proteins. The existence of such a process would not have been found in the absence of a comparative approach. It will also be important for future research to determine if these structural

alterations are just the indirect consequences of an age-related decrease in protein turnover. An increase in the half-life of the protein molecules could result in nonspecific tertiary structural alterations simply as a result of increased exposure to environmental insult. One of the most important aspects of the nematode for future work on aging is that the complete cell lineage of C. elegans is known and described. The organism has a determinate and strictly programmed pattern of cell division and cell differentiation that leads to the development of an adult organism. The origin, position, and fate of every cell are known and charted. Another important advantage of this organism is the excellent assortment of mutant genes, mutant strains, and genetic tools that are available for use by geneticists. In addition, its genome has been sequenced. Current research has identified two separate but interacting genetic systems controlling longevity in this species: one operating via regulation of stressresistance mechanisms and the other via regulation of metabolic timing. I discuss these findings in more detail in chapter 7. C. elegans therefore offers an excellent opportunity to investigate which aspects of the aging process are genetically determined and which are stochastic.

4.5.4.3 Insects: Drosophila as a Genetic Model

Perhaps a million different species of insects are known to exist today, and more new ones are being named and described each year. In fact, it

124 Chapter 4 Evolutionary and Comparative Aspects of Longevity and Senescence is likely that about 75% of all the animal species on earth are insects. They are the dominant terrestrial life forms on the planet, in terms of both number of species and number of individuals. As arthropods, all insects have an external skeleton, and their segmented body has three main divisions: head, thorax, and abdomen. The life history of insects is fundamentally different from that of the other invertebrates I have described. Immature forms of most insects have limited mobility and usually pass through their developmental stages close to the location where the adult female laid the egg. Young insects are voracious feeders. Their growth involves changes not only in size but also in form. In the holometabolous insects (about 90% of the extant species), this transition from immature to adult form involves a complete metamorphosis that is under hormonal and genetic control. This remodeling of the organism occurs within the pupal stage and results in the emergence of a sexually mature adult. The success and diversity of insects is reflected in the diversity of their life spans, ranging from 1 day for

mayflies to more than 60 years for termite reproductives. Despite this rich diversity provided by natural selection, only a few insect species have been studied in relation to aging. These include several species of the fruit fly (Drosophila), the common housefly (Musca domestica), a mosquito (Aedes stimulans), a wasp (Habrobracon serinopae), and a flour beetle (Tribolium castaneum). By far the overwhelming amount of work on aging in insects has been performed on Drosophila melanogaster, the geneticists’ favorite organism. Studies on aging are carried out on the adult form because the larval tissues show no signs of aging. Indeed, as pointed out in chapter 6, the presumptive adult or imaginal disc tissue found in the developing larvae shows no signs of senescence but can be kept alive indefinitely and still give rise to normal adult tissue. One interpretation of these findings is that only organisms that are no longer in their developmental phase age. Several obvious manifestations of senescence are detectable in aging insects; they are listed in table 4.6. Extensive reviews of the aging process

Table 4.6 Age-associated Alterations in Drosophila External morphology Increased damage to external structures Increased melanin pigment in sternites Behavioral changes Decreased geotactic and phototactic response Decreased ability to withstand environmental stress (e.g., starvation, insecticides, high temperature, low humidity) Decreased mating ability Decreased stamina, speed of locomotion, and flight performance Decreased chemoreceptor sensitivity Physiological, cellular, and biochemical changes Shrinkage of cortical area of brain, loss of basophila, and vacuolation of neurophil Increase in lipofuscin content of brain, heart muscle, gut, fat body, and Malphigian tubules Cell structure changes within alimentary canal Cell structure changes/possible myofibrillar degeneration in “heart” Loss of glycogen granules and possible myofibril degeneration in flight muscle Mitochondrial enlargement and loss of cristae Decrease in functional capacity of the mitochondria Decreased amount of rough endoplasmic reticulum in Malphigian tubule cells Decreased fecundity Decreased no. of functional ovarioles in females Decreased no. of spermatogonia/spermatocytes in males Various changes in steady-state activity of many enzymes Decline in protein synthesis ability Source: from Arking and Dudas (1989).

4.5 Comparative Aspects of Aging

in Drosophila and other dipterans have been conducted by Lamb (1978), Sohal (1983), Arking and Dudas (1989), and Rose (1996). These ageassociated alterations are comparable to the changes that take place in the human body (see chapter 5). This fundamental biological similarity is what allows us to (cautiously) extrapolate from one species to another. An almost universal indication of aging in insects is decline in stamina, speed of locomotion, and other behavioral parameters. Another common indicator is the decreased ability of the organism to resist stress, such as high temperature or insecticides. In fact, the investigation of different long-lived strains and mutants of Drosophila has led to the idea that resistance to stress is of particular importance. The beaten and worn appearance of wings and body parts is usually but not always a reliable indicator of aging. With the exception of the reproductive cells, the adult insect is postmitotic, yet cell death does not appear to be a major factor in insect aging, although insects do exhibit structural signs of degeneration and impaired function. The adult organism dies while its component cells are still alive. Age-associated structural changes in the nervous system do not involve the loss of neurons as much as they involve a reduction in cytoplasmic volume, an accumulation of lipofuscin, and other degenerative changes. The agedependent structural changes observed in the other tissues of the body all contribute to the physiological decrements associated with aging. The progressive accumulation of markers of damage, such as lipofuscin, carbonyl proteins, or peroxidized lipids, is probably the most consistent manifestation of aging in the postmitotic cells of the adult insect. There does appear to be a definite and marked decline in the ability of the aging insect to repair such damage. The overwhelming advantage of using Drosophila is that one may draw freely on the multitude of mutant strains, genetic and molecular procedures, and the very large body of knowledge accumulated during the past 80 years or so. But this advantage is not as great as it initially seemed. Arking and Dudas (1989) have pointed out that not every mutant is well suited for studies on

125

aging but that one must develop suitable strains. The unsuitability of ordinary strains was demonstrated some years ago by Ganetzky and Flanagan (1978), who compared numerous characteristics of two standard laboratory strains and found so many differences that it was not possible to determine which ones were potentially important. Another difficulty of using ordinary strains is that they are usually highly inbred. Crossing two such strains yields a hybrid that is usually more vigorous and longer-lived than either parent (Clarke and Maynard Smith 1955). These phenomena, known by the terms “inbreeding depression” and “hybrid vigor,” respectively, provide difficult technical obstacles to deciphering the genetic and physiological mechanisms involved. A way out of these dilemmas was shown some 20 years ago by several laboratories, which independently used long-term selection for longevity on moderately large outbred populations and thus developed stocks that did not exhibit inbreeding depression (Arking 1987b; Luckinbill et al. 1984; Partridge and Fowler 1992; Rose 1984). These are the suitable stocks alluded to earlier, and they will provide most of the data for this discussion. Additional information can also be cautiously obtained from genetically transformed stocks in which the techniques of genetic engineering have been used to introduce extra genes into otherwise normal flies. The caution arises from the many technical uncertainties inherent both in the technique itself and in the potentially confusing effects of introducing extra genes into an integrated response network. The most important findings of the genetic studies in Drosophila thus far obtained are that (1) the life span can be genetically shortened or genetically prolonged; (2) the physiological differences between the long-lived and normal animals are striking but are confined to a few key traits; (3) the strains developed in each laboratory appear to use different genetic and physiological mechanisms to bring about the expression of a similar phenotype; and (4) the genes involved in these proximate mechanisms have been or are now being identified. These findings are described in chapter 7, and their theoretical implications are discussed in chapters 9, 14 and 15.

126 Chapter 4 Evolutionary and Comparative Aspects of Longevity and Senescence Evolutionary theories of life history are based on the assumption that evolution acts so as to maximize the average fitness of organisms. What this means is that the organism should allocate its energy resources to somatic maintenance processes and to reproductive processes so as to maximize its Darwinian fitness, as predicted by the disposable soma theory (see figures 4.4 and 4.5). This supposition is now being tested in Drosophila because the only experimental study in which all of the necessary data (e.g.; longevity, fecundity, age-related energy production) exist is that collected for the Wayne State Drosophila strains as part of an artificial selection experiment for shortened and extended longevity (Arking, 1987b; Arking et al. 2000a,b; Luckinbill et al. 1984). It has been shown that the Wayne State control strain is at the predicted optimal fitness point in the normal environment ((Novoseltsev et al. 2002), and that the Wayne State long-lived strain is at a different optimal fitness point in its new environment. Thus, this basic tenet of life-history theory has been confirmed by use of Drosophila data. 4.5.4.4 Vertebrates: General Patterns of Aging

The vertebrates are a large group of animals (41,700 species) and is probably the most familiar to us, as it includes humans and our close mammalian relatives. All vertebrates are characterized by a dorsal vertebral column, or backbone, as their main structural axis. The body is originally segmented, but the obvious evidence of this segmentation is lost in development, save for the muscle segments associated with the vertebral column. There are seven living groups of vertebrates: the jawless fish such as lampreys, the sharks and rays, the true fish, the amphibians, the reptiles, the birds, and the mammals. We are most familiar with mammals because we keep so many of them as pets; thus we tend to assume that all vertebrates show age-related changes and longevity patterns similar to those of mammals. This assumption may be true, but it is best to review the relevant data. Table 4.7 gives the maximum life spans recorded for various vertebrates. More extensive lists

are presented in Comfort (1979) and in table 2.6. A study of these data permits the following observations. First, even though we lament the shortness of our days, humans are one of the longest lived species on the planet. Second, large animals tend to live longer than smaller ones. Third, on the basis of survival studies and morphological investigation, mammals and birds seem to senesce, provided they live long enough to demonstrate such changes. Comfort (1979) believes that reptiles, amphibians, and fish also senesce; however, the strength of this inference is blunted somewhat by the fact that the growth and life cycle patterns of these life forms may be so easily modified by diet, temperature, diapause, and so on, as to make it difficult to distinguish effects of aging from adverse environmental effects. In fish, for example, there is an inverse relationship between water temperature and life span. The bounds are set such that at too low or too high a temperature, the animal may live an exceedingly long or short life; but reproduction fails to occur at either extreme (Beaverton 1987). In both cases, the heritability of the life span falls to zero. Thus, in these organisms under such conditions, longevity and its inheritance are matters of the temperature adaptations of the reproductive process. I defer the detailed description of vertebrate aging to chapter 5, where I describe in detail the normal age-related changes in humans. Mammals. What Accounts for the Diversity? Much of the early interest in the identification of factors that affect mammalian aging was motivated by the need to understand why human longevity should be so much greater than that observed for even our nearest relatives. The relationship between body size and longevity was examined by Sacher (1959). Although the two variables are well correlated, there is still a lot of scatter in the data, not to mention that the relationship utterly fails to predict the longevity of humans. We are much longer lived than our body weight would predict when compared to other primates such as the gorilla (see table 4.7). A much better correlation was obtained when brain weight, which often appears to be relatively larger in longer-lived species, was taken into consider-

4.5 Comparative Aspects of Aging

127

Table 4.7 Maximum Recorded Life Spans for Selected Mammals, Birds, Reptiles, Amphibians, and Fish

Primates

Carnivores

Ungulates

Rodents

Bats Birds

Reptiles

Amphibians

Fish

Scientific name

Common name

Papio ursinus Macaca mulatta Pan troglodytes Gorilla gorilla Homo sapiens Felis catus Canis familiaris Ursus arctos Ovis areis Sus scrofa Equus caballus Elephas maximus Mus musculus Rattus rattus Sciurus carolinensis Hystrix brachyura Desmodus rotundus Pteropus giganteus Streptopelia risoria Larus argentatus Aquila chrysaëtos Bubo bubo Eunectes murinus Macroclemys temmincki Alligator sinensis Testudo elephantopus Xenopus-laevis Bufo bufo Cynops pyrrhogaster Rana catesbiana

Chacma baboon Rhesus monkey Chimpanzee Gorilla Man Domestic cat Domestic dog Brown bear Sheep Swine Horse Indian elephant House mouse Black rat Gray squirrel Porcupine Vampire bat Indian fruit bat Ringed-turtle dove Herring gull Golden eagle Eagle owl Anaconda Snapping turtle Chinese alligator Galapagos tortoise African clawed toad Common toad Japanese newt

Aphya pellucide

Cod Pike Halibut Guppy Sturgeon

Maximum life span (years) 45 40 53 54 122 28 34 47 20 27 46 70 3 5 24 27 19.5 31 35 44 46 68 29 58+ 52 100+ 15 36 25 16 20+ 40+ 60+ 6 82+

Source: Comfort (1964) and Lamb (1978); updated by Austad (1997).

ation. Taking both body weight and brain weight into consideration gave a much tighter correlation with life span than did either one by itself, and it permitted a much better prediction of the human life span. Sacher’s interpretation of this finding rested on the observation that metabolic rate is inversely proportional to body weight. If one assumes that the frequency of metabolic errors is directly proportional to metabolic rate, then it follows that larger animals will have fewer metabolic errors and hence longer life spans. Brain size is impor-

tant because animals with larger brains are assumed to have superior homeostatic mechanisms, which tend to reduce the incidence of metabolic errors. Sacher reasoned that larger animals will live longer than smaller animals because of a decreased frequency in metabolic errors, but that for animals of comparable size, the ones with larger brains will live longer for the same reason. The two traits therefore reinforce one another. Hofman (1983) also explored this relationship and arrived at similar conclusions. Unfortunately, this relationship between brain size and

128 Chapter 4 Evolutionary and Comparative Aspects of Longevity and Senescence longevity does not appear to be true. Not only are there serious logical problems inherent in a statistical analysis of correlations, and not only are other organs (liver, spleen, and so on) also correlated with longevity (Finch 1990; Gaillard et al. 1994), but the mortality coefficients of humans and other primates (e.g., rhesus monkeys; see table 2.6) suggest that we live long primarily because of our lower initial mortality rate (IMR) and not because our rate of senescence (mortality rate doubling time, or MRDT) is low. An alternative explanation is that smarter behaviors may affect the IMR, which, if true, suggests that selection for more effective behaviors and for the brain structures associated with them might have played a role in the evolution of human longevity. In other words, the long life spans of large-brained mammals could be the secondary consequence of selection for increased brain size (Wilson 1991). Increased brain size would have a selective advantage because smarter individuals should have an increased reproductive fitness relative to dumb ones. However, in order for large brains to develop, gestation times have to be long and litter sizes small because the construction of complex brains takes time and is resource demanding. The resulting decreased reproductive rate must be compensated for by an increased reproductive period. Of course, the width of the mother’s pelvic girdle sets an upper limit to the size of the neonate’s head, and thus sets an upper limit on brain size and function as well. Humans have escaped this constraint by normally being born prematurely, at least in comparison with our closest relatives, the chimpanzees. This timing makes it possible for our necessary brain growth to take place after birth, free of the constraints imposed by the pelvic girdle. However, this strategy results in an even longer postnatal developmental period and a correspondingly increased span of time during which parental care must be given. Thus, long life could be viewed as a pleiotropic consequence of selection for increased brain size taking place in K-selected species. This interpretation is consistent with the data, but it is also heavily dependent on unproven assumptions. Holliday (1995) pointed out that postreproductive factors might also affect longevity selec-

tion. He suggested that the selective advantage for parental care in humans led to the evolution of menopause as an adaption to ensure the survival of younger individuals, particularly since nonreproducing females cared not only for their own children but also for grandchildren or other young relatives (each of whom contain copies of their own genes and whose survival affects the reproductive fitness of the nonreproductive female; but see chapter 14 for a different view of this matter). I discuss these social interaction theories of longevity in more detail in chapter 14. A wide-ranging investigation of life span in various orders of mammals and birds was conducted by Prothero and Jurgens (1987). Their studies, summarized in figure 4.11, show that many species of birds, bats, and primates have a substantially longer life span than do mammals of comparable body size. In other words, the relationships between body size and longevity that were developed by Sacher (1959) and Cutler (1975) do not appear to apply equally to all animals. Line F (aquatic mammals) in figure 4.11 can be treated only as an approximation because of small sample sizes and indirect estimation of maximum life spans. The remaining data are reliable and show that birds, bats, and primates have life spans greater than those expected for mammals of comparable size. We have already discussed the hypothesis that primate longevity increased as a result of selection for increased brain size. In the case of birds and bats, this increase cannot be statistically attributed to an increase in brain size but may be related to the metabolic demands of flight (see the discussion of birds below). Thus, that hypothesis may be a valid explanation only for primates, which might explain the inability of Gaillaird and colleagues (1994) to replicate it in a variety of other mammals. If the data on aquatic mammals prove to be reliable, the explanation for their greater-than-expected life span may well involve their cardiovascular and respiratory adapations for long-term deep-sea diving. Other mammals may require other explanations. If longevity is attributable to some sort of cellular function, then a general correlation between body size and life span is not surprising.

4.5 Comparative Aspects of Aging

102

F Aquatic mammals rally gene ls a amm

129

Maximum life span (years)

Primates B

A

101

C

Bats

an in

Birds D E

fe sp um li im x a

M

pan in

M

fe s ean li

m

rally

ene als g

m

mam

100

10-1

10-3

10-2

10-1

100 101 102 Body weight (kg)

The fact that specific physiological and cellular specializations characteristic of particular species may interact with the animal’s life-history characteristics to indirectly result in a maximum life span that is greater than would otherwise be expected is also not surprising. In the case of primates, the explanation may involve behaviors and brain size; for aquatic mammals, it may depend on adaptations for diving; in birds and bats, the operative explanation may have something to do with the physiological requirements of flight. Birds. For a long time birds were believed to be short-lived creatures, exhibiting a life span comparable to that of laboratory rodents and possessing no characteristics interesting enough to tempt biogerontologists to overcome the handicaps associated with studying them. This belief probably grew out of the fact that in the past the only birds studied in any detail by biogerontologists were chickens, quail, and other poultry. These birds were probably chosen for their economic importance as well as for their convenience, but all of them are weak-flying, short-lived species. Thus, early conclusions were based on a small and skewed sample. Holmes and Austad (1995a,b) reviewed the characteristics of a wide variety of different birds and concluded not only that many birds are very long-lived, but that they may be particularly well adapted for the study of retarded aging (table 4.8). Their data show that birds are

103

104

Figure 4.11 Regression lines for mean and maximum life spans in various vertebrates as a function of body weight. The lines labeled A, B, C, D, and F refer to maximum life span in the specific groups; line E refers to mean life span in mammals generally. (After Prothero and Jurgens 1987.)

105

longer-lived and age more slowly than comparably sized mammals, both in nature and in captivity. For example, an inspection of table 2.6 reveals that even the Japanese quail, which is the shortest lived and most rapidly reproducing and senescing bird species yet documented (Holmes and Austad 1995a), nonetheless has a maximum life span of 5–8 years and an MRDT of 1.2. This MRDT value is about four times greater than that of the ordinary laboratory rodent and is indicative of a slower aging process than is found in rodents. The MRDT of the European robin is equivalent to that of humans; only its high IMR value of 0.5 restricts it to a 12-year maximum life span. Certain birds have life spans comparable to that of elephants, even though they are only a fraction of the elephants’ size. Scarlet macaws have lived in captivity for more than 90 years, ravens for 69 years, and the royal albatross for more than 50 years. Hummingbirds are the smallest bird species and have the highest metabolic rate, yet they have maximum life spans in nature of more than 12 years, values that far exceed those predicted on the basis of their body weight. Each of these species reproduces slowly and is not fecund, in contrast to the reproductive strategies adopted by common poultry. What is most interesting is that birds attain these high longevities despite having metabolic rates equal to or much greater than those of comparably sized mammals.

130 Chapter 4 Evolutionary and Comparative Aspects of Longevity and Senescence Table 4.8 Comparison of Mammal and Bird Body Mass, Longevity, and Reproductive Senescence

Species Mammals Mouse Human Domesticated birds Chicken Quail Wild birds Broad-billed hummingbird Barn swallow Canary Raven Wild sea birds Common tern California gull Northern fulmar Short-tailed shearwater

Body mass (gs)

Maximum recorded life span (years)

20 50,000

4 122

Reproductive senescence? Yes Yes

— —

20+ 7

Yes Yes

5 16 22 1200

14 16 24 69

Yes Yes — —

>20 >25 >42

No No No No

— — — —

Source: from Holmes et al. (2001) except for nothern fulmar, from Ricklefs and Finch (1995).

Because the production of potentially damaging oxygen free radicals (believed to be a major cause of senescent damage; see chapter 10) is thought to be proportional to the lifetime energy expenditure, then it appears as if many birds have a more efficient means of coping with freeradical damage. In addition, birds generally have much higher blood glucose levels and higher body temperatures than do mammals, which should accelerate the formation of glucose crosslinking of various macromolecules (thought to be a major cause of senescent damage; see chapter 10). Again, the supposition is that birds may possess qualitatively different processes for coping with these potential mechanisms of senescence, and this is supported by the data of chapter 10). These unique processes may extend beyond the amelioration of senescence. For example, the genome size of birds is only one-third that of mammals, and among the birds, the genomes of strong flyers are smaller than those of weak flyers (Hughes and Hughes 1995). The origin of these processes may be connected to the requirements of their life-history strategies as expressed within

the context of flight. Sorting out which of these unique avian properties is related to their delayed senescence and thus may offer us some insights into the development of future interventions promises to be an important future development. 4.5.4.5 Plants: Senescence as

a Developmental Process The area of plant aging and senescence has developed independently of the field of animal gerontology. This should not be surprising, considering that plant senescence arose out of the needs of plant physiologists and the agricultural community, whereas gerontology has its origins in medical science and evolutionary theory. There are both similarities and differences in the senescence patterns of plants and animals. Good reviews of this topic have been presented by Nooden and Thompson (1985), Nooden and Leopold (1988), Nooden and Letham (1993), and Nooden and Guiamet (1996). Refer to those sources for detailed information as well as to the summary from the prior edition of this text at http.//bio.wayne.edu/ prefhtml/arking/textbook/supplement.html.

4.6 Epilogue

4.5.4.6 The Modular Nature of Organisms

One important lesson the comparative approach has taught us is that basic cellular and physiological regulatory mechanisms are highly conserved across species. Perhaps the best example of such evolutionary conservation is the insulinlike signaling pathway (ISP), which regulates energy metabolism. Qualitative changes in energy metabolism lead indirectly to significant changes in longevity. The ISP can be best understood as constituting a major proximal mechanism by which the ultimate evolutionary forces act so as to increase or decrease longevity. Such a proximal control mechanism is of great interest. It may be understood as an example of a basic control module present in all cells and elaborated upon by natural selection as organisms increase in complexity. It is likely that the singular importance and generic mode of action of the ISP would have taken much longer to recognize without the constant use of the comparative approach to biogerontology, whereby conserved mechanisms allowed insights obtained with simpler animals to be swiftly translated to more complex organisms. This example alone proves the utility of the comparative approach to biogerontology. The proper study of humans is not limited to our species alone but involves our evolutionary history as well. 4.5.4.7 Public and Private Mechanisms of Senescence

Martin et al. (1996) pointed out that some senescent mechanisms are highly conserved within a species as well as across species and orders.

131

Other senescent mechanisms are highly restricted, usually being found only in particular individuals and their families. This observation led to the conceptual delineation of “public” and “private” aging or senescent mechanisms. Although many of the private senescent mutations and mechanisms are of clinical interest, we will likely learn more about the major aspects of the biology of aging by focusing our attention on the conserved public mechanisms and processes.

4.6 Epilogue We have focused on the scientific analysis of the evolution of aging and of the close relationship between reproduction and senescence in scientific terms. But to be fair, we must recognize that intimations of this strategy were earlier sensed by the artists among us. Consider, for example, Shakespeare’s Sonnet Number 12: When I do count the clock that tells the time, and see the brave day sunk in hideous night; . . . Then of thy beauty do I question make, That thou among the wastes of time must go, . . . And nothing ’gainst Time’s scythe can make defence Save breed, to brave him when he takes thee hence.

This page intentionally left blank

Cumulative survival

1.0 0.9 0.8 0.7

Part III

0.6 0.5

How Do We Age?

0.4

WT

0.3

p66+/-/p66

0.2 0.1

500

600

700

800 Time (days)

900

1,000

1,100

This page intentionally left blank

5

Human Aging

5.1 A Perspective on Human Aging If we are fortunate, we will age. What do we have to look forward to? The popular assumption is that the end of our life will most likely be made up of physical disabilities, mental incompetence, familial rejection, loneliness, and poverty—that aging will inevitably lead to severe loss of physical, mental, and social functions. This pessimistic view has found its way into our life and our literature. Today, it is becoming more and more clear that these dour assumptions and conclusions may not necessarily be completely true or accurate. It is increasingly evident that, as Rowe and Kahn (1987) pointed out, the mere association of physiological and cognitive deficits with age is insufficient evidence for us to conclude that these deficits are determined by age. To some extent, this misplaced emphasis might have been a natural consequence of the fact that most early research on aging was carried out in medical schools as part of the geriatrics program and that aging thus may have been viewed as a special case of the disease process. As pointed out in chapter 3, not only is aging an individual process that is easily confounded with the pathologies of covert illness and disease, but it is superimposed on the substantial heterogeneity characteristic of the human species. A good portion of this heterogeneity may be due to genetic factors, but an even larger proportion can be traced to environmental differences. A variety of cultural and psychosocial factors can play a significant role in modulating the aging process.

The Duke First Longitudinal Study (Palmore, 1982) found that such nongenetic factors could contribute as much as 16 years to longevity in men and as much as 23 years in women (see chapter 8 for details). Furthermore, our social and physical lives do not exist in separate worlds with no communication between them. Rather, it is now certain that these cultural and psychosocial factors work by affecting our neuroendocrine and immune systems and in this manner modulate the physiological and cellular levels of body function. The brown antichinus and the Pacific salmon, for example, die as a result of stress-related, hormone-induced physiological dysfunction. Psychological stress shortens our telomeres. Our physical and social worlds are linked by well-known biological mechanisms. These processes, in turn, provide the physical basis underlying the concept that people who have led a well-integrated life often demonstrate less loss of function than others. One example of such an individual is David Gill, whose exercise data are shown in figure 3.9. These data show no age-related decrement in heart rate and oxygen uptake between the ages of 35 and 65 years. This is by no means an isolated example, as will become clear in the remainder of this chapter. These newer insights have shown us that aging is much more plastic and individual than was once thought. Two lines of evidence support the plasticity of aging. One is the dispassionate description of agerelated changes, which will fill the remainder of this chapter. The other is anecdotal and serves mainly to show us the possibilities. Many people today are alive, vital and active at advanced ages;

135

136 Chapter 5 Human Aging these people have aged successfully. Some are famous, most are not. Many of our fellows are old in their years, but young in their impact on our society. They have achieved their impact because they have remained professionally active and effective. They have not retired from life. These examples should make us realize that the loss of our mind and independence is not an inevitable effect of the aging process. Such loss may be more common than is desirable, but it is not inevitable. This concept of successful aging should not be interpreted overoptimistically so as to suggest that elderly individuals will show no decrease in capacity or function with age. Of course they will. The point is not that age-related changes will not occur; rather, my focus is to describe the inevitable changes and to illuminate the mechanisms by which the other changes may be modulated so that we each may strive to age successfully and retain our effectiveness as individuals and as members of society.

5.2 The History of Human Mortality and Longevity Most of our prehuman ancestors were not long lived. The long life span characteristic of modern humans arose in the Upper Early Paleolithic among our Homo sapiens ancestors (Caspari and Lee, 2004; Cutler, 1975). It may well be that the extension of our potential longevity to the current norm is what allowed us to become fully human, if for no other reason than it gave us the gift of time to learn new things and develop new skills. Longevity also gave us the time to transfer this cultural knowledge to our offspring (see Figure 14.7), and so expand our cultural evolution to the symbolic level. The ancient Egyptians described the maximum human life span as 110 years (D. E. W. Smith 1993), which isn’t too far off from our present measurement of 122.5 years. Roman funerary inscriptions describe individuals who lived into their 80s. Excavations from medieval cemeteries reveal skeletal remains from individuals in their 60s (D. E. W. Smith 1993). Although these life spans are similar to those seen

today, it would be a serious error to assume that the life expectancy and age structure of those early societies were similar to our own, as the quantitative data of table 5.1 indicate. A 65-yearold man or woman is so common as to excite no attention today, but they would probably have been a rare spectacle in most early societies. Life expectancy at birth in most places in the past ranged from 25 to less than 50 years. Infants and children must have accounted for a large number of those dying early, for this is the only logical explanation for the rise in the life expectancy between birth and 5 years of age in Massachusetts in 1789. Presumably, children died of infectious diseases, women died in childbirth, and men died in accidents. Combining such short life expectancies with the high infant mortality rate explains why the human population had such a low rate of increase for such a long time. Altering the environmental conditions responsible for such conditions led to a slow increase in life expectancy, but still it took England about 300 years (from the mid-1500s to the mid-1800s) to increase the life expectancy by only 7 years. Change has since accelerated. The more recent accomplishments are illustrated in figure 2.14, which shows that during the 20th century the median longevity for females has increased about 40% (from 58 to 81 years), while the mean life expectancy at birth for both sexes combined has increased about 50% (from 50 to 75 years; D. E. W. Smith 1993). This progress has been uneven. In a demographic sense, undeveloped and poor countries such as Sierra Leone are lagging behind the United States by about 150 years. This difference reinforces the observation that longevity depends on the social environment. Had our ancestors been able to indulge in time travel, they would have been amazed at the increase in life expectancy and would have remarked about how this single phenomenon has transformed society. The more astute among them undoubtedly would have pointed out that this increase in life expectancy is really due to a striking decrease in premature mortality and not to any fundamental tinkering with the aging processes. And they would have been correct, for the time being.

5.3 The Relationship between Aging and Disease

137

Table 5.1 Life Expectancies in Different Times and Places Life expectancy Place and time

At birth

England

1541 1661 1781 England and Wales 1838–1854 1871–1880 1881–1890 1891–1900 United Kingdom 1983–1985 Sweden

Massachusetts

United States Canada Japan USSR Bangladesh Sierra Leone

Male Female

1816–1840 1851–1855 1871–1880 1881–1890 1891–1900 1985 Male Female 1789 1855 1890 1895 1901 1986 Male Female 1984–1886 Male Female 1987 Male Female 1985–1986 Male Female 1988 Male Female 1985–90 Male Female

33.7 35.9 34.7 40.9 43.0 45.4 46.0 71.8 77.7 41.5 42.6 47.0 50.0 52.3 73.8 79.7 28 39.8 43.5 45.3 47.8 71.3 78.3 73.0 79.8 75.6 81.4 64.1 73.3 56.9 55.9 39.5 42.6

At age 1

At age 65 or at indicated ages

71.5 77.4

13.4 17.5

73.3 79.2 age 5 = 41

14.7 18.5 age 60 = 15

71.1 78.0 72.6 79.3 75.0 80.8 65.0 73.9 63.7 61.5

14.7 18.6 14.9 19.2 16.1 19.7 12.3 15.8 12.2 12.0

Source: from tables 1.2, 1.3, and A-2 of D. E. W. Smith (1993).

5.3 The Relationship between Aging and Disease In 1900, the three leading causes of death in the United States were respiratory diseases, digestive diseases, and central nervous system diseases. In 1986, they were cardiovascular diseases, malignancies, and accidents (D. E. W. Smith 1993). In 1996, the prediction was made that coronary disease would no longer be a major public health problem by early in the 21st century (M. S. Brown and Goldstein 1996). Not only have the

proximate causes of death been completely transformed during the 20th century, but the current death rate is less than half what it was in 1900. Fewer of us die prematurely, and we die from different immediate causes than was the case 100 years ago. Now that the norm is a full life, the question arises as to what insight, if any, an investigation of the diseases of the elderly may give us into the biology of aging. Human age-related pathology is usually considered the province of geriatric medicine and is often overlooked in discussions of experimental biogerontology. Holliday (1995)

138 Chapter 5 Human Aging maintains that the distinction between natural aging and pathological aging is artificial. No single individual suffers from all the age-related diseases, but no single individual exhibits all the manifestations of normal aging either. Therefore we must examine populations to assess the whole spectrum of age-related changes seen in humans. In this process, we usually exclude all people who are suffering from an age-related disease, presumably because we believe there is something qualitatively different between normal aging and pathological aging. However, this distinction is arbitrary and a wasteful oversight of a huge amount of data. It is useful to view age-related diseases as each representing a particular outcome of the failure to maintain a particular anatomical structure or physiological process, as indicated in table 5.2. In this light, the utility of viewing diseases as systemic failures that highlight the weak points of the evolved anatomical and physiological design of the organism and thereby allow us to identify and investigate them becomes obvious. Would the biology of cell division have been a concentrated focus of research if no one had ever died of cancer? Note, for example, that the two leading causes of death in 1986 (cardiovascular diseases and cancers) are attributable, respectively, to failure to prevent damage to and otherwise maintain the linings of the blood vessels and to the signal transduction and other mechanisms that regulate cell division. Both of these processes

Table 5.2 Relationship between Cell or Tissue Maintenance and Human Age-related Diseases Failure of maintenance Neurons Retina, lens Insulin metabolism Blood vessels Bone structure Immune system Epigenetic controls Joints Glomeruli Source: after Holliday (1995).

Major pathologies Dementias Blindness Type II diabetes Cardiovascular and cerebrovascular diseases Osteoporosis Autoimmune disorders Cancer Osteoarthritis Renal failure

are in the forefront of modern biological research because of the social and political forces that provide research funding for the study of these common age-related diseases. We study them to improve our understanding of the etiology of the disease in question and to develop procedures for better alleviating the disease or delaying its onset. So in one sense, studying age-related diseases as individual entities apart from one another, as we do today, is part of modern geriatrics and biogerontology. But it is not enough. In Holliday’s (1996a) view, this strategy is seriously flawed because the focus on disease keeps us from integrating our knowledge by studying all the cellular and molecular changes that precede and bring about the loss of reserve capacity that occurs in healthy people in the absence of disease and that must also prepare the way for the overt onset of any disease. Holliday (1996b, p. 90) has stated that “the study of gerontology must have a central position in biomedical research” if we are to have a realistic hope of reining in ever-increasing health care costs without adversely affecting quality of care or length of life. Whether biomedical science will be reorganized in this way cannot be foretold, nor is it particularly germane to our central goal of understanding aging (but see chapter 15). However, Holliday’s view that we should incorporate the study of disease into the study of gerontology by viewing diseases as systemic failures that identify biological weak points is germane to our goal. In fact, I have already discussed this problem along with a detailed presentation of one example in chapter 3. Following Holliday’s advice has allowed us to unify the biology of aging at a functional level instead of arbitrarily continuing the division between normal and pathological aging. The mechanisms of aging identified in this chapter are discussed in greater detail in many of the following chapters, as are the kinds of interventions necessary to prevent or delay the onset of systemic failures. In addition, adopting Holliday’s point of view integrates these functional levels with the evolutionary models discussed in chapter 4, for here we confront the gory and personal details of what the disposable-soma theory, for example, im-

5.4 An Overview of Theory

plies when it states that selection pressures shift the allocation of energies away from somatic maintenance and toward reproduction. The large increases in longevity obtained during the 20th century were based on the containment and curing of disease (figures 2.14 and 2.15). That strategy is at the point of yielding diminishing returns. The chronic age-related diseases are the great killers of today. It has been estimated that curing cancer or heart disease or diabetes would add only about 4–7 years to the average female life span today, and simultaneously curing all three of them would add about 15 years (Olshansky et al. 1990). Those extra years would be spent in an aging state susceptible to other age-related diseases. Given the huge financial costs involved in conquering disease, and given the financial constraints facing health care today, then it might not be wise public policy to try to eradicate these diseases one by one; the money might be better spent elsewhere. This approach has been labeled the “retail approach” to staying healthy, and I deal with it in detail in chapter 15. An alternative strategy is now becoming realistic. All of the chronic diseases are age-related diseases whose symptoms reflect the organism’s failure to maintain a particular organ system. If one could retard the aging processes, then one should be able to simultaneously retard some large number of allegedly independent diseases and thereby incur much lower direct financial costs in the quest for continued health (the “wholesale approach”). It has been shown that genes associated with the extension of life span in mice dying of cancer are also associated with an extended life span in mice dying of nonneoplastic diseases (R. A. Miller et al. 2002b). Much of the biomarker work done in both humans and laboratory animals (see chapter 3) leads to the conclusion that there is a general aging process. In other words, many late-life diseases may be under the strong influence of some common form of genetic control. Intervene at those (currently unknown) controls and one should be able to delay or inhibit the onset of numerous diseases. I describe known genetic mechanisms that might fulfill this role in chapters 7 and 14

139

and discuss their implications for both individuals and society in chapter 15. In the meantime, as I progress through the description of aging changes in humans, I point out examples of the manner in which aging changes set up the particular organ system for possible failure. These linkages between aging and disease constitute the evidence supporting the wholesale approach to disease control.

5.4 An Overview of Theory In previous chapters I briefly described aging in selected experimental organisms. In this chapter I describe the aging processes in mammals, using humans as an example. A multitude of changes take place with the passage of years. Even if I had the space to describe every such change, I could still not chronicle all of them but would have to be selective. How, then, do we choose which variables to describe and which ones to ignore? To a large extent, adherence to the CPID (cumulative, progressive, intrinsic, deleterious) criteria outlined in Chapter 1 ensures that we will recount only the significant age-related changes and omit many interesting but probably unrelated time-dependent changes. A second factor that we must consider is the role of disease. If we regard disease as indicating the weak points in the body’s structure and function, then some of the more common breakdowns should be examined, if only as an empirical test of our view of the role of disease. A third factor influencing our narrative is the demands of theory. Theoretical explanations of the biology of aging make some facts more important than others. For example, knowing that extended longevity is hereditary in laboratory animals makes one sensitive to evidence of different patterns of gene action in long-lived animals as opposed to shorter lived animals. Or being aware of the theories suggesting that cross-linking between macromolecules plays an important role in the aging processes might indicate that the details of connective-tissue aging should be given special significance. If you have not already done so, you may wish to skip ahead and look over the

140 Chapter 5 Human Aging theories summarized in tables 9.3 and 9.4 to gain a brief idea of the kinds of facts the theories require us to see.

5.5 Plasticity and Patterns of Aging We often talk about aging as if it were a unitary process that all members of the species undergo in the same manner. Perhaps this idea arose from the superficial resemblance of the old survivors to one another; they usually had gray hair and wrinkles and suffered from the slings and arrows of time’s misfortune. Perhaps the idea arose by analogy with development, which all members of the species undergo in the same manner. No matter how the idea arose, it is not correct, as suggested by the evidence presented in table 3.1 and figure 3.7. Note the wide variability in the response of individuals for the single trait of creatinine excretion. Aging occurs gradually, with some processes beginning to decline in a person’s 20s or 30s, while other processes remain relatively untouched until the late 60s and beyond. Superimposed on this population heterogeneity is a substantial amount of individual heterogeneity, arising from the person’s genetics, development, and environment, suggesting that the gradual nature of aging is not true for many, perhaps most, members of the species. If different traits can begin their decline at different ages in different individuals, and if most traits can change more or less independently of one another, then older adults are a heterogeneous group characterized primarily by their individual patterns of aging. In addition, the results from the Human Aging Study begun in 1955 by the National Institute of Mental Health found that much of what was popularly called aging is really a function of disease, sociocultural effects, and lifelong personality traits (Butler 1995). Once we correct for the confounding effects of such variables, we find that the age-related declines in function that occur in healthy adults are lower than the popularly assumed levels, although they are extraordinarily variable and individual. This individual heterogeneity must be kept in mind even as we present data based on statistical analysis of

groups of people: The mean does not adequately describe the individual.

5.6 Age-related Changes in Humans: Detailed Survey 5.6.1 Overall Anatomical Changes Table 5.3 lists age-related changes in anatomical measurements and indices. The relationships involved likely are not simple. In chapter 3, I discussed in some detail the parameters affecting changes in weight with age. There is no reason to believe that the factors governing the other indices listed in table 5.3 are any less complex than that one. A close examination of these indices yields no simple or overall detectable pattern of growth. Instead, one is faced with a mosaic of differential growth patterns, in which some features reach their peak values in the third decade of life (e.g., facial index), a few reach their peak values in the eighth decade (e.g., thoracic index), and the others are scattered throughout the intervening years. Because these indices are primarily skeletal measurements, postmaturational bone growth must be taking place. Furthermore, this highly localized growth is taking place not only against a background of slow overall growth through the fourth decade followed by decline thereafter, but also against the background of other complex age-related skeletal alterations, such as simple osteoporosis, and even more extreme abnormal alterations. These changes in the skeletal system are further confounded by ethnic and/or environmental effects. This heterogeneity of growth patterns is not limited to the skeletal system. Cross-sectional data show that about half the internal organs attain their peak weight in the fourth decade and then decrease in weight, while the rest of the internal organs attain their peak weight sometime between the fifth and eighth decades. Clearly, there is neither a single nor a simple pattern of normal growth, but probably one related to the normal functioning of each organ. Accompanying these size and weight changes are changes in the composition of the body. We

5.6 Age-related Changes in Humans: Detailed Survey

141

Table 5.3 Changes in Anatomical Measurements and Indices Probably Due to Aging Weight Stature Span Thoracic index Biacromial diameter Relative shoulder breadth Chest breadth Chest depth Sitting height Relative sitting height Head circumference Head length Head breadth Cephalic index Cephalo-facial index Total face height Facial index Upper face index Nose height Nose breadth

Increase to 50; decline from 60 Increase through 30–34; decline from 40 Increase through 30–34; decline from 40 Increase through 70–74 Increase through 35–39; decrease from 55 Increase through 45–49 Increase through 50–54 Increase through 50–54 Increase through 35–39; decline thereafter Slow decline after 49 Increase through 35–39; slow decline after 54 Increase through 50–54 Increase to 40 and slight decline thereafter Decline from 35 Rise through 75–79 Increase through 30–34; decline thereafter Increase through 25–29; decline thereafter Increase through 30–34; decline from 55 Increase through 55–59 Increase throughout all age groups

Source: from Rossman (1977).

each become literally a different person as we grow older. The amount of fat in the body increases with increasing age in both absolute and relative terms, and this increase in fat is accompanied by both absolute and relative decreases in the amounts of cell solids, bone minerals, and water. These changes are consistent with the longitudinal and cross-sectional, cross-cultural data in figures 3.2–3.5, which demonstrate the reality of the changes in body build we each undergo.

5.6.2 Changes in the Skin and Connective Tissue More money and effort are probably expended by individuals in the effort to hide and disguise the normal age-related changes occurring in the hair and skin than in any other organ system. The cosmetic industry thrives on our vain desire to remain young, and fortunes await those who can enhance the naturalism of our illusions. The importance of a discussion on skin in a gerontology text has little to do with morbidity or mortality. Very few patients die of old skin or succumb to

skin failure. The importance of skin appearance is primarily psychological, for we read our mortality in our skin. This phrasing has literal as well as poetic meaning, as the discussion of human biomarkers in chapter 3 made clear (see figure 3.19). Since much of our communication with one another is nonverbal, we probably depend on subtle interpretations of each other’s skin to tell us much about the other’s health, social status, personality, and the like. The emotional impact of skin aging should not be underestimated, nor should we ridicule the serious attempts by cosmetic companies and health professionals to alleviate the impact of skin aging and maximize a positive self-image. It may be more true than we know that we must feel good about ourselves or suffer the consequences. 5.6.2.1 Normal Structure and Function

The skin is probably the largest organ of the body both in surface area and in weight, constituting about 16% of the body weight (Bloom and Fawcett 1968). Our skin holds us in. It is the boundary between our bodies and the outside world, and

142 Chapter 5 Human Aging we depend on it for our literal physical integrity. The skin is composed of several layers. The outermost layer is the epidermis, a specialized epithelial cell layer. Under the epidermis is the dermis, a vascularized layer of connective tissue. The innermost layer is a loose layer of connective tissue called the hypodermis. Various structures penetrate these horizontal layers: hair follicles, sweat glands, sebaceous glands, and so forth. The color of the skin depends on three factors. First, the skin itself is yellow, owing to the presence of the organic pigment molecule carotene. Second, the blood showing through from the underlying dermis implants a reddish hue. Finally, the presence of varying amounts of melanin granules, present in the epidermis, produces shades of brown through black. We are constantly losing our skin. New cells are constantly produced by cell division in the basal layer of the epidermis and are moved upward by the subsequent appearance of newer cells beneath them. As they move up and outward, they synthesize huge amounts of the protein keratin, the same protein that forms hair and nails. As the cell accumulates massive amounts of keratin, it becomes metabolically inactive, dies, and leaves behind only a flaky residue of a cell, which sooner or later is shed from the surface. The whole process takes between 2 and 4 weeks. The skin of the palms and soles is thicker than the skin of the rest of the body because of localized thickening of the keratinized dead cell layer of the epidermis in those regions. The main function of the dermis is to provide an extensive and tough matrix both to support the many structures embedded in it and to provide a junctional base to which the epidermis may adhere. The dermis is composed the fibroblasts that produce the extracellular matrix connective-tissue fibers, most of which are collagen and the remainder elastin. The collagen molecules are flexible but offer great resistance to a pulling force; they prevent the skin from being torn by overstretching. Elastin is a springy material that maintains normal skin tension but that will stretch to allow movement of the underlying muscles and joints. The main cellular components of hypodermis are mainly adipocytes. The collagen and elastin fibers of the hypodermis are continuous with

those of the dermis. Depending on the region of the body and the individual’s nutritional state, variable numbers of fat cells are located in this layer. This layer serves as a shock absorber against trauma to the internal organs, as a storage depot for high-calorie food reserves, and as insulation against excessive loss of body heat. Finally, skin is an active element of the immune system (Edelson and Fink 1985). The keratinocytes produce not only the protective outer layers of dead cells filled with keratin, but also hormonelike molecules (interleukin and thymopoietin; see chapter 13) that are capable of profoundly affecting lymphocyte function. In addition, the dermis contains small dispersed populations of two other types of cells thought to play a controlling role in immunological function. The Langerhans cells and the Granstein cells present antigens (foreign molecules) in the skin to specific helper or suppressor T-lymphocytes, which tend to migrate into the epidermis from elsewhere in the body. Depending on the detailed characteristics of these intercellular interactions, the immune response will be either enhanced or suppressed. Both animal and human studies have shown that particular abnormalities of the immune system are closely associated with particular defects in the structure and/or function of skin cells. 5.6.2.2 Age-related Changes

The skin’s ability to function as an effective boundary layer depends on the maintenance of its structure. However, the three main protein components of connective tissues (elastin, collagen, and proteoglycans) undergo extensive intermolecular cross-linking and side-chain modifications (Bailey 2001). These structural alterations lead to functional changes: cross-linking gives rise to stiffer tissues, and side-chain modifications give rise to altered cellular–extracellular matrix interactions. These dysfunctional structural changes take place in all connective tissue components, including skin, tendons, lungs, and vascular system, with increasing age. Their ubiquity makes it possible to estimate the extent of functional loss in internal tissues by sampling the skin (Gogly et al. 1998). In addition, it must be recognized that skin

5.6 Age-related Changes in Humans: Detailed Survey

is an active tissue in which matrix proteins and other molecules are continuously being synthesized and recycled. Age-related loss of function occurs not only because of the structural alterations mentioned above, which happen at a more or less steady rate, but also because of a slower matrix turnover. This results in a longer dwelling time of the molecules of the extracellular matrix and so increases the probability that these molecules will be adversely affected by the structural damages. The normal age-related changes encountered in human skin were described by Kligman et al. (1985), Kaminer and Gilchrest (1995), and by Fossel (2004). The epidermis becomes slightly thinner with age, and the dermis thins more than the epidermis. There is a marked decrease in the density of the dermal papillae and a consequent decrease in their interdigitation, as well as an alteration in their cell–cell adherence within each papillae. These stubby interlocking protrusions of the epidermis and the dermis hold the two layers in close contact with one another. The result of their loss is that the epidermis is held less tightly to the underlying dermis. This alteration accounts in part for the looser feel of aged skin. There are reported to be no significant changes in the fine structure of the epidermal cells themselves. By the eighth decade, however, keratinocyte turnover in the epidermis slows by about 50%, and this more or less parallels the thinning of the epidermis. The thinning of the dermis is associated with a change in the weave of the collagen fibers located in this layer, resulting in less collagen per unit of surface area. The fiber bundles become larger and coarser, with larger spaces between them. The whole texture of the collagen becomes looser. There are several different varieties of collagen molecules, and this change in texture may have something to do with changing levels of synthesis and/or degradation of these different types. Alterations in either variable will affect the net synthesis level and alter the turnover time. The fact that males have a thicker dermis than females may help explain why female facial skin appears to deteriorate more readily with aging, particularly after menopause. In addition to the

143

changes in collagen, equally significant transformations take place in the elastic fiber network. A significant number of such fibers are lost from the upper layers of the dermis, although this may not happen in all cases (Robert and Robert 1988). This fiber loss is accompanied by a disorganized increase in the elastic fibers in the lower dermis. The architecture of this localized increase is abnormal; the fibers are thicker, longer, more disarranged, and less elastic. The net effect of this series of fibril transformations is looser skin that is more prone to wrinkle. The microvasculature also changes with age; the arterioles, capillaries and venules in the dermis and hypodermis become very sparse and irregularly formed. Atrophy of the hypodermal layer is one of the factors that makes it more diffcult for the aged to modulate their heat loss. However, this atrophy is not a generalized phenomenon occurring over the whole body. It usually occurs on the face and the backs of the hands; it usually does not occur about the waist or thighs, as all too many of us can testify. One of the most obvious signs of skin aging is the formation of wrinkles. Their origin was an enduring puzzle, but the old idea that the wrinkle was likely to be the visible manifestation of an abnormal structure has now been verified (Conter-Audonneau et al. 1999). As the skin undergoes senescence, it has decreased amounts of collagen IV and VII, and an abnormal elastin is found more prominently in the dermis. The important point may well be not so much the actual changes in any one type of collagen or elastin as the changes in the relative amounts (i.e., ratios) of the affected types. In the areas where future wrinkles will develop, the elastin fibers are interrupted, and the dermal collagen atrophies. There is also a decreased amount of chondroitin sulfate in the papillary dermis in the area of the future wrinkle, asymmetrical variation of glycosaminoglycans on edges of the wrinkle, and atrophy of the hypodermis in the area of the wrinkle. Thus wrinkles owe part of their origin to the decreased and disorganized fibril network of the dermis and part to the loss of the smooth padding provided by the fat cells of the hypodermis. The loss of this padding, the decreased elasticity,

144 Chapter 5 Human Aging and the looser binding of the skin layers allow the skin to be pulled downward more easily by the force of gravity, and so it sags. A decrease in the muscle mass with age also contributes to the loss of firmness of the aging skin. In addition to these structural changes, there is an age-related decrease in the DNA repair capacity of the skin and other tissues, a process which has implications for the genomic stability of the cells (Hadshiew et al. 1999; also see chapter 10). This may arise from the slower cell division rates and longer turnover times characteristic of older skin. Exposure of human skin to UV-A radiation results in a dose-dependent decrease of catalase activity. The activity is recovered over a period of time that is more or less coincident with the renewal time of new cells arising from the deeper layers (Hellemans et al. 2003). This, in turn, suggests that the radiation (and other toxic) exposures of the skin result in permanent inactivation of the various defense systems in the affected cells, and the tissue can only regain its normal defense levels by generating new, undamaged cells. In this scenario, anything that decreases stem cell replenishment of the damaged cells results in a loss of function that can contribute to the spiraling damage to the cell’s genome and proteome.

5.6.3 Changes in the Skeletal System Cartilage and bone are the specialized connective tissues that make up the skeletal system. Each of these tissues consists of cells and fibers such as collagen and elastin embedded in a nonliving matrix produced and secreted by the cells. Both the amount and the rigidity of this matrix distinguish these hard skeletal tissues from the relatively soft muscles. In both cartilage and bone, the living cells are isolated in small cavities within the matrix, and most of the tissue bulk is made up of the matrix. 5.6.3.1 Cartilage

Normal Structure and Function. Cartilage has the capacity for very rapid growth. Its matrix

imparts to it a considerable degree of stiffness. Together, these two properties—rapid growth and rigidity—make cartilage a particularly favorable skeletal element first for the embryo, in which the entire skeleton is first formed in a cartilage model and only later replaced by bone; second, in the growing bones of the young individual; and finally, in the joints and articulating surfaces of the bones in the adult. Because of its involvement in bone growth, cartilage is not an inert tissue, but a fairly delicate indicator of various metabolic disturbances, such as nutritional or hormonal deficiencies (Bloom and Fawcett 1968). Age-related Changes. With age, cartilage loses its translucence. Fewer cells are present, and the protein matrix undergoes some complex and still incompletely understood changes. The bluish color typical of young cartilage changes by age 20 to an opaque yellowish color. By age 30, cracking and fraying of the surface of cartilage joints begins to be visible. Although the aging cartilage cells retain the ability to make matrix, the rate of synthesis decreases relative to that of degradation, and the types of fibrils present in the matrix are altered (Tonna 1977). The most important regressive change is that of calcification, in which minute granules of inorganic calcium salts are deposited in the matrix. As the granules become enlarged and merge with one another, the cartilage becomes hard and brittle. This calcification of the matrix interferes with the ready diffusion of nutrients and waste products to and from the cartilage cell, and the cells die. With their death, the calcified matrix subsequently is slowly resorbed (Leeson and Leeson 1970). This process of calcification and resorption is a normal part of the phenomenon whereby cartilage is transferred into bone or whereby broken bones repair themselves. The altered rate of fiber turnover (e.g., synthesis and degradation) leads to a regressive change known as asbestos transformation, in which short, closely packed, coarse fibers, totally unlike any collagenous fibers, are deposited in the matrix. These silky fibers may spread over large areas and may lead to a softening of the matrix or even to

5.6 Age-related Changes in Humans: Detailed Survey

the formation of cavities within it (Bloom and Fawcett 1968). The cartilage cells show a decreased level of biosynthetic activity and proliferative ability. Dying cells are not often observed in growing cartilage but are quite common after growth has ceased. Cell death often leads to the formation of microscars. The matrix of cartilage in particular, and perhaps of all connective tissues in general, has often been viewed as an amorphous, passive, and uninteresting tissue component. Although the earlier literature indicated the existence of continuous and complex age-related changes in the matrix, the fundamental nature of these alterations was not understood (Balazs 1977). Other than collagen and elastin, the main component of the cartilage matrix is now known to be a very large and very elaborate protein molecule called proteoglycan. This molecule consists of a core protein, which is identical in animals of all ages, and to which are bound many molecules of chondroitin sulfate. The chondroitin sulfate electrostatically binds large volumes of water to the proteoglycan molecule, and this hydrated structure accounts for the resiliency of young cartilage. As the cartilage cells age, they synthesize and secrete chemically different forms of chondroitin sulfate, the major characteristic of which is that they are smaller molecules and consequently can bind less water. The increasing inability to function as a cushion at the joints results in tissue damage, inflammation, and in some cases, symptoms of osteoarthritis. This changing biosynthetic pattern is part of the normal developmental program of the chondrocytes. Alternatively, the existence of these altered forms of chondroitin sulfate may also be due to postsecretory damage to the molecule within the matrix and to slower turnover of such altered molecules in the aging extracellular matrix (Squires et al. 2003). Programmatic alteration of the biosynthetic pattern is also seen in other tissues. The transition from cartilage to bone is marked by the cessation of cartilage-specific type II collagen and its associated polysaccharides and the onset of bonespecific type I collagen and its associated calcium salts. Analogous changes occur in the connective tissues of muscles and, most likely, in other tissues as well. The potential complexity and vari-

145

ety of the different types of connective tissues are illustrated by the fact that there are at least nine forms of collagen alone, each of which appears to possess different functional and tissue specificity. The sequential replacement of these and other cellular components appears to be a normal and general phenomenon which decreases with age. 5.6.3.2 Bone

The adult human skeleton is composed of bone. Certain lifestyle alterations we will likely encounter stem from age-related changes that take place in our bones. Although bone approaches cast iron in its tensile strength, it is less than one-third as heavy (Bloom and Fawcett 1968). The architecture of its construction ensures the greatest strength with the greatest economy of materials and weight. Bones are dynamic living structures, surprisingly responsive to metabolic, nutritional, hormonal, and mechanical factors. The state of our bones is a good reflection of the manner in which we have lived and will live. Normal Structure and Function. There are two types of bone: trabecular or spongy bone and cortical or compact bone. Spongy bone is characterized by numerous interwoven partitions called trabeculae. The trabeculae branch and unite with one another to form a meshwork or honeycomb, the intercommunicating spaces of which are filled with marrow. The honeycomb provides strength to the bone with a minimum of weight. The pattern of this meshwork is determined by the mechanical functions of the individual bones. Bones of the axial skeleton are mostly spongy bone. The vertebrae, the flat bones of the hips and pelvis, and the ends of the long bones are all spongy bone. Compact bone, on the other hand, appears solid except for microscopic spaces. Compact bone has a layered bone matrix arranged in a manner determined by the distribution of blood vessels that nourish the bone. The bone is traversed by a continuous and complex system of canals that contain the blood vessels and nerves of the bone. The skull, jaw, and shafts of the long bones are examples of compact bone. No sharp

146 Chapter 5 Human Aging boundary may be drawn between the two types of bone; the differences between them depend on the relative amount of solid matter and the size and number of spaces in each. They both contain the same kinds of cells and calcified matrix. The bone cells themselves reside in small cavities (or lacunae) within this matrix. Radiating from each cavity are many narrow channels that interconnect with neighboring lacunae to link all the bone cells together in an intercommunicating network. The matrix is secreted and continuously turned over by the bone cells and consists of two main components, an organic protein phase and an inorganic salt phase. The organic phase consists of collagenous fibers embedded in an amorphous medium that cements them together and that contains various protein–carbohydrate complexes. The fibers are laid down in the gel in a distinct, probably helical, pattern. The inorganic matter of bone consists of submicroscopic salt crystals of calcium and phosphate apatite. These long, slender salt crystals are lined up alongside the collagenous fibers as if each were reinforcing the other—an efficient arrangement for resisting mechanical stresses. The bone mineral content increases during growth and development, reaching a maximum of about 65% of the dry weight of the bone in a healthy adult. In persons afflicted with rickets or other bone pathologies, the bone mineral content may be as low as 35% (Bloom and Fawcett 1968). Age-related Changes. Bone is a dynamic tissue that is constantly being remodeled throughout life. This remodeling results from the two separate processes of changes in both osteoblast/osteoclast activities and in matrix turnover. The first process involves the resorption of the bone in one location and the deposition of new bone elsewhere. Osteoblasts are associated with the formation of new bone tissue and are invariably found on the advancing surfaces of growing bones. Osteoblasts have been viewed as “sophisticated fibroblasts” (Ducy et al. 2000) because of their ability to synthesize the extracellular matrix. Closely associated with the process of bone resorption are morphologically different cells called osteoclasts, which are multinucleated cells de-

rived from the fusion of several cells of the macrophage family (Teitelbaum 2000). Osteoclasts are often found in small depressions on the bone surface, in cavities that arise as a result of their erosion of the bone surrounding the cell. Because the spongy bone accounts for 90% of the total surface area of bones and because this remodeling takes place on the bone surfaces, it is not surprising that remodeling is more widespread in spongy bone than in compact bone. Bone remodeling begins during the fetal period, accelerates to a peak during infancy and childhood, and continues throughout adult life at a much reduced level (Riggs and Melton 1986). The pattern of loss in spongy bone differs in several ways from that in compact bone. In both sexes, bone loss begins at least a decade earlier in spongy bone than in compact bone, and in women, the kinetics and patterns of spongy-bone loss differ considerably from those of compact-bone loss in both the premenopausal and the postmenopausal aspects of the life cycle. The balance between bone synthesis and degradation is variable and usually in flux as a result of the changing balance between the regulatory hormonal, metabolic, and behavioral inputs (Fossel 2004). These changes can favor osteoclast activity over osteoblast activity, and they can also alter the rate of turnover of the extracellular matrix. If the osteoblasts are more active than the osteoclasts and the matrix turnover rate decreases, then these conditions (e.g., high matrix formation plus high matrix accumulation) favor the net accumulation of matrix as seen in bone growth or healing. Conversely, if the osteoclasts are more active than the osteoblasts, then these conditions (e.g., high matrix loss plus low matrix accumulation) favor the net loss of bone matrix as seen in osteoporosis. After age 40, the balance shifts so as to favor resorption, and this major age-related change may lead to an age-related pathology. If we can one day understand how to specifically direct this reversible modulation of cell activity, we might be able to effectively abolish the most important age-related changes that take place in the bone. Recent events have made this goal appear more plausible than before. It turns out that osteoblasts secrete at least two factors

5.6 Age-related Changes in Humans: Detailed Survey

(M-CSF and RANKL) that indirectly control the differentiation of osteoclasts by binding to two specific receptors (c-fms and TNFa respectively) in the osteoclast progenitor cell (Chien and Karsenty 2005). The point of this regulatory interaction is to regulate the bone mass by maintaining a balance between bone synthesis and resorption; defects in the system lead to either osteoporosis (significantly decreased bone mass) or osteopetrosis (pathologically increased bone mass). Osteoblast differentiation, on the other hand, is controlled at the local level by the Wnt signaling pathway, but we do not yet know the details of the signaling process nor indeed how it is controlled. In addition, hormonal pathways provide a systemic level of regulation of bone remodeling. Both parathyroid hormone and estrogen affect bone resorption. The former usually leads to an increase in osteoclast differentiation, possibly by altering the RANKL system (Boyle et al. 2003); although an intermittent application of parathyroid hormone surprisingly favors bone formation instead of bone resorption. This last observation points to the existence of a more complex signaling system than has been anticipated. Estrogen is known to favor bone formation, and both direct and indirect models of how it might do so have been put forward. It might act directly on osteoblasts by either regulating the expression of RANKL, TNFa, and other cytokines; or it might inhibit the occurrence of apoptosis in these cells. However, the level of the estrogen receptors are quite low in bone cells, and this casts some doubt on these being the major effects. Alternatively, estrogen might stimulate the secretion of various bone regulatory molecules by other cells such as T-lymphocytes (Weitzmann et al. 2002) or by pancreatic b cells (Dacquin et al. 2004). Finally, leptin is known to inhibit bone formation, and this observation suggests that bone mass and body weight are co-regulated as well. Aging-related Pathologies. Two bone pathologies, osteoporosis and arthritis, appear to be related to aging. In general, beginning in our 30s, the bone resorbed by the osteoclasts exceeds the amount of new bone synthesized elsewhere by the osteoblasts (figure 5.1). This shift in the bal-

147

ance between resorption and synthesis is the basis for the universality of bone loss with age. The progressive loss of bone matrix, and thus its ability to bind appropriate minerals within bone, during later adult life appears to be responsible for the simultaneous and progressive loss in bone strength. The cross-sectional data of figure 5.2 illustrate the age-dependent changes in bone mineral content, which reaches its maximum value during the 30s and then declines beginning in the 40s. (The statistically significant difference in the values for the left and right ulnae can be understood as the results of differential use of the bones by the predominantly right-handed population.) Neatly paralleling these changes in bone mineral density are the age-dependent changes in bone strength (figure 5.3). Osteoporosis, defined as a decrease in bone mass with no change in the chemical ratio of mineral to protein matrix (Schlenker 1984), is not a disease entity separate from aging but is a more extreme version of the normal processes of bone loss. This bone loss is perhaps the most characteristic age-related change of the skeletal system. Although an intrinsic change, it is nonetheless one that may be successfully modulated by various environmental factors. Since the pathological effects of osteoporosis are associated with loss of strength and subsequent susceptibility to fractures, the more bone mass one has as a young adult, the better off one will be as an aging adult. The simplest way to increase bone mass is to use your body. “Use it or lose it” is one way to summarize the left–right discrepancy of the data in figure 5.2. Exercise, whether that of a professional athlete or that more typical of an ordinary person, is beneficial (Buskirk 1985; see chapter 6). The gonadal hormones play an important role in determining the rate of skeletal maturation. In normal human development, the progress of skeletal growth is intimately related to the developmental state of the reproductive system. This relationship is seen not only in conditions of precocious or delayed sexual development, with their corresponding effects on skeletal maturation, but also in pregnant women and postmenopausal women. The maternal skeleton is to some extent a calcium reserve during pregnancy for

148 Chapter 5 Human Aging

9 Bone formation

T

8

7

6

Bone width (mm)

Figure 5.1 Age-related changes in the amounts of bone formation and resorption in Ohio white females. Curve T is a measure of cumulative cortical bone formation. Curve C is a measure of cortical bone resorption. Curve M is a measure of the amount of bone, the net difference between formation and resorption. (After Garen 1975.)

Amount of bone C

5

4 M Bone resorption

3

2

1

0

10

20

30

40

50

60

70

80

Bone mineral per unit volume of cortical bone (mg/mm3) (mean sem)

Age (years)

1.3

Right ulna

1.2

Left ulna 1.1

1.0

0.9

15–19

20–29

30–39 40–49 Age (years)

50–59

60–69

70–79

Figure 5.2 Age-dependent changes in the mineral density of cortical bone in women. The reduction in mineral density from age 40 on parallels the decrease in average mineral concentration. The values for the right ulna are significantly higher than those for the left ulna, highlighting the probable role of exercise in these predominantly right-handed people. (After Doyle 1969.)

148

5.6 Age-related Changes in Humans: Detailed Survey

149

Compressive strength of the third lumbar vertebra (pounds per square inch)

1,200 Male Female

1,100 1,000 900 800

700 Figure 5.3 Age-related changes in vertebral compressive strength in 137 male and female cadavers. (After Weaver and Chalmers 1966.)

600 500

400

300 200

100

0 10

20

30

40

50

60

70

80

90

Age (years)

calcification of the fetal skeleton and during lactation to replace the calcium lost in the mother’s milk. Such changes normally are very slight, but if they are superimposed on a severe nutritional deficiency or an impaired absorption of calcium, perhaps due to low levels of vitamin D, severe regressive changes leading to pathological fractures may result (Exton-Smith 1985). Diet is clearly an important modulating factor in the development of this age-related change. Gender is important as well, for the sex differences in bone loss are dramatic (figure 5.4). At any given age, bone mass is greater in men than in women. The rate of bone loss, however, is usually higher in women. A male with a 4000-gram skeleton will lose about 450 grams (12%) during

a 30-year period. Over an equivalent time period, a female with a 3000-gram skeleton will lose about 750 grams (25%) of her bone mass, much of it in the years immediately after menopause. These last two factors—hormones and gender—can be particularly synergistic, as in the case of postmenopausal women. Estrogenic hormones tend to protect bone from the stimulating effect of parathyroid hormone on the osteoclasts. When estrogen levels decline at menopause, bone resorption increases because of the increased sensitivity of the osteoclasts to the parathyroid hormone. Quantitative measurements of the amount of bone resorption in postmenopausal women have shown it to be about equal to 425 milligrams of calcium per day, while bone forms

150 Chapter 5 Human Aging

Metacarpal cortical area (mm2)

60

50

40

Men—high-calcium district Men—low-calcium district Women—high-calcium district Women—low-calcium district

30

30

35

40

45

50 55 Age (years)

60

65

70

75

Figure 5.4 The loss of bone from the metacarpals in two populations with different peak adult levels. Calcium intake in the high-calcium group was about 1000 mg in men and 875 mg in women; the comparable values for the low-calcium group were about 450 mg and 400 mg, respectively. Note that both groups lose bone mass at about the same rate, but that at age 75 the population with the highest peak value (the male population from the highcalcium area) had as much bone as the other male population had a its peak. Also note that gender differences overwhelm the dietary differences. (After Matrovic et al. 1979.)

at a rate of only 387 milligrams per day. The result is a net daily loss of 38 milligrams. At this rate, an average postmenopausal female loses 1.5% of her bone per year (Schlenker 1984). Bone loss can be measured by a variety of different methods, each of which yields different numbers, but they all show the same patterns of age-dependent loss (Exton-Smith 1985). Estrogens also affect the levels of calcium absorption and excretion. Thus, estrogens have a dual effect: indirect suppression of remodeling and improved efficiency in the utilization of dietary calcium. The loss of bone strength with age (see figure 5.3) has been attributed to at least two different processes (Whitbourne 1985). An increased porosity arising from the continuous bone remodeling reduces the structural strength of the bone. The remaining bone also becomes more brittle with age—paradoxically via an increased mineralization of the remaining bone tissue. As a result, the bone of an elderly person, when subjected to pressure, is more likely to snap and cause a clean fracture. Such fractures are less likely to heal. The

bone of younger persons is more flexible, because of the higher organic composition of its matrix. When subjected to pressure, it bends and cracks in such a way that a complete break is unlikely. Arthritis is an age-related bone pathology that attacks the joints. Bones are joined to one another by joints, whose smooth functioning is made possible by the strength and elasticity of the tendons and ligaments, by the smooth cartilage lining the opposing surfaces of the joint, and by the synovial fluid that lubricates the whole assembly. The agerelated changes described thus far affect the entire joint system. The tendons and ligaments become less resilient and less able to transmit the forces that act through them as their component collagen and elastin fibers undergo degenerative changes in composition and overall geometry (see the next section of this chapter on muscle tissue for a description). The same processes affect the synovial membranes and make them less flexible. The synovial fluid becomes thinner and less viscous, and its biochemical composition is altered (Balazs 1977).

5.6 Age-related Changes in Humans: Detailed Survey

The cartilage becomes scarred and calcified, thinner and less resilient. All these intrinsic changes act both to decrease the functional efficiency of the joint (Tonna 1977; Whitbourne 1985). The description of osteoarthritis, a degenerative joint disease, greatly resembles the description of normal age-related changes in the joints; the distinction between normal and pathological age-related changes is very difficult to make, particularly since there are more than 100 different conditions that affect the joints and may play a precipatating role (Tonna 1977; Whitbourne 1985). It is instructive to realize that people with arthritis are often suffering from other ailments as well. There is no doubt, however, that osteoarthritis is agerelated, regardless of whether it originates with alterations in the bone or in the cartilage. Some evidence suggests that mechanical abnormalities of the joint, such as those that might arise as a result of injuries, poor posture, or immobilization, are the underlying predisposing factor that alters the course of normal age-related changes toward degenerative joint disease (Frymoyer 1986).

5.6.4 Changes in Muscle Tissue 5.6.4.1 Normal Structure and Function

Our muscles and our bones shape and define us. There are three different types of muscle: skeletal (or voluntary) muscle, cardiac muscle, and smooth (or involuntary) muscle. The present discussion of muscle aging is limited to the skeletal muscles, in part because these constitute most of the muscle mass and in part because much is known about them. The structure of the skeletal muscle is a classic illustration of biological organization and one that clearly shows the interrelationships of one organizational level to another, from the molecular level to the gross anatomical level. Thus, a description of this tissue allows each of us to grasp the interrelationships between structure and function that lie at the heart of modern cell biology. A brief description of the normal muscle structure is also necessary to understand the origin and nature of the age-related changes and what can be done about them.

151

The large individual skeletal muscles (figure 5.5a) are made up of slender, stringy muscle bundles (figure 5.5b). These bundles are associated in various patterns and held together with connective tissue to form the anatomically familiar muscle types such as the biceps or the gastrocnemius (calf muscle). Each bundle can move independently of its neighboring bundle. Each muscle bundle is composed of muscle fibers (figure 5.5c), which are formed from individual muscle cells that have fused together. The nuclei of the muscle cells are found on the outside of the muscle fiber. Because it corresponds to the cellular level, the muscle fiber may be regarded as the basic unit of organization in the muscle. The muscle fiber, in turn, is composed of many myofibrils (figure 5.5d). These myofibrils have a highly regular and periodic structure based on the ordered arrangement of two different families of protein molecules, the actins and the myosins. This ordered molecular arrangement gives rise to the periodic, or striated, appearance of the muscle fiber and underlies the muscle’s ability to contract. Each of these repeating periodic units is called a sarcomere (figure 5.5e). As shown there, the myosin molecules and the actin molecules are spatially situated within the myofilament in an alternating arrangement such that the two interdigitate. The muscle contracts and exerts a force via the tendon on the bone to which it is attached when energy is used to slide the thin actin filaments inward over the overlapping thicker myosin molecules. This process decreases the length of each sarcomere. The muscle is composed of lengthwise repeating sarcomeres, and the muscle as a whole shortens by the cumulative total decrease in length of all the sarcomeres. This contraction is initiated by a nerve impulse delivered by motor nerves to individual muscle fibers. Although all the muscle fibers appear to have the same physical structure, it is possible to sort them into two distinct physiological types depending on their innervation and speed of contraction. The fasttwitch and slow-twitch fibers differ from each other in their inventory of various enzymes; they also differ in their appearance. The flight muscles of birds (the familiar “white meat” of the dinner table) are examples of fast-twitch muscles, while the birds’

152 Chapter 5 Human Aging

Figure 5.5 The organization of skeletal muscle from the gross anatomical level to the molecular level. Parts a through e are pregressive enlargements at the levels indicated. See text for discussion. (After Bloom and Fawcett 1968.)

152

5.6 Age-related Changes in Humans: Detailed Survey

legs and thighs (the “dark meat”) are examples of the slow-twitch muscles. Fast-twitch muscle fibers develop the rapidly accelerating muscle contraction that we normally associate with strength. Slowtwitch fibers are involved primarily in activities such as postural adjustments, which require prolonged and enduring, but not necessarily quick, muscular exertion (Whitbourne 1985). The two fiber types differ on the genetic level as well. Duchenne’s muscular dystrophy is caused by a sex-linked gene and eventually leads to the destruction of the skeletal muscles. The gene defect that causes this condition results in the absence of a particular protein, dystrophin, from at least one population of the fast-twitch muscle fibers, which leads to their subsequent degeneration (Webster et al. 1988). Presumably the functional differences between the different classes of muscle fibers have their origin in gene-controlled structural differences. The muscles exert their effects by contracting and thereby exerting a pulling force on the bone via the interposed ligaments or tendons. These connective-tissue bundles increase efficiency and precision by allowing the muscle to be located at some distance from its site of action, as well as allowing for several muscles to exert different effects on the same bone. The tendons are composed of collagenous connective tissue; the ligaments contain substantial amounts of elastic fibers that allow them to act as a spring and thus permit heavy weights (such as the head of four-footed mammals) to be raised or lowered with very little muscle effort. 5.6.4.2 Age-related Changes

The elderly commonly have less muscle mass than do younger individuals (see chapter 11). This muscle atrophy is thought to be brought about by a decrease in both the number and the size of the muscle fibers (McCarter 1978), although the decrease in size is disputed (M. Brown 1987). Muscle atrophy with age may take place somewhat differently in the fast-twitch and slowtwitch fibers, the former decreasing both in number and in size, the latter only in number (Gutmann and Hanzlikova 1972; Newton and Yernas 1986). The fast-twitch fibers are thought

153

to atrophy because the nerves that innervate them die; the fibers apparently cannot be maintained without innervation. In any event, the decline in specific force exerted by both types of fibers contributes to the age-related decline in mechanical performance (Gonzalez et al. 2001). It was once thought that muscle cells could not be replaced when they die, but we now know there is a low but continuous supply of new muscle satellite (or stem) cells that do supply new muscle fibrils. However, the replacement level seems to be low and can be modulated by growth factors and environmental means in animals (Nnodim 2000), and so we may view the skeletal muscle as being mostly a nondividing tissue in humans. Exercise is well known to stimulate the growth of skeletal muscle, which it brings about mostly by causing an enlargement in the size of existing muscle fibers. In contrast, disuse, malnutrition, or denervation causes the muscle structure to atrophy (McCarter 1978). This remarkable ability of the skeletal muscle to respond to environmental influences makes it difficult to determine the nature of the changes taking place during aging. Decreased exercise and adaptation to a sedentary lifestyle by some people would be expected to alter muscle function significantly. It is difficult to completely rule out such extrinsic factors affecting the aging process. However, some sort of an intrinsic age-dependent atrophy clearly takes place (McCarter 1978), and it involves at least some structural destruction (figure 5.6). There are known changes in skeletal muscle gene expression that take place with aging (Jozsi et al. 2000). Older healthy men show higher expression of genes associated with the response to stress-induced damage both before and after exercise, relative to younger men. These data indicate that the generally decreased response of older muscle to exercise may be the result of prior long-term senescent damage to the gene expression network (see chapters 7 and 14). Another process that can give rise to agedependent muscle atrophy is known as sarcopenia and arises as a result of mitochondrial damage (see table 11.3 and figure 11.4). The deteriorated muscle fibril is replaced initially by connective tissue and eventually by fat (Inokuchi et al. 1975). The

154 Chapter 5 Human Aging

Figure 5.6 A longitudinal section of fibers from the lateral omohyoid muscle of an 18-month-old male rat. Note the degenerating myofibrils alongside apparently normal myofibrils. Also evident are enlarged mitochondria, sarcoplasmic reticulum, and a degenerated neighboring muscle fiber. (From McCarter 1978).

muscle fibers that do not atrophy appear to undergo some major metabolic changes, changes that appear to arise from age-dependent alterations in the neuronal regulation of the muscle fiber (Gutmann and Hanzlikova 1972). One result of these biochemical alterations is that the enzymatic differences between the fast-twitch and the slowtwitch muscle fibers are greatly decreased, and the functional differences in the speed of contraction between the two fiber types are consequently diminished as well (McCarter 1978). The energy metabolism of the muscle fibers also appears to decrease, possibly as a result of degenerative changes taking place in the mitochondria. In addition, there is an age-related decrease in the density of insulinlike growth factor 1 (IGF-1) receptors in mouse muscle and in the ability to activate the signal transduction pathway associated with those receptors (Li et al. 2003). This IGF-1 pathway is a major longevity determinant process (see chapter 7) and if the same changes take place in human muscle also, then these alterations may play a role in the age-related degradation of that system in skeletal muscle (Reynolds et al. 2002). The decreased muscle strength brought about by these changes is complicated by concurrent neural, circulatory, and psychological changes in the aging adult, all of which may nonspecifically cause fur-

ther decrements of muscle functioning. Sarcopenia, whatever its specific origin, is very common and affects 25–50% of older men and women in an agerelated fashion. Low muscle strength is associated with mortality (see figure 3.15), and so preventing sarcopenia would be useful. Alterations in the nerves innervating the muscle likely are an important extrinsic aging factor. One indication that the intrinsic structural atrophy may not be primarily responsible for the decrease in muscular strength observed in aged adults is the remarkable fact that muscles isolated from old individuals and tested under laboratory conditions show no decrease in their dynamic properties when compared with the muscles of younger adults. The dynamic properties are retained despite the presence of marked structural degeneration in the older muscles (McCarter 1978). These observations suggest that extrinsic factors play a major role in the aging of muscle. It has been suggested that exercise-induced hypertrophy of the muscle fibers also brings about a reinnervation of existing muscle fibers that makes up for the fibers that have atrophied (Whitbourne 1985). The beneficial effects of exercise are discussed in chapter 6. Inasmuch as muscle action is transmitted by means of our tendons and ligaments, it is reasonable to conclude that some of the effects of aging on muscle performance may originate in the agerelated alterations that take place in these connective tissues. The elderly are particularly vulnerable to tendon rupture (Shephard 1982), in large part because of the loss of elastic tissue and the alterations in collagen structure that reduce the ability of the tendon to elongate. Following maturity, the collagen of tendons and ligaments undergoes very little turnover and replacement. The basic aging process in these structures consists of alterations in protein structure (see chapter 9) rather than changes in concentration. Aging of collagen is characterized by progressive insolubility in various reagents, increased chemical stabilization, and increased stiffness. These changes are assumed to be due to chemical cross-linking of the collagen fibrils to one another. The mechanisms that underlie these agerelated changes are just now being unraveled.

5.6 Age-related Changes in Humans: Detailed Survey

One line of evidence came from studies of diabetes. The effects of this disease on many organs and tissues are similar to those that often develop at a later age in the normal elderly. For this reason diabetes is sometimes described as accelerated aging. These and other observations led to the idea that the nonenzymatic attachment of glucose to certain proteins in the body gives rise to a series of chemical reactions that result in the formation of irreversible cross-links between adjacent protein molecules, the advanced glycosylation end products (AGE products; see chapter 10). Animal studies have shown that exposing collagen to high concentrations of glucose, either in the test tube or in the body of a diabetic mouse, leads to extensive cross-linking (Cerami et al. 1987). Test tube studies have also shown that drugs such as aminoguanidine can block this cross-linking effect if the protein is incubated with glucose. AGE products also stimulate the macrophages to produce and release tumor necrosis factor and interleukins, two growth factors with diverse effects. These factors then stimulate nearby cells such as fibroblasts to synthesize and secrete collagenase and other extracellular proteases that can digest the cross-linked AGE products (Brenner et al. 1989; Vlassara et al. 1988). Thus there appears to be an in vivo system that remodels the connective tissue. These findings, as well as the observed effectiveness of caloric restriction (see chapter 7), suggest that alleviation of normal aging of tendons and ligaments may not be impossible.

5.6.5 Changes in the Cardiovascular System 5.6.5.1 Normal Structure and Function

In vertebrates, the function of distributing nutrients to the tissues and collecting waste products from them for transport to the appropriate excretory organs is carried out by the cardiovascular system. This system consists of a muscular pump (the heart) and two continuous systems of tubular vessels: the pulmonary circulation and the systemic circulation. The pulmonary circulation

155

carries blood to and from the lungs; the systemic circulation serves all the other tissues and organs of the body. In both of these circulations, the blood pumped from the heart passes progressively through large and small arteries, capillaries, small and large veins, and back to the heart. Nutrients and waste products are exchanged primarily in the capillary network of each circulation. Most organs are also served by a network of capillaries belonging to the lymphatic system, which gathers up the fluids lost by diffusion from the blood during its passage through the capillaries and returns those fluids to the bloodstream. During this return passage, the body fluids pass through a series of lymph nodes in which the immune system can detect foreign invaders in the components of the lymph fluids. Figure 5.7 shows the tissue structure of these various vessels. The simplest vessel is the capillary, composed of a single layer of endothelial cells (a specialized form of epithelial cell). Endothelial cells also line the inside of both arteries and veins as well as the heart, so they provide a continuous lining throughout the cardiovascular system. Capillaries have an average diameter about equal to that of the red blood cell. They form a tubular meshwork in an organ, the tightness of which is determined by the level of metabolic activity characteristic of the organ. Nutrients and waste products are exchanged across and through the capillary walls. The flat, thin endothelial cells are highly adapted to ensure that this process is quick and efficient. Capillaries in some organs, such as the kidney, contain pores that further enhance the process. Capillaries and their supporting cells in the brain appear to actively inhibit the transport of many molecules out of the vascular system into the nervous tissue; this inhibitory function has given rise to the concept of the blood–brain barrier. Arteries and veins, regardless of size, show a common pattern of organization (see figure 5.7a,b). Next to the endothelial lining is an elastic layer composed variously of branching elastic fibers and collagenous fibrils. The middle layer consists of smooth (involuntary) muscle cells, circularly arranged. The outer layer is composed mostly of various connective-tissue elements, longitudinally

156 Chapter 5 Human Aging

(a)

(b) Valve Endothelium Basement membrane Elastic layer Smooth muscle Elastic layer Connective tissue

(c)

Artery

Vein

Artery

Vein Arterioles

Capillaries

Venules

Figure 5.7 Structure of the blood vessels: (a) an artery, (b) a vein, (c) a capillary.

arranged. The structural differences between the arteries and veins reside mostly in the differential contribution of each of these three layers to vessel. In general, the veins have a larger diameter than that of the arteries, but their walls are much thinner. The larger internal volume of the veins allows them to accommodate the same amount of blood as do the arteries, but at a much lower pressure. As a consequence, venous walls can safely develop to be thinner than arterial walls (see figure 5.7a,b). Arterial walls have significant amounts of both the muscular and the elastic layers. The muscular layer allows arteries to contract or dilate and thereby differentially regulate the distribution of blood to individual organs. The elastic layer makes the arterial walls flexible and expansible. Only part of the force of contraction of the heart goes into advancing the column of blood in the vessels; the rest goes into expanding the arterial walls. The elastic recoil of the vessel wall during the interval when the heart is not contracting serves as an auxiliary pump, forcing the column of blood forward even during diastole. This process smoothes out the pulsatile nature of the blood flow, allowing a continuous flow with an intermittent pump (Bloom and Fawcett 1968).

During its relaxation phase (diastole), the heart actively expands, in part because of a mechanical recoil action attributable to the arrangement of connective tissue in the heart. This recoil action creates suction and helps draw blood into the ventricles (Robinson et al. 1986). The heart is the rhythmically contracting muscular portion of the cardiovascular system. It is composed of three main layers, each of which is homologous with the three layers of the blood vessels. These layers are the innermost endocardium; the middle myocardium, or muscular layer; and the outermost epicardium, or connective tissue layer. The cardiac muscle cells are similar in structure, although not identical, to the skeletal muscle cells described earlier. Cardiac muscle differs primarily from skeletal muscle in that it is not under voluntary control. The myocardium is thinnest in the atria and thickest in the left ventricle. These structural differences probably stem directly from the functional differences between the two chambers: The highest pressures in the circulatory system are found in the left ventricle. The muscle bundles are arranged in sheets that wind about the atria and ventricles in complex patterns and are attached to the cardiac skeleton,

5.6 Age-related Changes in Humans: Detailed Survey

a dense connective tissue structure that acts as the central supporting structure of the heart. Connective tissue fibers consisting of collagen and elastin are known to form a netlike structure around each cardiac muscle cell. It has been suggested that this meshwork actively contributes to protecting the muscle cell from either overcontraction or overextension and, in either case, to assisting the cell in moving back to its original or resting state (Robinson et al. 1986). This suggestion implies that the aging processes taking place in connective tissue can have a direct effect on the age-related changes in performance of the heart muscle. The motor impulse that triggers each heartbeat arises autonomously near the top of the right atrium. It is then conducted throughout the heart in a coordinated manner by highly specialized cardiac muscle fibers. The heart is more than just a pump; it is also an endocrine organ. The atria secrete a protein hormone, known as the atrial natriuretic factor, that plays an important role in the regulation of blood pressure, blood volume, and the excretion of water and salts (Cantin and Genest 1986). It is now known that there are three different cell lineages that appear during normal heart development and give rise to the heart. These lineages are distinguished by different types of cell receptors (Chien and Karsenty, 2005). One lineage gives rise to the atrial myocytes, another to the ventricular myocytes, and a third to the cells giving rise to the heart’s conduction system. Lineage specific defects in various genes which guide the development of each of these lineages have been implicated in certain human congenital heart diseases. It may well be that they may play some role in lineage-specific age-related diseases as well. 5.6.5.2 Age-related Changes

Structural Changes. Fewer age-related structural changes take place in the heart than one might naively suppose to be the case, given that cardiovascular disease is a leading cause of death. In fact, most experienced pathologists recognize that age cannot be accurately determined from an inspection of the heart. One of the major changes that can be observed is an increase in coronary artery

157

disease with age. Various studies suggest that about half of the elderly population have symptoms of this cardiovascular disease; thus structural changes in the cardiovascular system must be interpreted with caution, since atherosclerosis is possibly the best example of an age-related disease. I discussed the problem of distinguishing between aging-related changes and diseaserelated changes in chapter 3. Each layer and region of the vascular tree is affected differently by the aging process. The endothelial cells of the intimal (innermost) layer (see figure 5.7) become more irregular in size and shape. It is now believed that a major cause of atherosclerosis may be the stress-induced damage to the endothelial cells that line the entire circulatory system. This focus on the role of stress on the endothelial cell has become central to our improved understanding of aetherogenesis (see chapter 12). The elastic and smooth-muscle layers of the intima increase dramatically with age, by up to 40% in the aorta. In the thoracic aorta, this thickening is due to an increase in the elastic layer; in the abdominal aorta it is due to the proliferation of smooth muscle. Much of the thickening in the elastic layer results from fragmentation, redistribution, and thinning of the elastic fibers. This process is accompanied by increased calcium-binding activity by the elastin. These changes may represent some sort of tissue response to prolonged stress because the magnitude of these alterations is greater in the more highly stressed aorta than in the less stressed pulmonary artery. The relationship of such alterations to the genesis of atherosclerosis is being investigated (Bates and Gangloff 1986). In the heart, the only genuine age-related change observed is the approximately 30% increase in the thickness of the left ventricular wall, an increase that is due to cellular hypertrophy rather than to hyperplasia (Hangartner et al. 1985). Most heart muscle cells do not have the capacity to divide; they are postmitotic cells that must last for the life of the individual. Such stable cells would be expected to display progressive morphological and/or chemical alterations reflecting age-related changes. In fact, one of the best-characterized age-related changes is the

158 Chapter 5 Human Aging increase in cardiac lipofuscin pigment as a function of age (see chapter 11). The role of cardiac stem cells and their role in cell and tissue senescence is discussed in chapter 12. Functional Changes. The importance of not confusing occult disease states with normal agerelated changes is well illustrated by the data of table 5.4 (Shock et al. 1984) and by the information in figure 5.8 (Lakatta 1985). Most of the resting cardiac functions show no age-related changes, save for the systolic blood pressure. As noted in chapter 3, it is both reasonable and plausible to attribute at least some of this change to the stiffening of the arterial walls, as just described. The hypertrophy of the left ventricular wall can then be interpreted as an adaptive response of the heart muscle to this increasing load on the heart. Cardiac output is the product of heart rate and stroke volume and is probably the most important single overall measurement of cardiac performance, for it represents the ability of the heart to meet the oxygen requirements of the entire body. Initially, cardiac output was believed to decline with age (see figure 5.8a); however, populations screened for occult disease show no age-related change in cardiac output (see figure 5.8b). When the same healthy subjects were stressed by exercise, however, their maximum workload did decrease with age (figure 5.9). This decrease appears

to be due to a decrease in the maximum attainable heart rate and thus may represent another true age-related change. This decrease in workload is not thought to be due to a decrease in the maximum contractility of the muscle fiber itself (Lakatta 1985). Animal studies have suggested that the decrease in workload may be the result of age-related alterations in the intrinsic pacemaker activity of the heart and of significant decreases in the concentration of norephinephrine in the heart (Goldberg 1978). Norephinephrine is the principal neurotransmitter of the adrenergic system that innervates the heart. In any event, decrease in maximum heart rate presumably results in a failure of the heart to supply sufficient oxygenated blood to the leg muscles. This cardiovascularinduced muscle fatigue results in a significant reduction in the maximum attainable workload. 5.6.5.3 Age-related Pathologies

One of the more serious age-related cardiovascular pathologies is atherosclerosis. If present, the atherosclerotic process usually begins early in life, progresses during the middle years, and culminates in clinical disease toward the later years of the life span. Age is a nonreversible risk factor for atherosclerosis. Because atherosclerosis is a multifactorial disease, it has proven difficult to sort out the intrinsic age-related changes from the environmental factors that exert their effects over

Table 5.4 Effect of Adult Senescence on Resting Cardiac Function Population

Variable

Institutionalized, Unscreened for occult coronary artery disease (age range = 19–86)

Active in community life, Screened for occult coronary artery disease (age range = 24–79)

Heart rate Stroke volume Stroke-volume output Cardiac output Cardiac index Peripheral vascular resistance Peak systolic blood pressure Diastolic blood pressure

Slight decrease Decrease Decrease Decrease Decrease Increase Increase No effect

No effect No effect No effect No effect No effect No effect Increase No effect

Source: from Shock et al. (1984).

5.6 Age-related Changes in Humans: Detailed Survey

159

(a)

Cardiac index (l/min/m2)

5.5

4.5

3.5

2.5

Figure 5.8 The relationship between age and cardiac index (the volume of blood passing through the heart per unit of time). (a) Males had no apparent circulatory disorders but were recovering in the hospital from other ailments. A strong age-related decline in cardiac index is evident. (After Brandfonbrener et al. 1955.) (b) The Baltimore Longitudinal Study on Aging male and female subjects were known to be free from cardiovascular problems. No agerelated decline is seen in these healthy subjects. (After Rodeheffer et al. 1984.)

1.5 20

30

40

50

60

70

80

(b)

Cardiac index (l/min/m2)

10

8

6

4 Male Female 2 20

30

50 40 Age (years)

a long time. Nonetheless, substantial progress has been made. Bates and Gangloff (1986) offer a more detailed overview and references. On the basis of extensive autopsy observations, Stary (1986) described the progression of atherosclerosis (summarized in table 5.5). Not all

60

70

80

thickenings of the arterial wall are pathological, for certain localized increases in the thickness of the intima (especially at branch points in the vessel) are found in all normal coronary arteries. The diffuse thickening of the intima is the hallmark of this pathology. Such lesions are composed of

160 Chapter 5 Human Aging

Maximum workload achieved (watts)

200 175 150 125 100 75 50

Male Female

20

30

50 40 Age (years)

60

70

80

Figure 5.9 The effect of age on the maximum workload during upright bicycle exercise in Baltimore Longitudinal Study on Aging participants (the same population as in figure 5.8b.) The end point in all instances was muscle fatigue. (After Rodeheffer et al. 1984.)

assemblages of macrophages and smooth-muscle cells embedded in an extracellular matrix containing elastic fibers. These four components of the plaque—macrophages, smooth-muscle cells, extracellular matrix, and elastic fibers—are unevenly distributed into two distinct layers (figure 5.10). In the early stages, the innermost, or luminal, layer is rich in matrix and poor in elastic fibers. It contains a loose and irregular arrange-

ment of the smooth-muscle and macrophage cells. Conversely, the underlying musculoelastic layer is poor in matrix and rich in elastic fibers. In advanced lesions, this layer also contains dense and orderly arrangements of lipid-rich smooth muscle cells and macrophage foam cells layered above a thick extracellular lipid core (figure 5.11). The foam cells are derived from macrophages which have become full of oxidized, low-density lipoproteins. This lipid or necrotic core is composed of the partly degraded lipid droplets of dead macrophage foam cells. Finally, other secondary processes also take place, such as the deposition of calcium in the core and the slow formation of a collagenous cap over the intima region above the core. The result is massive thickening of the intima and narrowing of the arterial channel to critical limits. The mechanisms underlying this process are numerous and complex, but two obvservations stand out. First, the localized increased number of cells in the plaque suggests that an increase in cell proliferation is an important part of the disease process in atherosclerosis (see chapter 12). Second, multiple lines of evidence suggest that the involvement of high blood cholesterol (particularly the oxidatively modified low-density lipoprotein, or LDL, fraction) is required for the initiation and formation of atherosclerosis (Chisholm and Steinberg 2000). These two insights have been woven together into our current

Table 5.5 Development Stages of Artherosclerotic Plaques Lesion Type

Descriptive name of Lesion

Characteristics

I II

—— Fatty streak

III IV

Preatheroma, intermediate lesion Atheroma, atheriosclerotic plaque

V VI VII

Fibroatheroma, complicated lesion Ulcerated fibroatherone fibrous plague

MFC as isolated cells in intima; no extracellular lipid Extracellular lipid layers of MFC; lipid rich SMC; some extracellular lipid All of above plus many small pods of extracellular lipid. All of above plus large confluent extracellular lipid pod replacing much of intima All of above plus collagen cap above core All of above plus ulceration of surface Massive thickening of intimately collagen layers, intra/extracellular lipid negligible

Source: adapted from Stary (1986). Note: MFC, macrophage foam cells; SMC, smooth muscle cells.

5.6 Age-related Changes in Humans: Detailed Survey

Figure 5.10 A type II lesion (see table 5.4) in the coronary artery of a 25-year-old woman who died in an automobile accident. Eccentric thickening contains a submerged fatty streak. (From Stary 1986.)

concept, according to which any one of several factors may injure the endothelial lining of the artery. Blood platelets adhere to the wound site and release platelet-derived growth factor (PDGF). This compound stimulates the smooth muscle cells to proliferate and heal the wound. At this point, the process is still reversible, unless something intervenes to convert this transient injury into a chronic situation with sustained release of PDGF. One of the most important sustaining factors (but not the only one) is the chronic elevation of LDL in the plasma. Animal studies have demonstrated that hyperlipidemic serum will stimulate the proliferation of arterial medial cells grown in vitro. Adding high-density lipoprotein obtained from serum of normal animals to this in vitro system reduces the rate of cell division (Wissler and Vesselinovitch 1986).

161

Figure 5.11 A type IV lesion, or atheroma (see table 5.4), in the coronary artery of a 23-year-old male who died by violence. Note that extracellular lipid is concentrated at the core and replaces much of the musculoelastic intimal layer. Macrophage foam cells and lipid-laden smooth cells are layered above the core. (From Stary 1986.)

Many important aspects remain to be understood, but the basic theme is clear: Atherosclerosis is an age-related pathology that depends on the dynamic interplay of genetic and environmental factors affecting lipid metabolism, cell proliferation, cardiovascular hemodynamics, and other biological, social, and cultural variables. The complexity of the issue may be illustrated by two other observations. First, animal studies have shown that diets containing very high levels of lipids will produce only fatty streaks (type II lesions; see figure 5.10) when fed to prepubertal animals, but will yield coronary atherosclerotic plaques (type IV and V lesions; see figure 5.11) when fed for the same amount of time to young, sexually mature adults (Clarkson et al. 1987). This finding suggests that the sex hormones play an important role in modulating environmental insults.

162 Chapter 5 Human Aging

(a)

270

Serum cholesterol (mg/dl)

250

230

210

190 (20) 170 30 (b)

(67)

(65) 50

40

(31) 60

(36) 70

80

270

250 Serum cholesterol (mg/dl)

The second observation concerns the longitudinal changes in serum cholesterol levels as determined in the Baltimore Longitudinal Study on Aging (BLSA). Figure 5.12 shows that serum cholesterol levels decline significantly as a result of social and cultural dietary changes but that these decreases are superimposed on what appears to be an age-related increase in these levels. Theory predicts that these dietary changes will give rise to a decrease in serum cholesterol levels, and they obviously did—but the observed changes are much greater than the predicted drop. Clearly, other unknown factors are at work here as well. Incidentally, almost all the BLSA subjects shown in figure 5.12 have elevated serum cholesterol levels relative to what is now considered normal (which is about 150–170 milligrams per deciliter). During the same period of time as shown in figure 5.12, the age-related incidence of coronary artery disease began to decline, presumably as a result of long-term changes in diet, cigarette smoking, hypertension detection and treatment, and possibly other factors (Hazzard 1986a). These observations reinforce the perception that such stressors can damage the endothelial cell and set off the cascade of negative effects that can result in atheroma formation (see figure 12.8). Third, it is a common assumption that atherosclerotic lesions once formed are permanent. This assumption is not correct. Significant regression of coronary atherosclerotic lesions has been observed in the Lifestyle Heart Trial (Ornish 1993). In this clinical trial, participants were asked to make conventional (control group) or extraordinary (experimental group) lifestyle changes. The experimental group changes included a low-fat (10% of total calories) vegetarian diet without caloric restriction, no caffeine, limited alcohol, moderate exercise, and stress management techniques. Of the 22 patients in the experimental group, 82% showed regression of lesion size coupled with a significant reduction of symptoms (e.g., angina). Of the 19 patients in the control group, 53% showed a progression of lesion size coupled with an increase in symptoms. There is a suggestion in the data that lesions may regress more easily in women than in men. The fact that

230

210

190 (40) 170 30

(67) 40

(87) 50

(54) 60

(52) 70

80

Age (years)

Figure 5.12 Longitudinal changes in serum cholesterol concentrations. Line segments indicate the mean slope of changes in serum cholesterol for each age decade. Each line is drawn with the midpoint at the mean cholesterol; the length along the x axis represents the mean time span over which the longitudinal data were gathered. (a) Longitudinal change during the period before the drop in cholesterol (1963–1971.) (B) Longitudinal change during the period in which cholesterol levels fell (1969–1977.) The number of subjects for each mean slope is given in parentheses. (After Hershcopf et al. 1982.)

lifestyle changes can bring about a significant reversal of atherosclerotic lesions suggests that this condition cannot be viewed as an inevitable age-related disease, but rather as the outcome of socioculturally related diet and lifestyle habits acting on the background of different genetic or inherent susceptibilities.

5.6 Age-related Changes in Humans: Detailed Survey

5.6.6 Changes in the Respiratory System 5.6.6.1 Normal Structure and Function

Figure 5.13 shows the anatomical structure of the respiratory system. The branching airways make up the main components of the bronchopulmonary system. These airways are as follows: nasal cavity, pharynx, larynx, trachea, primary, secondary, and tertiary bronchi, bronchioles, and alveoli. The bronchi, bronchioles, and alveoli are contained within the lungs. The many series of tubes formed by the branching of the bronchi are known as the respiratory tree. The walls of the trachea and the primary bronchi contain cartilage and smooth muscle. The cartilage gradually disappears as the bronchi branch, so the bronchioles contain only smooth muscle. No muscle surrounds the alveoli. Gases are exchanged between air and blood in the alveoli, thin-walled sacs that are richly supplied with a dense network of cap-

163

illaries. These blood vessels bulge out into the alveolar sacs, thereby presenting a large surface area to the alveolar air. Collagenous and elastic fibers form a tenuous supporting framework for these sacs and capillaries. The alveolar cells are extraordinarily thin and present little obstacle to the free diffusion of gases throughout all parts of the alveoli and into the capillaries. Ventilation of the lungs involves the inspiration and expiration of air. In the normal resting state, inspiration usually lasts about 2 seconds and expiration about 3 seconds. This rhythmicity is under the alternating control of inspiratory and expiratory neurons located in the medulla. Various chemoreceptors monitor the oxygen, carbon dioxide, and hydrogen ion (pH) concentrations and transmit signals to the respiratory center to help regulate respiratory activity. The central chemoreceptor is located in the medulla and responds to the presence of excess CO2 in body

Nasal passage Oral cavity Rings of cartilage

Pharynx Larynx Trachea

Bronchi Bronchioles Ribs

Pleural cavity Diaphragm Figure 5.13 The structure of the human respiratory system.

164 Chapter 5 Human Aging fluids (hypercapnia). The peripheral chemoreceptors are located in the neural tissue of the carotid body and the aortic body and are located at specific points on the arteries of the same name. They respond to reduced O2 concentration in body fluids (hypoxemia). Thus, either hypercapnia or hypoxemia will normally increase the rate and depth of breathing. Inspiration is an active process brought about by contraction of the diaphragm and elevation of the ribs. The increased volume reduces the intrapulmonary pressure to a level below the atmospheric pressure, so air flows into the lungs. Expiration is a passive process brought about by relaxation of the diaphragm and of the rib muscles. Relaxation of these muscles allows the elastic recoil of the chest wall to reduce the volume of the chest cavity. This action, when combined with the elastic recoil of the lungs, increases the intrapulmonary pressure and forces air to flow out of the lungs. Several parameters are important in understanding and measuring lung function; they are defined in table 5.6. In normal individuals, the volume of air in the lungs depends primarily on body size and build. The different volumes and capacities also change with body posture, most of them decreasing in the sitting position and increasing in the standing position. 5.6.6.2 Age-related Changes

The analysis of respiratory function is intimately involved with the functional ability of other systems, such as the cardiovascular or muscular systems and varies as a function of body height. In addition, respiratory function can be readily impaired by disease states such as emphysema or by exposure to environmental pollutants, including those associated with smoking (Klocke 1977). Nonetheless, carefully designed studies make it possible to identify what appear to be the effects of age on respiratory function (Shock 1985). There seem to be very few, if any, changes in agerelated respiratory functions at rest. Marked age effects appear, however, when the respiratory system is made to perform under stress.

Measurement of respiratory volumes is a convenient clinical test. The amount of air that can be forcibly expelled in 1 second (the forced expiratory volume, or FEV) is often used and has been found to possess some predictive value (see figure 3.17). Even in populations screened for occult respiratory pathologies, the FEV decreases strikingly with age. This decrease in FEV occurs even though the total lung capacity is known to remain constant (Klocke 1977); therefore, the residual volume must increase. This increase in residual volume affects particularly the lower part of the lungs, leading to localized changes in ventilation. Another measure of respiratory function is the ventilatory rate, or the minute respiratory volume, defined as the volume of air inspired in a normal breath (the tidal volume) multiplied by the frequency of breaths per minute. This value averages about 6 liters per minute in males and females. At submaximal levels of work, a low ventilatory rate is desirable, suggesting that the lungs do not have to work hard to oxygenate the blood. There are no clear-cut age effects on this parameter at low levels of exertion (Whitbourne 1985). At maximal exertion, however, the ventilatory rate appears to decrease with age. Young adult males can sustain a maximum breathing capacity of about 125–170 liters per minute, but only for about 15 seconds. However, they can maintain a respiratory volume of 100 to 120 liters per minute for prolonged periods of time. This value represents a 20-fold increase over the tidal volume. The maximum breathing capacity decreases to about 75 liters per minute at 85 years, a 50% decline. The elderly have lost their reserve. These age-related changes in respiratory function have been plausibly interpreted as arising from intrinsic age-related changes taking place in the connective tissue components of the lungs. I have described the intrinsic age-related changes in the structure of the elastin and collagen fibers. Such changes occur in the lungs and result in decreased elasticity of the alveoli. The elastic recoil of the lungs is its tendency to resist expansion as it becomes filled with air, and this property helps keep the airways open during expiration. A decrease

5.6 Age-related Changes in Humans: Detailed Survey

165

Table 5.6 Pulmonary Volumes and Capacitiesa Numerical value in normal young adult (ml) Term

Definition

Female

Male

500

500

2100 800

3000 1200

1200

1200

2600

3500

2000

2400

3400

4700

4600

5900

Volume 1. Tidal volume 2. Inspiratory reserve volume 3. Expiratory reserve volume 4. Residual volume

Volume inspired and expired during quiet normal breathing Volume inspired during forced inspiration Volume forcefully expired at end of normal tidal expiration Volume of air remaining in lungs after most forceful expiration

Capacities 5. Inspiratory capacity

6. Functional residual capacity 7. Vital capacity

8. Total lung capacity

aAdapted

Sum of volumes 1 + 2; amount of air a person can breathe in beginning at normal expiratory level and distending lungs to maximum amount Sum of vols. 3 + 4; amount of air remaining in lungs at end of normal expiration Sum of volumes 1 + 2 + 3; maximum amount of air one can expel from lungs after a maximum inspiration plus a maximum expiration Sum of volumes 1 + 2 + 3 + 4; maximum volume to which lungs can be expanded with greatest possible inspiratory effort

from data in Guyton (1966).

in elastic recoil results in a decreased ability of the alveoli and associated structures to remain open during expiration, thereby causing air to be trapped inside them. As the residual volume increases, the FEV decreases. The entire process is accelerated by an increase in the size of the alveolar ducts and a decrease in the number of alveoli, thus decreasing the surface area available for gas exchange (Whitbourne 1985). The same changes in connective tissue also affect the dynamics of the chest wall. The increased rigidity of the ribs and muscles makes it more difficult for the lungs to increase to their full volume. This rigidity increases the amount of work that must be done by the respiratory muscles during the breathing cycle. This increase in the amount of work combines with the loss of elastic recoil of the lungs themselves to further decrease the maximum volume of air that can be brought into the lungs.

As a result of these kinetic alterations, the lungs of older adults are less able to provide sufficient ventilation and gas exchange to meet the body’s oxygen demands at maximum levels of exertion. Smoking adds chemical insult to these intrinsic age-related changes in the respiratory system. We worry a great deal about ways to offset these intrinsic changes, yet many of us tend to overlook the fact that one of the strongest predictors of shortened longevity is smoking. It is an extraordinarily difficult addiction to break, as I can testify, but the benefits of quitting are (eventually) worth the agony. 5.6.6.3 Age-related Pathologies

Emphysema is an obstructive disease in which the lungs lose their ability to ventilate properly, excessive air accumulates in the alveoli, and the supply of freshly oxygenated air to the alveoli is

166 Chapter 5 Human Aging sharply restricted (Spence 1988). This disease is believed to develop as a gradual response to chronic irritation of the bronchial tree by smoke, repeated infections, or another such irritant. The chronic irritation induces the production of excessive amounts of mucus within the airways. The buildup of mucus gradually restricts the airflow to the lungs. The trapped air causes the alveoli to remain inflated at all times, and this condition eventually damages the walls of the alveoli. In the repair process, damaged areas are replaced with fibrous tissues that are much more impermeable to gaseous exchange than healthy tissue is. This fibrous connective tissue also decreases the elastic recoil of the lung, making expiration less efficient and more difficult. People suffering from emphysema consequently have an increased residual volume, a decreased vital capacity, and a decreased minute respiratory volume. The afflicted individuals suffer from “oxygen hunger” and develop hypoxemia and hypercapnia. The disease also places an extra load on the heart as it attempts to compensate for the hypoxemia by pumping more blood to the lungs.

5.6.7 Changes in the Digestive System 5.6.7.1 Normal Structure and Function

The primary function of the digestive system is to transform food into a form that the body’s cells can use and to provide for the uptake of nutrients and the excretion of wastes. It also has secondary endocrine functions, which are intimately involved with its digestive and metabolic processes (Uvnas-Moberg 1989). The digestive system develops as a long tube, topologically outside of the body, with regional specializations for different functions. Components of this system include several highly specialized organs, which may be viewed as belonging either to the alimentary canal or to the accessory organs. Although the digestive tube is regionally specialized, the structure of each region is fundamentally similar and consists of four concentric layers. The innermost layer, the mucosa, is made up of epithelial tissue, an underlying basement mem-

brane, and connective tissue with a thin coating of smooth muscle in some places. The next layer is the submucosa, which is made up of connective tissue and contains nerve fibers and blood vessels. Next is the muscularis, or muscle layer. In most regions the muscularis consists of a double layer of smooth muscle, with the inner layer arranged circularly and the outer layer arranged longitudinally. Coordinated contractions of these two layers produce the peristaltic motion that propels food along the digestive tract. At several points, the circular layer thickens into a heavy band or sphincter, which regulates the passage of food from one portion of the digestive tract to the next. The outermost layer is called the serosa and is composed of connective tissue. Food in the alimentary canal is digested by secretions that enter the tube from accessory glands and/or organs. Most of these secretions are different types of enzymes, but they also include nonenzymatic fluids such as hydrochloric acid in the stomach or emulsifying agents (bile) in the small intestine. A layer of mucus protects the stomach and small intestine from digesting themselves; failure of this mucous layer underlies the formation of ulcers. The small intestine is approximately 6 meters (20 feet) long. Its effective length is much greater, however, because of the great increase in surface area provided by numerous microscopic fingerlike projections, called villi, that line the mucos of the small intestine. The surface area is further enhanced by the presence of microvilli, small hairlike cytoplasmic projections on the surface of each epithelial cell making up the villus. It is estimated that because of these adaptations, the total surface area of the human small intestine is approximately 300 square meters, or about the size of a doubles tennis court. Most digestion takes place in the upper 25 centimeters of the small intestine, known as the duodenum. In addition to enzymes, the small intestine receives from the pancreas an alkaline fluid that neutralizes the stomach acid. This neutralization is essential because the pancreatic enzymes are inactivated by the acid environment of the stomach juices. The remainder of the small intestine is concerned mostly with the absorption of nutrients.

5.6 Age-related Changes in Humans: Detailed Survey

5.6.7.2 Age-related Changes

The cells of the small intestine are exposed to a brew of nutrients mixed with varied metabolic breakdown products, enzymes, acids, microorganisms, and other potentially harmful factors. These components must take a toll on the intestinal cells, for they are continuously replaced throughout life. With a turnover time of about 4 days, the intestinal cells are perhaps the most proliferative of the body’s cells. The stem cells that give rise to this continuously renewing lining of the intestine are located near the base of each of the intestinal villi. The fact that old flies and old mice often die of infectious diseases possibly obtained from the gut raises the possibility that a diminuition of the stem cell’s proliferative ability might result in a weak spot in the gut of older animals, which microbes might be capable of exploiting. Few other true age changes are apparent in the digestive system, once the populations being examined have been screened for the presence of occult gastrointestinal diseases and/or pathological dieting habits such as high levels of alcohol ingestion. Cultural and/or economic factors that may restrict the types of food available to the aging individual are another complicating factor. This topic was reviewed and summarized by Whitbourne (1985) and by Spence (1988); the discussion that follows is drawn from these sources. At least two general changes affect the digestive tract: The muscle contractions become weaker, resulting in a slowing down of the peristaltic motion of the alimentary canal, and the glandular secretions tend to diminish somewhat. These alterations are the result of more fundamental agerelated changes, such as those already described affecting muscle performance (see chapter 11). The decrease in glandular activity leads to regionally specific alterations. The mouth becomes drier as the volume of saliva diminishes. Gastric secretion can diminish by as much as 25% by 60 years of age. In the small intestine, atrophy of the mucosal lining mildly reduces the absorption rate. All four layers of the large intestine atrophy, leading to a weakening of the intestinal wall and a concomitant increase in the incidence of diverticulosis.

167

The most significant of these age-dependent changes in digestive function concern the relationship between vitamin D and calcium absorption. Calcium absorption in the intestine decreases after age 70. Normally calcium is absorbed via two different processes. One process is passive diffusion across the intestinal cell membranes from the high concentrations in the chyme (the liquid contents of the small intestine) to the lower concentrations in the cell. The other process is active transport of calcium, regulated by the blood levels of the active form of vitamin D (calcitriol). Thus, the decreased absorption of calcium might be due directly to reduced serum levels of active vitamin D. In this view, the decreased calcium uptake might be due to paradoxically higher blood levels of calcium as a result of bone resorption, which results in a decreased level of active vitamin D formation and a consequent continued low level of calcium absorption. However, this age-related change (if that’s what it is) can be modulated by proper nutrition and exercise (see chapter 6).

5.6.8 Changes in the Excretory System 5.6.8.1 Normal Structure and Function

Healthy adult kidneys are about 12 centimeters long and 6 centimeters wide. Their weight ranges from 120 to 150 grams and is not appreciably correlated with body size. Internally, the kidney is divided into two distinct regions, the outer cortex and the inner medulla. The striated appearance of the medulla is due to the presence of many arrays of tubules and ducts that are more or less parallel. The renal arteries branch elaborately, giving rise to afferent arterioles that subdivide into a capillary bed, the glomerulus, contained within Bowman’s capsule. Capillary convergence creates efferent arterioles that in turn subdivide into a network of capillaries surrounding the tubules before heading back into the venous circulation. The nephron is the functional part of the kidney responsible for eliminating chemical wastes from the bloodstream and for regulating the salt and water balances of the body. Each kidney is made

168 Chapter 5 Human Aging The urine flows through the ureters to the bladder, where it is stored until excreted via the urethra. The bladder is a hollow muscular organ that can hold about 600 milliliters of fluid. Stretch receptors in the bladder wall initiate muscular contractions that will empty the bladder—usually when the bladder is only about half full.

up of millions of nephrons. As illustrated in figure 5.14, the nephron consists of a very long, regionally specialized and coiled kidney tubule, intimately encircled by a network of capillaries connecting the arteriole and the venule. About 20% of the liquid portion of the blood is filtered out under pressure in the glomerulus and is then selectively reabsorbed throughout the rest of the tubule such that essential components are recovered by the blood, and waste products are retained in the urine. The functional integrity of the nephron is directly related to the maintenance of this normal morphology. In turn, blood pressure is largely dependent on renal function (see chapter 9). The kidney filters an enormous volume of fluid. Approximately 125 milliliters of fluid per minute, or 180 liters per day, is filtered through the glomerulus into each Bowman’s capsule. This filtrate contains digested foods, minerals, and waste products, but no blood cells or large protein molecules. Such components are essential for normal body functioning. Fortunately, about 99% of this fluid is resorbed in the renal tubules, so the body produces only 1–2 liters of urine each day.

5.6.8.2 Age-related Changes

The aging kidney gradually loses mass, most of which occurs in the cortex as a result of intrarenal vascular changes. These vascular changes take place independent of hypertension, although they are aggravated by that condition. The genesis of these changes is not clear, but one hypothesis suggests that they are due to a thickening of the visceral layer of the glomerulus (which is normally tightly opposed to the glomerular capillaries). This condition would then give rise to an impaired retention of proteins in the blood, resulting in proteinuria (Samiy 1983). This proteinuria precedes and might then cause the glomerular sclerosis and

Glomerulus Bowman’s capsule

Figure 5.14 The structure of the nephron and its blood supply. About 20% of the liquid portion of the blood filters from the glomerulus to the tubule system. All regions of these tubules are surrounded by a separate capillary bed, the peritubular capillaries, which reabsorb much of the minerals, nutrients, and fluid from the filtrate. The resulting urine is transported out of the kidney via the collecting tubules.

Nephron

Proximal tubule

Distal tubule Capillaries

Arteriole

Venule Artery

Vein Loop of Henle Vasa recta (capillaries)

Collecting tubule

5.6 Age-related Changes in Humans: Detailed Survey

vessel abnormalities. However, this point is far from settled. One school of thought suggests that the loss of renal function is a disease process initiated by the high protein content of most human diets. The resultant high-solute load delivered to the kidney causes chronic renal vasodilation. This dilation initiates a series of reactions leading to progressive glomerular sclerosis. As glomeruli are destroyed, the solute load in the remaining nephrons is increased, and a disastrous positive feedback cycle is initiated (Dworkin et al. 1984). What is clear is the widespread degeneration of the blood vessels that constitute the glomerulus, a change that effectively removes that nephron from any further role in renal function. Renal blood flow begins to decline by 10% per decade beginning at age 40. The glomerular filtration rate also declines from a steady value of 120 milliliters per minute at age 40 to 60–70 milliliters per minute at age 85. Tubular functions decrease at the same rate as glomerular functions, implying that the nephron fails as a unit (McLachan 1978). Because the kidney has tremendous reserve capacity, age usually has little effect on the body’s ability to maintain normal fluid and electrolyte balances under normal conditions. When the elderly are stressed, however, their ability to cope is not as good as that of younger people, and the decreased number of nephrons begins to be evident in longer response times and a general failure to maintain homeostatic equilibrium. Both cross-sectional (Tobin 1981) and longitudinal (Shock et al. 1979) studies have demonstrated that creatinine clearance (a measure of kidney function) declines with age (see figure 3.6) and appears to be an intrinsic change. The longitudinal study also showed that there is a great deal of normal variability in this parameter between individuals. Some people show a remarkable maintenance or even improvement with age in some cases of their renal function. Interestingly, subjects who died during the study showed a greater rate of decline in renal function during the 10 years preceding death than did survivors. These results bring to mind other data suggesting the existence of individual differences and the heterogeneity of the human

169

population. The genes affecting hypertension mostly affect the kidney and its ability to extract salts from the glomerular filtrate (see figure 9.1). The population variability in renal function and blood pressure likely reflects the interaction of each individual’s environment with their particular genetic constitution.

5.6.9 Changes in the Nervous System 5.6.9.1 Normal Structure and Function

We each base our fundamental knowledge of ourselves as individual personalities on our conscious awareness. The complex actions, thoughts, perceptions, and emotions that constitute this awareness are made possible by the actions of the nervous system. Age-related changes that affect the functioning of the nervous system may well underlie changes in the processes that are most important to each of us. For this reason alone, I am not surprised by the high level of interest shown by both professionals and laypeople in this topic. The structural complexity of the nervous system and the as-yet fully understood processes by which brain activity is transduced into mental processes repay our interest with more questions than answers. The best I can do here is offer a brief and partial summary of the neurological basis for some of these age effects as observed in the central nervous system (CNS) and to allude to some of their possible effects in intellectual functioning. More detailed and comprehensive reviews of the topic may be found in Whitbourne (1985), Finch and Landfield (1985), Duara et al. (1985), Cotman and Holets (1985), Rogers and Bloom (1985), Shock et al. (1984), Haug and Eggers (1991), and Snowdon (2003). Figure 5.15 shows the overall anatomy of the human CNS. The different regions of the brain are involved with different aspects of sensory, motor, and/or mental activities. The human brain is the most complex living structure known to us; it is probably safe to say that it is the most complex structure in our solar system. Much of the brain’s ability to function normally depends on the high level of integration of its various components.

170 Chapter 5 Human Aging

Motor cortex Sensory cortex Corpus callosum

Thalamus

Cerebral cortex

Optic stalk

Eye Amygdala Hypothalamus Reticular formation

Pituitary Hippocampus Pons Cerebellum

Medulla Spinal cord

Figure 5.15 The brain, showing the major structures active in sensing and internal regulation, as well as in the limbic system and brain stem. The eye and optic stalk are shown connected to the hypothalamus, from the lower surface of which the pituitary emerges. (After Bloom et al. 1985.)

The 100 billion or so neurons in the human brain do not function in isolated units the way the cells of the liver or kidney do; the brain is more than just 100 billion isolated nerve cells. Instead, the neurons are linked together in networks of circuits, by means of which informational messages are passed from sending cells to receiving cells. An individual neuron may transmit messages to as few as one or two other neurons or to as many as 1000 or more other neurons. These circuits are complicated and overlapping. A message is transmitted from one end of a cell to the other as an electrical impulse. This impulse cannot jump the gap, or synapse, between individual cells, so it must be converted to a chemical signal at the start of the synapse and then converted back to an electrical impulse at the other side of the synapse before it can continue through the circuit. “Neurotransmitter” is the generic term for any of the diverse chemicals that regulate this intercellular transmission of the message across

the synapse. Many of these different chemical transmitters are found only in particular nervoussystem pathways. Given the proper conditions, a nerve cell is either transmitting an impulse or not; it is either “on” or “off.” It seems incredible that our complex thoughts and mental activities could be reduced to such a simple state. Yet it is instructive to remember that the most subtle and abstract written thoughts of humankind can be completely encoded in the equally simple 1/0 binary system that underlies our digital age. Animal studies are now allowing us to describe learning in molecular biological terms. The gray matter of the brain is composed of the nerve cell bodies; the white matter contains no nerve cell bodies or dendrites and is formed by the myelinated nerve fibers. The gray matter is located on the surface portions of the brain and is known as the cerebral cortex, the cerebellar cortex, and so on. Most of the cells of the CNS

5.6 Age-related Changes in Humans: Detailed Survey

are neuroglial cells and not neurons proper. The neuroglial cells are a diverse group of nonnervous cells that assist in supporting and maintaining the neurons, thereby contributing to their functional ability. 5.6.9.2 Age-related Changes

Gross Structural Changes. Many studies of the changes in brain weight or volume as a function of aging have been undertaken. Some of these studies, particularly the early ones, appear to have been strongly biased in their methods of sample collection and/or measurement (see Gould 1981). The conclusions of such racially or ethnically biased investigations are of course suspect. Nonetheless, several objective studies have established that the human brain loses both weight and volume from youth to old age (see Brody and Vijayashankar 1977 for review). Most studies agree on the general pattern: The brain increases in size from about 357 grams at birth to a peak size of about 1300 grams at age 20 years. This weight is maintained until about 55–65 years of age, which marks the beginning of a progressive decline in weight through the age of 80 years (Davison 1987). The result can be as much as an 11% decrease in mean brain weight, or a 6% decrease in mean brain size, in the elderly relative to the young adult. Several problems are inherent in such studies, not the least of which is the fact that such cross-sectional correlations might be spurious because the mean body weight of people has increased during the past century. Thus, these studies would be confusing age effects and birthcohort effects. However, clinical observations show that mean brain weight and volume change with age, even in clinically normal elderly patients (M. Fossel, personal communication). However, the distribution of this loss over the patient population, and whether certain life histories might predispose one to show brain shrinkage, is not clear. A more important problem is the assumption that the decrease in brain weight or size is due to the loss of neurons, and this assumption was often used to justify dis-

171

criminatory social policies. It turns out that the assumption is not true. Use of modern techniques on fresh brain tissue has shown that there is no decrease in the mean neuron number of the cerebral cortex between the ages of 20 and 110 in either males or females (Haug and Eggers 1991). What does occur in older people is a decrease in the size of the nerve cell body and a decrease in synaptic density in some regions. There may also a decrease in glial cell numbers. Changes of this type are known to accompany cognitive decline in rats (see tables 13.2 and 13.3; figure 13.4). It may well be that it is loss of synaptic activity and complexity, rather than just a simple loss of numbers, that underlies cognitive loss. This supposition is supported by the data obtained from the Nun Study (Snowden 2003). The Nun Study is a longitudinal study of 678 Catholic sisters 75–107 years of age who are members of the School Sisters of Notre Dame congregation. Data collected for this study include early and middle-life risk factors from the convent archives, annual cognitive and physical function evaluations during old age, and postmortem neuropathologic evaluations of the participants’ brains. Analysis of these data revealed a strong inverse relationship between linguistic ability in early life and the risk of mortality in late life (Snowden et al. 2000). It may be that suboptimal neural development might lead to a decreased synaptic complexity, which would manifest as a lower linguistic ability. The simpler connectivity of the neuronal network (see chapter 13) can lead to decreased buffering against future neuronal insults and thus a higher risk of succumbing to a neurodegenerative disease. These facts, plus the data discussed in chapters 13 and 14, suggest that the trajectory we will trace as older adults is to some extent been predetermined by the quality of our neural development when we were young. Which means, of course, that the failure to provide each child with an optimal diet, a challenging environment, and emotional support is likely to result in long term suboptimal neural performance and increased levels of neurodegenerative diseases. We either pay to ensure optimal neu-

172 Chapter 5 Human Aging ral development in early life or we pay to alleviate the symptoms of suboptimal functioning in late life. But we will pay. Microscopic Structural Changes. If the normal functioning of the brain depends on the normal architecture and arrangement of the cells which compose it, then we might expect that unusual cellular- or subcellular-level changes in that structure might give rise to abnormal mental functioning. A neural circuit may be substantively altered by the loss of its component cells and/or by a reduction in the number of other circuits it interacts with, so in this section I examine cell loss and morphological alterations. The loss in overall brain weight and size through the life span was once thought to be due to the loss of individual neurons with age. Earlier studies reported that neurons were lost in selected layers and regions of the aging human cerebral cortex, but not in most brain stem structures. Losses in the number of interneurons in the lower spinal cord have been reported. The most extensive losses, which may range as high as 25–45%, appear to occur in the layers containing the associative neurons of the cerebral cortex. These regions are probably the ones ultimately associated with such mental processes as thought and memory (Duara et al. 1985). These data probably gave rise to the popular folklore that we lose 100,000 neurons every day. Almost all of these older studies agreed that the dramatic decreases in neuronal number occur mostly after the age of 60 years or so. Phrasing this as a daily rate was neither useful nor justifiable. It is even less justifiable now, since recent studies using more sensitive techniques have shown that the number of neurons in the human brain does not appear to decrease with age (Pakkenberg et al. 2003). Indeed, the difference in total neuron number over the age range from 20 to 90 years is less than 10%, which is less than the difference observed between men and women of the same age. Neuronal loss does not seem to happen. Glial cell numbers decline with age, but the difference is not statistically significant. If it is not the absence of neurons that accounts for the nonpathological changes in mental ability, then perhaps it is the complexity of the connections between the neurons. The death

of a neuron can affect, even disable, the functioning of a neuronal circuit; however, so can a decrease in the number of neurons participating in the circuit. The dendrites represent the major route for incoming nerve impulses to reach the neuron; thus, an alteration in their number may reflect an underlying functional change at the level of the synapse. Furthermore, because most synapses terminate on the dendritic spines, an estimate of their number provides us with an index of synapse abundance and distribution. In humans, the number of dendrites, and of the dendritic spines, decreases with age (figure 5.16), although this decrease does not affect all brain regions in the same way. Some areas show no changes; some areas show decreases of up to 25% (Cotman and Holets 1985). In this manner, a functional decrement in our thought process may be brought about by means other than neuronal cell death. These progressive changes transform the delicate, multiply-branched neuron into a simple-looking, almost deformed, cell shorn of its many processes. Such changes are believed, but not yet proven, to lead to neuronal cell death. Dendritic decreases are associated with various dementias of middle and old age. However, the data are not robust enough to permit any strong inference regarding normal aging. Several studies (summarized by Cotman and Holets 1985) show that presumably normal individuals may display either no change or even an increase in the number of their dendritic spines with age. Animal studies have shown that new synapses, indicated by an increase in the number of dendritic spines, can form in the adult brain as a result of either partial denervation or various manipulations of the external environment, or by incorporation of newly formed neruons derived from neural stem cells (see chapter 12). Although such synapse replacement is somewhat slower in aged animals relative to young ones, the older individuals end up with the same number of normal synapses at the end of the process as do young animals. This retained capacity for dendrite growth may compensate to some extent for the agerelated decline in number of neurons. This neuronal plasticity in the adult brain appears to be enhanced if the animals are fed a choline-enriched

5.6 Age-related Changes in Humans: Detailed Survey

(b)

(a)

(d)

(f)

173

(c)

(e)

(g)

(h)

Figure 5.16 A comparison of Golgi-stained neurons from rats and humans. (a–c) Pyramidal cells from the auditory cortex of a 3-month-old rat (a), a 34-month-old rat (b), and a 36-month-old rat (c). (From Vaughn 1977.) (d–f) Summary of the progression of senile changes in human cortical pyramidal cells. (From Scheibel 1978) (g, h) A more detailed view of the rat neurons, showing the loss of the dendritic spines. (From Feldman 1926) The neuron in (g) is from a 3-month-old rat, the one in (h) from a 34-month-old rat.

diet and/or are raised in an enriched environment (an environment in which there are sufficient opportunities for exploration and activity; see Cheal 1986). These findings have implications for human biology and society. More recently, animal studies have shown that in certain defined structures, adult neurons that are lost as a result of cell death may be replaced by the transplantation of certain fetal nerve cells to the affected site. This kind of surgical intervention was explored as a possible clinical

alleviation of Parkinson’s disease in humans. An interesting variant of this approach is the implantation of genetically engineered fibroblast cells into an affected organ (Rosenberg et al. 1988). This experimental approach might allow genetic tailoring of connective tissue cells, altering them so that they can synthesize desirable molecules not usually made by such cells. After transplantation, these altered cells might not only join in the structure of the particular organ, but might also supply necessary molecules to the organ. It

174 Chapter 5 Human Aging has already been shown in animal studies that such grafted cells, genetically engineered to secrete nerve growth factor, survived and exerted a reparative influence on the CNS of surgically damaged rats. The use of stem cells, either embryonic or adult, are now thought to offer a better outcome; the utility of such grafts is being investigated on animal models. The implications of this strategy for gerontology and for society will depend on whether the procedure is clinically successful and whether such success will cause a revision of any laws restricting its use. The progressive accumulation of autofluorescent pigment within the cytoplasm of some long-lived cells is one of the oldest known manifestations of aging in the cell. Such age pigments have been found in many cell types of most eukaryotes and are discussed in more detail in chapter 10. In general, the number of neurons that contain significant amounts of lipofuscin, a yellowish pigment, increases with age. It has been generally assumed that lipofuscin interferes with the normal activity of the neurons and that the progressive accumulation of this pigmented waste product causes the malfunctioning and eventual death of the neuron. As a result, the amount of lipofuscin pigment is often used as a rough indicator of age. This interpretation is still far from being universally accepted, especially since certain neurons in the brain stem normally accumulate large amounts of lipofuscin and other pigments with no apparent deleterious effect (Whitbourne 1985). Changes in the Function of Neurotransmitters. One might suspect that many functional changes in the brain are not heralded by a visible morphological alteration in the neurons. Instead, one could postulate that even small biochemical changes bring about drastic changes in neuronal functioning. Neurotransmitters would be prime candidates as the targets for such masked changes. The neurotransmitters are a heterogeneous group of relatively simple molecules functioning as the chemical messengers that transmit nerve impulses from one neuron to another. In target cells, the neurotransmitters would also activate signal transduction mechanisms, and alterations in these

mechanisms could also contribute to a functional change in the nervous system (Fulop and Seres 1994). Alterations in neurotransmitter metabolism may decrease the functional ability of a neuron while leaving no structural lesion visible to the eye. Individuals who strive to understand just what kinds of changes occur in neurotransmitters with age are faced with an almost impossible task because of the many problems involved: The amounts of neurotransmitter involved are quite small; the quantities are localized at the synapse, which is microscopic; different neurons within the same nerve bundle may use different neurotransmitters; the same neurotransmitter may be used by nerves of very different specificity; different brain regions often show conflicting changes; and the neurotransmitter levels may change dramatically within minutes after death. Nonetheless, progress in this area has been made and has been reviewed by Rogers and Bloom (1985), Carlson (1987), Pedigo (1994), Roth and Joseph (1994), Goldman et al. (1994), and Sugawa et al. (1996). As summarized by Whitbourne (1985), the most striking age-related reductions in neurotransmitter activity are seen in the hippocampus for acetylcholine, in the substantia nigra and the striatal pathway for dopamine, in portions of the brain stem for norepinephrine and serotonin, and in the thalamus for gamma-aminobutyric acid (GABA). These changes in neurotransmitter levels are often paralleled by changes in both the number of receptors (the membrane proteins to which the neurotransmitters associate; Pedigo 1994) and their binding affinity (Petkov et al. 1988). Motor deficits in experimental animals are closely correlated with loss of striatal dopamine receptors, believed to result from reduced rates of receptor synthesis (Roth and Joseph 1994). The regulation of receptor mRNA synthesis may not be entirely autonomous because it can be manipulated by diet, exercise, or hormone treatment (Roth and Joseph 1994). A decrease in neurotransmitter receptor density might lead to diminished prejunctional neurotransmitter reuptake, inhibition, or facilitation. Such a situation might result in only minor aging-related changes in neurotransmission at rest but yield a significantly diminished range of modulation of the

5.6 Age-related Changes in Humans: Detailed Survey

nerve signal due to aging when the organism is stimulated (Docherty 1996). The activity of the neurotransmitter-synthesizing enzyme, choline acetyltransferase, is regulated by nerve growth factor (NGF; Kerwin et al. 1993). Both animal and human studies indicate that the loss of cholinergic neuronal activity in the hippocampus is the earliest age-related deficiency observed, decreasing after the fourth decade (Araki et al. 1993; Kerwin et al. 1993). This suggests the possibility that regional changes in NGF receptor density may serve as an indirect level of neurotransmitter synthesis control. The cholinergic system is involved in neural systems that control memory and learning. Loss of cholinergic activity may have widespread effects on mental functioning. Work in rodents has indicated that transplants of cholinergic-rich fetal neural tissue can sometimes alleviate the behavioral effects associated with age-related cholinergic decline (Ridley and Baker 1993). More useful, perhaps, is the demonstration that adenoviral vectors are capable of transferring neurotransmitter receptor proteins to the brain (Ikari et al. 1995). The possible use of targeted gene therapy to reverse deficiencies in specific neurodegenerative disorders will command great interest in the near future. The important conclusion from all these studies is that normal aging does not bring with it a universal decrease in neurotransmitter activity. However, the localized and specific decreases that we see appear to play an important role in the development of age-related changes in behaviors. Changes in Cerebral Metabolism. Another approach to identifying age-related functional changes in the brain is to monitor changes in the energy metabolism of the aging brain, the idea being that alterations in brain output might reflect changes in the amount of energy available to run the brain. This topic was reviewed by Duara et al. (1985). In humans, glucose is the main substrate for energy metabolism in adults eating a normal diet. Almost all the glucose taken up by the brain is oxidized and used for the production of ATP (adenosine triphosphate). The production of this energy-rich molecule is pro-

175

portional to oxygen consumption and therefore to the rate of cerebral blood flow. One can safely use radioactive glucose or measure the magnetic spin of the phosphate molecule or measure cerebral blood flow to estimate the levels of glucose and ATP within the living brain. Thus one can noninvasively measure any of these three parameters and obtain an estimate of the brain’s overall metabolic activity. At least 100 noninvasive studies have been done on the age-related metabolic changes in the human brain. These studies are consistent in showing that brain metabolism never increases with age: it either remains the same or it decreases. In subjects without covert disease, aging is not correlated with decreased cerebral blood flow (CBF) while at rest (Atkinson et al. 1992). However, mental stimulation revealed significant age-related decreases in regional CBF and in regional glucose consumption, a correlation suggesting that regional hypometabolism may underlie the age-related increases in cognitive dysfunction seen in some people (Baron and Marchal 1992; Eberling et al. 1995; Grady 1996; Grady et al. 1994). Some support for this hypothesis is given by the rough correlation observed between the extent of cortical hypometabolism in an individual and whether that individual displays normal, subnormal, or severe levels of cognitive defects (Baron and Marchal 1992). Animal studies with the senescence-accelerated mouse (SAM) show that the early impairment of memory in these animals is closely correlated with the decrease in cerebral glucose metabolism but not with other metabolic indicators (Fujibayashi et al. 1994). In addition, there is some indication that characteristic regional hypometabolic patterns are associated with specific neural diseases as well as with healthy or normal aging (Moeller et al. 1996). The regional metabolic declines may be the result of neuronal cell death (Meyer et al. 1994), degradative changes in the extracellular matrix of brain microvessels (i.e., the blood–brain barrier; Robert et al. 1997), and or systemic changes in the cerebral blood supply as measured at the carotid arteries. Animals with few observable age-related neuropathologies, such as rats, exhibit little decline in brain metabolism, whereas animals with extensive age-related

176 Chapter 5 Human Aging neuropathologies, such as humans, exhibit significant declines in brain energy metabolic indices. The studies discussed above are beginning to shed some light on the relationship of these two variables. Changes in Mental Abilities. The anatomical and physiological evidence presented so far indicates that our neural circuitry changes and our rate of cerebral metabolism decreases as we grow older. Are these age-dependent decrements in structure and function correlated with age-dependent alterations of mental activity? One standard measure of mental activity in our society has long been the intelligence test. Intelligence is what scientists term a “soft” or “fuzzy” concept. We are not sure what it is, but we are sure that we know how to measure it. This attitude has engendered much unfairness and bad science in the past, as has been ably recounted by Stephen Jay Gould in his 1981 book The Mismeasure of Man. The now discarded stereotype that senility is an invariant accomplice of aging was based in large part on the misinterpretation of cross-sectional data. The BLSA studies have shown the existence of a continual age-related quantitative decline in tests of memory and of decision performance (Shock et al. 1984). This is depressing, but it doesn’t fully support the old stereotype, for the data also show that certain of the age differences (those having to do with learning new material) are strikingly lessened (but not abolished) if the older individuals are allowed to learn the test material at a slower pace. This observation suggests that the earlier cross-sectional results may have confounded intelligence with speed of test taking. Thus, the adoption of appropriate learning strategies by the elderly may substantially compensate for and alleviate any real decreases in learning performance. This effect may be particularly true in situations where a knowledge and appreciation of past events play a significant role in one’s real-life performance. Wisdom and experience count for something. Perhaps Francis Bacon summed it up best when he wrote, “Young men are fitter to invent than to judge; fitter for execution than for counsel; and fitter for new projects than for settled business” (see Bartlett, 1919).

We should be careful not to swing to the other extreme: maintaining that aging brings with it no decrement in mental function and that faculties such as memory, learning, and reasoning persist undiminished until late in life. In fact, people’s mental faculties tend to diminish significantly shortly before death (Bank and Jarvik 1979). As we grow older, we experience a decrement in short-term memory ability. The anatomical structures involved in memory are known (Mishkin and Appenzeller 1987). Oxidative stress-induced damage to these structures in individual animals is highly correlated with impaired mental function (see figure 13.4). There is certainly a correlation in humans between cognitive and physical performance (Tabbarah et al. 2002) and between intellectual disability and life expectancy of older people (Bittles et al. 2002). The biological basis of memory is not yet known in detail, but it seems certain that here, too, the adoption of mnemonic aids can help. The important thing is that we remember; how we remember is not the issue. The loss of memory must be the saddest blow. The brain cannot function normally without a constant influx of sensory perceptions—perceptions that form our view of the world. It is now common knowledge that strong people in the prime of life can be broken and brainwashed by the manipulation of their sensory environment. If sensory deprivation can wreak such havoc on younger individuals, then it is fair to ask what sorts of functional decrements affect our sensory organs with age and what effects such decrements might have on our normal mental functional ability. Whitbourne (1985) reviewed this question in some detail, and much of the following discussion is drawn from that source. Of the five traditional senses, only taste appears not to be affected by any well-characterized intrinsic age-related changes. This observation is consistent with the anatomical evidence, which shows no age-related decrease in the number or distribution of taste buds and receptors on the tongue. Although cross-sectional data show an age-related decline in salt and sweet taste detection, these results were probably confounded by other factors, such as smoking and denture use.

5.6 Age-related Changes in Humans: Detailed Survey

In any event, even this possible decrement is far below the taste thresholds found in normal life. There is, however, some data suggesting that poor nutrition, especially of vitamins and minerals, may lead to taste changes, which may then reinforce the poor nutrition (Chauhan et al. 1987). Our sense of touch undergoes mixed changes. Sensitivity to touch on the skin of the hand decreases, even though this region is normally much more sensitive than the rest of the body surface. However, touch sensitivity decreases nowhere else on the body. These observations are consistent with the anatomical evidence, which shows that three out of the five different types of touch receptors undergo little if any age-related changes. Only the Pacinian corpuscles and Meissner’s corpuscles seem to display structural and numerical changes with age. The ability to detect pain, an adaptive response of great value in the aged, is so confounded with extraneous cultural and personality factors that it is not possible to draw firm conclusions regarding the aches and pains of old age. Smell is a chemical sense akin to taste. It has both a practical and an aesthetic role in informing us of the dangers and the pleasures of our environment and can, in some of us, trigger the release of specific memories. The way in which odors are transformed into neural signals by the smell receptors is not understood, except that this transformation must result from the interaction of the olfactory receptor molecule with the molecule responsible for that odor. We do not understand how to classify odors, nor do we agree on how to measure them. Nonetheless, the crosssectional data appear to support the view that both the ability to detect odors and the ability to identify them experience an age-related decline from age 60 onward. There is a gender difference in these abilities, males scoring poorer than females in both categories. The data from 712,000 respondents enrolled in the National Geographic Smell Survey (Corwin et al. 1997) suggest that environmental influences, such as factory work, adversely affected both men and women. The effects of age, sex, and exposure to noxious events or agents appear to interact so as to produce a progressive olfactory deficit.

177

Cross-sectional data indicate that there is an age-related loss in hearing, which is especially pronounced at high frequencies. Superimposed on this loss is a gender difference in which men have poorer hearing at moderate and high frequencies than do women. Some or all of this gender difference may be due to the greater environmental noise that men used to be exposed to in their traditional occupations. Hearing is vulnerable to environmental insults. Thus, we may expect to encounter an increased number of very deaf rock guitarists (and perhaps concert goers) of both sexes in the decades to come. The anatomical site of this decreased sensitivity to high-frequency tones (termed presbycusis) lies in the inner ear. At least four different mechanisms (involving alterations in the sound receptors, their neurons, their blood supply, and the basilar membrane) are capable of producing presbycusis. Agerelated anatomical changes also occur in the middle ear, but these do not appear to result in a hearing decrement. We are visual creatures. Most of us probably dread the loss of vision more than any of our other senses. Our language reflects this bias: When we understand something, for example, how often do we say, “I see”? Our eyesight is a critical feature of adaptation to the environment, for vision and memory combine to let us know how people and things are arranged in space and time. The eye makes vision possible by using the cornea, lens, and iris to focus a controlled amount of light onto the receptor cells in the retina. True agerelated changes occur in various parts of the eye, the net effect of which is to reduce visual acuity and color discrimination and to affect sensitivity to light. The cornea undergoes several age-dependent changes, the most important of which is its decreased curvature and thus decreased refractive power. These decrements can be a real problem, since the cornea is responsible for most of the refraction and focusing of the light entering the eye. The lens is the other structure involved in refraction. The lens capsule is elastic, and its shape can be altered from a spherical form to a more flattened form by the action of the ciliary muscle. This change in shape allows us to focus the light

178 Chapter 5 Human Aging rays onto the appropriate region of the retina. The lens itself never stops growing, so it becomes thicker and stiffer with age. In fact, the mass of the lens triples from its original value by the age of 70. The greater density and stiffness of the lens make it more difficult for it to change shape. As a result, its refractive power drops considerably with age. By the age of 60, the lens is incapable of accommodating to focus on objects at close distance. So we hold the newspaper farther and farther away from our eyes—until we wear the glasses we so obviously need. As the lens increases in thickness and density, it also becomes less transparent and thus transmits less light into the eye. This decrease is not uniform across the spectrum; it is aggravated by the increasingly yellowish tinge of the lens to selectively absorb the blue and violet portion of the spectrum. The result is impaired ability to discriminate colors and impaired night vision. This reduced transmission of light by the lens is further aggravated by the decreased ability of the iris to open to its widest, as a result of the atrophy of the iris muscles. The effect of this atrophy is to reduce the maximum size of the iris and hence to decrease the amount of light that can enter the eye. These changes in our sensory functions appear to be true age-related changes. Most of them we can compensate for to some extent, either by prosthetic devices (glasses, hearing aids, and so on) or by adaptive changes in behavior. The brain regulates and integrates all the activities of the body. Even small and localized changes in neural structure or function have the potential to bring about far-reaching changes in the coordinated functioning of the body, should they affect the appropriate regions of the CNS. The deleterious effects on our thought processes that flow out of the localized death of certain cholinergic neurons that are characteristic of senile dementia of the Alzheimer’s type (SDAT) are proof of this statement. These integrative activities of the brain are most easily observed in the neural regulation of endocrine functions and of immune functions. In recent years, it has become clear that the nervous, endocrine, and immune systems are functionally integrated. Rather

than viewing them as three separate physiological systems, it now appears to make more sense to view them as three overlapping components of the body’s intercellular communication system. The link between the nervous and endocrine systems has been long known. The hypothalamus–pituitary axis constitutes the principal interface between the two systems (figure 5.17). Together the two are usually referred to as the neuroendocrine system; this system controls a host of vital body functions. Thermoregulation is a good example. Body temperature is monitored by external thermoreceptors in the skin, as well as by internal thermoreceptors in the hypothalamus. The latter measure the temperature of the blood passing through the region. When a drop in blood temperature is sensed, peripheral autonomic nerves act to constrict the skin capillaries and shunt blood toward the core, to erect fur or feathers to trap a layer of warm air next to the skin ( “goose bumps”), and to induce shivering to generate heat. When the blood temperature exceeds the body’s physiological set point, these heat gain mechanisms are inhibited and heat loss mechanisms (the reverse of the heat gain processes, plus evaporative cooling by sweating) are activated. Although the number of sweat glands is not noticeably reduced in the elderly, older people require a much longer time for the onset of sweating under conditions of moderate exercise than do students of college age (Finch and Landfield 1985). This impaired adaptive response of the heat loss mechanisms results in an increased heat load in the elderly and consequent increased morbidity and mortality. These and other similar observations (Whitbourne 1985) suggest that the basic defect in thermoregulation resides in the delayed response of the peripheral structures (sweat glands, capillaries, and so on) to CNS signals and not in the neuroendocrine control mechanisms themselves. However, Shock (1977, 1985) has made the reverse argument, suggesting that the impaired adaptive response is due to a narrowing of the range within which the neuroendocrine control mechanisms can work effectively. The question as to the site of the defect(s) is still unresolved. In either event, the net effect is that the body’s

Hippocampus –

Hypothalamus + –

–

Nerves of the central and autonomic nervous systems

Thyroid Gonads Adrenal cortex Glucocorticoids

Thymus Other immunesystem centers

Pituitary

– +

Thymosin α1

–

–

+ –

Endocrine system

Other brain centers

Sense organs

Immune system

Light Sound Smell Taste Touch

179

Nervous system

5.6 Age-related Changes in Humans: Detailed Survey

Interleukin-1

Figure 5.17 A schematic representation of the intercellular communication system. The hypothalamus–pituitary axis defines the neuroendocrine system, a key regulatory system composed of both neural and endocrine components. In a similar manner, the nervous system also overlaps with the immune system. The glucocorticoid pathway illustrates the role of negative feedback (–) on the nervous and endocrine systems. The thymosin a1 pathway illustrates the positive (+) and negative feedback effects of the immune system on the nervous and endocrine systems. Via these communication circuits, external stimuli, interpreted as stress by the brain, result in the pathophysiological effects of glucocorticoids, such as increased incidence of cancer, metabolic disorders, and so on.

integrative and homeostatic mechanisms are impaired. The aging individual is less able to handle the stresses of daily living and is more likely to succumb to an environmental stress that he or she could easily have adapted to in the past. The nervous system and the immune system affect each other. The nervous system appears to communicate with the immune system directly via neuronal connections to the immune-related structures, such as the bone marrow and the thymus, and indirectly via the endocrine and neuroendocrine system. The most important aspect of biological communication, feedback regulation, has been demonstrated among these several systems. A good example of feedback regulation is shown in figure 5.17, which depicts the interactions between glucocorticoids (a product of the adrenal cortex) and thymosin alpha-1 (a peptide hormone secreted by the thymus gland). Glucocorticoid production is under the proximate control of the hypothalamus–pituitary axis, which is under negative feedback control by the hippocampus (a part of the limbic system). The gluco-

corticoids have varied effects on other parts of the body, not the least of which is suppression of particular immune functions, especially after prolonged exposure to glucocorticoids secreted by the adrenal gland (see figure 13.4). The many observations showing that increased or prolonged stress increases susceptibility to certain diseases, such as cancer, may have their physiological basis in this immunosuppressive activity. The high glucocorticoid levels associated with prolonged stress also kill many of the regulatory neurons in the hippocampus, thereby reducing the effectiveness of the CNS control of glucocorticoid levels. Aging rats show a similar loss of neurons, and such impaired regulatory mechanisms may underlie the observed higher incidence of cancers in the elderly. The thymic hormone thymosin alpha-1 stimulates the immune system. This hormone is found in the hypothalamus, but with a circadian rhythm inverse to that of glucocorticoids. Thymosin alpha-1 stimulates glucocorticoid production, probably by its stimulation of hypothalamic-releasing

180 Chapter 5 Human Aging factors, and thymosin alpha-1 is probably inhibited by glucocorticoids. Thymosin alpha-1 completes the feedback cycle and ties all three systems together. Furthermore, thymosin stimulates the production of interleukin-1 by macrophages (white blood cells). Interleukin-1 is identical to the molecule used by the hypothalamus to regulate body temperature. Macrophages produce and secrete interleukin-1 only when actively engaged in fighting an infection. This signal induces the hypothalamus to increase the body temperature. The resulting fever decreases the viability of invading bacteria and hence contributes to recovery. Data are still being gathered, but it seems possible that some of the deleterious degenerative diseases and pathologies that characteristically affect the elderly have their origin in an upset of a particular aspect of this intercellular communication system. This concept is explored in detail in chapters 13 and 14. 5.6.9.3 Aging-Related Pathologies

Alzheimer’s Disease. Individual neurons also exhibit other sorts of age-related changes, the etiology and significance of which are still being deciphered. Neurons normally contain very slender processes, known as neurofilaments. These neurofilaments are intimately involved with the internal transport of neurotransmitter molecules from their site of synthesis in the cell body to their site of usage at the synapse. Some older people display changes in these processes such that the slender filaments become transformed into thickened and twisted black fibrils of much greater prominence within the cell, these fibrils are called neurofibrillary tangles. In intellectually normal old subjects, clusters of such affected neurons may be found at certain specific sites (the anterior temporal lobe, for example), although they may be very rare in other brain regions (such as the neocortex). Persons suffering from senile dementia may have a greater number of such neurofibrillary tangles in the anterior temporal lobe than do normal individuals. This phenomenon has been studied extensively in the case of Alzheimer’s disease. Alzheimer’s disease (AD), also known as senile dementia of the Alzheimer’s type (SDAT),

is a degenerative disorder of the CNS that results in a progressive loss of memory and other intellectual functions of sufficient severity to interfere significantly with normal activities of daily living and social relationships. It is differentiated from other benign age-associated episodes of forgetfulness by its inevitable, progressive, and irreversible declines in memory, time and space orientation, performance of routine tasks, language and communication skills, abstract thinking, learning ability, and, finally, personality changes and impairment of judgment (Katchaturian and Radebaugh 1996a). Age, genetics, culture, and socioeconomic status are known risk factors for AD. This disease is a major health problem and promises to become worse in the future. The overall incidence of the disease is about 600 per 100,000 population or, to put it more realistically, about 5–10 percent of the people over the age of 65 (Evans 1996). The prevalence of AD increases with age, the percentage affected doubling for every decade that people live beyond the age of 65. Thus, a study in East Boston determined a prevalence of 0.6% for people 65–69 years old, 1.0% for persons 70–74 years old, 2.0% for persons 75–79 years old, 3.3% for persons 80–84 years old, and 8.4% for persons 85 years of age and older (Evans 1996). This strong association of age with AD suggests two possibilities: first, that there might be an intimate coupling of the disease with an underlying process of aging, possibly involving the age-related increase of damage processes coupled with the age-related decreased efficiency of repair mechanisms (Martin et al. 1995); and second, that the disease might be dependent on some slow, time-dependent processes that take a long time to reach a critical threshold. In the latter case, of course, AD might not fit the strict definition of an age-related change. Recall the description of polycystic kidney disease in chapter 3, a time-dependent disease in which the longterm hyperplasia has no ill effects until a certain threshold is reached after about 50 years of slow overgrowth. Should AD actually be a similar sort of time-dependent process, then this would imply that individuals fated to develop AD should have characteristic abnormalities in their brains at an early, presymptomatic age. Data from the

5.6 Age-related Changes in Humans: Detailed Survey

Nun’s Study (Snowden 1997) suggests that an important early predictor of AD may be lower levels of mental activity at young ages. The adage to “use it or lose it” applies with especial vigor to our minds. Estimates suggest that more than 100,000 Americans die of AD every year. In 1980, approximately 2.9 million persons had the disease; in 1996, about 4 million persons were believed to have the disease; and it is projected that by 2050, approximately 10 million persons will be afflicted with AD. However, this projection assumes that current conditions will not change, which is likely not to be the case. Still, the projection provides us with some numbers to discuss. Most of the increase in AD prevalence is projected to occur in the over-85 age group. The prevalence in the younger age groups will likely either stay constant or increase only moderately (Evans 1996) in the absence of any effective interventions. Clearly, the increase in the senescent phase of our lives and the consequent changing age structure of the population do not come without a price, as discussed in chapter 15. A natural history of AD indicates that the average patient first exhibits symptoms at about 72 years of age, is diagnosed at about 75, is institutionalized at about 77, and dies at about 81. The disease appears to progress faster in males (84 months) compared to females (108 months); the sex difference is most notable in the last, or institutional, phase (25 versus 52 months); however, this sex difference is not seen in all studies (Jost and Grossberg 1995). Autopsies have shown that the brains of people who die with AD are abnormal (Gearing et al. 1995). At a gross level, there is an atrophy of the neocortex and often of the hippocampus and amygdala as well—all key sites involved in thinking and memory. In addition, the ventricular system is often enlarged, contributing to the atrophy. Finally, melanin-pigmented neurons are often much paler than normal. At a microscopic level, there are three more or less characteristic neuropathologies: neuritic plaques, neurofibrillarytangles, and amyloid angiopathy (Wolf et al., 1999). Neural, or senile, plaques are found in both intellectually normal individuals and people af-

181

fected by the various dementias. Two major subtypes of these plaques are recognized: neuritic and diffuse. The neuritic plaques, which are the ones strongly associated with AD, are spherical structures about 80 mm in diameter, with thickened neuronal processes (or neurites) surrounding a central fibrillar core of an abnormal protein called amyloid (discussed below). In additon, these neuritic plaques contain dense bodies thought to be the remains of lysosomes, mitochondria, and paired helical filaments. The diffuse plaques, in contrast, lack the abnormal neurites, appear more amorphous, and contain little if any fibrillar amyloid (they do contain amyloid, but in a diffuse form). Diffuse plaques are often found in brains of cognitively normal individuals, as well as in brains of AD individuals. They may represent an early stage in the development of neuritic plaques; Selkoe (1997) has suggested that they represent very early lesions that may or may not progress to mature, symptom-producing lesions, depending on many factors, including longevity. In individuals who suffer from AD, the number of neurofibrillary tangles (NFTs) in the affected brain areas is approximately six times greater than the number present in the most severely affected intellectually normal person. NFTs are found mostly in the hippocampus and the cerebral cortex; they have not been found in the cerebellum or spinal cord. The tangle is intracellular and is composed mostly of paired helical filaments (PHFs), which, as their name suggests, are protein filaments helically twisted about each other in pairs. These PHF proteins are abnormally phosphorylated forms of the tau protein (which is usually associated with microtubules) and are complexed with another protein, ubiquitin, which is normally used by the cell to label proteins destined for degradation (Mori et al. 1987). The PHFs are resistant to degradation in these patients. Note that although NFTs invariably are found in AD patients, they can be found at a lower density in cognitively normal individuals, as well as in the brains of patients suffering from other neurological disorders. We can see these plaques and tangles, but perhaps what we cannot see is even more important. We cannot see the loss of synapses and the

182 Chapter 5 Human Aging death of neurons. The breaking of neuronal circuits and the isolation of important brain centers from their normal partners cannot be directly seen, although we can clearly observe the results of these losses in the transformed personality and diminished behaviors of our loved ones. The third neuropathological feature of AD is amyloid angiopathy, or the deposition of amyloid protein within the walls of the blood vessels of the meninges and cortex. The severity of the deposition varies widely, and the same feature is sometimes found in the brains of older normal individuals. What is the amyloid protein and what role does it play in the genesis of AD? Furthermore, what role do our genes play in the genesis of this disease? The answers to these two questions are intertwined. There are both inherited and noninherited forms of AD. Families have been identified in which the incidence is very high, affecting members of four or five generations (Wurtman 1985). The inheritance pattern is consistent with the idea that the defective aging is transmitted as an autosomal dominant. Thus individuals need inherit only one copy of the aberrant gene to develop full-blown AD. Estimates are that between 40 and 75% of AD patients suffer from some sort of a genetically transmitted form of this disease. The remainder of the affected individuals display a milder, nonfamilial form that tends to become apparent later in life; its etiology is not yet clear. For reasons of convenience and accessibility, most of the research attention has focused on the inherited forms of AD, albeit with the understanding that the two forms of the disease may share some or even many common mechanisms.

Four genetic alterations underlying familial AD are now known; they are listed in table 5.7. Note that all four mutations result in the increased production of amyloid protein or its variants. I return to the genetics in a moment but first describe the amyloid protein. “Amyloid” is a generic term that describes proteins with a beta-pleated sheet structure. The amyloid proteins involved in AD are derived mostly from the APP (amyloid precursor protein) gene located on chromosome 21. The APP gene is a large gene (approximately 400 kilobases of DNA) that is alternatively spliced to yield several transcripts that code for a family of amyloid (Ab) proteins ranging in size from 695 to 779 amino acids (Sandbrink et al. 1996). The APP gene is expressed ubiquitously in mammals, by both neural and nonneural cells. It is highly conserved in vertebrates, and homologous proteins have been identified in the fruit fly Drosophila and the nematode Caenorhabditis elegans. In all these cases, the APP protein is a transmembrane protein apparently involved in cell–cell signaling processes. Significant amounts of newly synthesized APP protein appear at the cell surface; some of these molecules may be cleaved at particular positions by (unknown) proteases. It is not the APP protein itself that seems to cause AD, but rather certain of these cleaved fragments. The cleavage normally yields a 40-amino-acid peptide (Ab1-40); this fragment appears to play no role in the pathogenesis of AD. However, when cleaved to yield a 42- or 43amino-acid peptide (Ab1-42/43), these slightly larger fragments appear to nucleate rapidly into amyloid fibrils and apparently give rise to the structural abnormalities described above.

Table 5.7 Genetic Factors Predisposing to Alzheimer’s Disease Chromosome

Gene defect

21

APP mutations

14 1 19

Presenilin 1 mutations Presenilin 2 mutations ApoE4 polymorphism

Source: after Selkoe (1998).

Age of onset 50s 40s and 50s 50s 60s and older

Ab b Phenotype Increased production of total Ab peptides or of Ab1-42/43 peptides Increased production of Ab1-42/43 peptides Increased production of Ab1-42/43 peptides Increased density of Ab plaques and vascular deposits

5.6 Age-related Changes in Humans: Detailed Survey

All four known genetic mutations (table 5.7) involved in familial AD cause an increase in the production of Ab1-42/43 peptides and/or an increase in the density of Ab plaques. These results were confirmed with transgenic mice containing a mutant APP gene (Hsiao et al. 1996). Together, these results strongly indicate that the accumulation of Ab-42/43 in the brain is an early and invariant event in the development of AD pathology (Selkoe 1997), yet the long time periods before the appearance of symptoms suggest that these pathogenic peptides accumulate very slowly. The APP gene is located on chromosome 21, the same chromosome that is involved in Down’s syndrome (Tanzi et al. 1987). All individuals afflicted with Down’s syndrome develop symptoms indistinguishable from those of AD by age 50, and they show diffuse plaques as early as age 12. These observations were some of the earliest evidence suggesting an important role for the APP gene in the etiology of AD. The presenilin genes (see table 5.7) are located on chromosomes 1 and 19. They code for two homologous transmembrane proteins that are also involved in cell–cell signaling. Dewji and Singer (1996) suggested that PS1 and PS2 (presenilin 1 and 2) normally interact directly with the APP protein in an evolutionarily conserved intercellular signaling mechanism. Mutations in either the PS1 or the PS2 gene apparently affect the manner in which cells handle the APP protein, and these mutations are believed to be responsible for the increased production of the pathogenic Ab-42/43 fragments via the cell’s ordinary protein-processing mechanisms. Finally, the ApoE4 gene plays a role in AD in that individuals homozygous for that particular ApoE allele are significantly more likely to develop AD than are individuals with the ApoE3 or ApoE2 alleles. As described in more detail in chapter 8, this situation might arise because the ApoE4 allele is not capable of binding to the tau microtubule protein, thus allowing the unbound tau protein to be hyperphosphophorylated abnormally and thus giving rise to the NFTs characteristic of AD. The ApoE3 and ApoE2 alleles are capable of binding to tau, so the NFT formation is delayed or suppressed (Kamboh 1995; Schacter

183

et al. 1994). In addition, the ApoE4 allele has a lower level of antioxidant activity, and this decreased protective effect probably also plays a role in the pathogenesis of AD (Miyata and Smith 1996). These various genetic processes allow a local accumulation of the self-aggregating Ab1-42/43 peptide. As it accumulates in its insoluble (plaque) form, this protein injures nearby neurons, either directly via neurotoxicity or indirectly via inflammatory reactions on the microglial cells. The mechanisms that produce this damage are becoming clear. We now know that the Ab1-42/43 peptide can induce oxidative damage by binding specifically to a particular receptor protein that has a limited expression within the adult CNS (Yan et al. 1996). This specifically expressed receptor may well convey site specificity to an otherwise ubiquitous protein. The receptor protein normally binds to molecules mediating neurite outgrowth, but it can also bind to the Ab1-42/43 peptide. The Ab1-42/43 peptide can generate reactive oxygen intermediates (which cause oxidative damage; see chapter 10); when bound to the receptor, however, it triggers within the cell an additional and sustained production of oxidants, resulting in oxidative stress and neuronal toxicity (Yan et al. 1996). The receptor–Ab1-42/43 combination also activates the microglial cells, which react by secreting cytotoxic cytokines and by mounting other aspects of an inflammatory response (Yan et al. 1996). In either case, the resultant oxidation severely damages the neurons and decreases their ability to resist subsequent stresses. This decreased oxidative resistance is likely exacerbated by the low levels of mitochondrial activity (and therefore low energy levels) characteristic of AD patients (Davis et al. 1997); in fact, the low energy levels may be due to mitochondrial damage induced by the oxidative stress. Other metabolic changes likely include altered tau phosphorylation and PHF formation in neuritic plaques and NFTs. To the above description, we need to add the very important role of inflammation. As discussed in chapter 13, alterations in input signals of various types can set off a cascade that leads to

184 Chapter 5 Human Aging decreased neural activities, demylineation of the axon, initiation of an autoimmune reaction from the glial cells, and eventually to vulnerability and impaired cognition (figure 13.4). The altered redox imbalance leads to the secretion of proinflammatory cytokines from the microglial cells and so continues and exacerbates this deleterious cascade (figure 13.6). The clinically important consequence of this train of events is eventual synaptic loss and/or neurotransmitter defects, both resulting in an altered neural circuitry and function (Selkoe 1997). However, there is no strict relationship between the level of damage and the extent of the observed behavioral modifications. Education, culture, and socioeconomic status allow for significant modulation of the outcome. The Nun Study (Snowden 1997) combines a long-term, indepth longitudinal study of the behavior and cognitive abilities of a large number of nuns with detailed autopsy data at death. The unselfishness of the nuns to contribute their bodies for postmortem study has allowed the interesting finding that high levels of education, verbal ability, and/or continuing mental activity allows at least some individuals to function at normal behavioral and cognitive levels, despite the ravages of AD. If mental activity stimulates complex neural circuits, the resulting redundancies may allow for multiple additional pathways by which information can get in and out. Using your mind may keep you from losing it. We owe the nuns our thanks for having shown us the truth of this old adage. This pathology most likely has a complex developmental history—more complex than was first thought when the amyloid gene (APP) was first discovered on chromosome 21 (Glenner 1988; Tanzi et al. 1987). For example, it has recently been learned that high levels of insulin might accelerate the accumulation of amyloid plaques because the same enzyme (insulysin) is thought to degrade both amyloid and insulin (Craft et al. 2003). High insulin levels might allow amyloid proteins to escape degradation. This possibility is supported by the fact that diabetics have twice the risk of expressing AD as do non-diabetics (Taube 2003) (but see chapter 7).

Unraveling this etiology is of prime importance in the near future, both for the insight it will give us into normal brain function and for the clinical interventions it may yield so that we may better alleviate or cure the multitudes otherwise doomed to the slow death of their memory and their personality. Other Disorders. The molecular genetics of the mind is not limited to the study of SDAT. New insights are being obtained into non–age-related illnesses such as manic–depressive psychosis (bipolar disorder) and schizophrenia. Genetic studies have suggested that bipolar disorder is due to any one of three different genes causing a predisposition to the psychosis (Robertson 1987). Three different populations yielded three different inheritance patterns, and hence three different locations, of the gene involved, thus indicating again the genetic heterogeneity of the human species. The gene on chromosome 11 might involve an enzyme responsible for the synthesis of the catecholamines, an important class of neurotransmitters. Schizophrenia has long been suspected of having a familial genetic basis. Recent reports suggest that this illness also is quite heterogeneous; in different populations the disease apparently involves different chromosomes. The important point for us to gather from these early reports is that one day a more complete understanding of the molecular biology of the mind, and of how different biochemical processes, whether age-related or not, can affect our mental functioning, will be possible.

5.6.10 Changes in the Immune System Our immune system saves us from certain death by infection. The various responses of this system destroy and eliminate invading organisms and any toxic molecules produced by them. The destructive nature of these responses makes it imperative that the system respond only to cells and molecules that are foreign to the host and not to those of the host itself. This ability to distinguish between self and nonself is a fundamental feature of the immune system. The immune system has

5.6 Age-related Changes in Humans: Detailed Survey

mechanisms by which it first recognizes a large number of diverse and unrelated stimuli, then sorts them into self or nonself categories, and finally translates the detection of the latter group into an “on” signal for the appropriate type of immune response. The complexity of the system, and the abstractness of the foreign stimuli, are suggested by the similarity of the words used to describe the operations of both the nervous and the immune systems—for example, “learning,” “short-term” and “long-term memory,” and “recall.” One reason for the delayed appreciation of the immune system mechanisms is that they are highly dependent on subtle biochemical signals rather than on more obvious morphological structures. 5.6.10.1 Normal Structure and Function

The bone marrow and thymus are the principal structures of the immune system and serve as the source of precursor cells. The spleen and lymph nodes are the secondary structures and serve as the sites at which immunity is initiated. The cells of the immune system consist mostly of B- and T-lymphocytes. These cells come in many specific subtypes, all of which have cell-surface receptors that can respond to a limited group of structurally similar antigens. The antigen is the nonself stimulus molecule that triggers the highly specific immune response. The B-lymphocytes are responsible for humoral immunity, which they confer on the body by producing and secreting specific antibody molecules into the blood and lymph circulation. The immunoglobulin molecules then bind specifically to the antigen (such as a bacterium or toxin) that induced their formation and thereby inactivate that antigen. The T-lymphocytes are responsible for the cell-mediated immune responses. This is a heterogeneous set of responses. One of these responses involves stimulation of the growth and differentiation of B-lymphocytes (and thus regulates the humoral antibody response). Another response involves production of a T-cell subpopulation that can directly recognize and destroy foreign, or nonself, cells. This recognition process is important in graft and transplant rejection.

185

5.6.10.2 Age-related Changes

The most obvious morphological change in the immune system is the age-related involution, or shrinkage, of the thymus, which becomes obvious at the time of sexual maturity (figure 5.18). This reduction in size is due primarily to the atrophy of the cortex, which is responsible for the production of the various thymic hormones required for the maintenance of immune functions. The resulting decrease in the levels of thymic hormones (figure 5.19) is paralleled by the decrease in the number of component T-lymphocytes. The relatively large number of immature T-lymphocytes found within the involuted thymus suggests that the decrease in competent cells reflects the decreased capacity of the thymus gland to promote differentiation of the many immature lymphocytes contained within it. It is clear that immune senescene results in a selective decrease of some secretory factors and that the lowered level of such factors might give rise to alterations in the composition of immune cell populations, and hence in immune function (table 5.8). Oddly, the total number of circulating lymphocytes does not change significantly, despite the involution of the thymus. However, the proportions of the different subpopulations of T-lymphocytes do change with age. These alterations may be responsible for the observed agedependent decrease in natural antibody titers and the concomitant increase in autoantibody titers. The cell-mediated immunological reactions responsible for the rejection of foreign skin grafts and tissues also depend on T-lymphocytes and show an age-related decrease in their ability to perform these functions. For example, in the aftermath of a flu epidemic, the nonimmunized young and old individuals were examined to determine the levels of their immune response to this antigenic stimulus. Among the young individuals, 90% had antibodies against all three flu strains in their nasal fluids; only 63% of the older individuals attained the same level of protection. Furthermore, in a population of individuals examined for their ability to respond to a specific antibody (the yeast Candida), a disproportionate

186 Chapter 5 Human Aging

Section of thymus gland Newborn

10 years

20 years

40 years

60 years

Diseases of aging (cancer, autoimmune diseases, infectious diseases)

Thymusdependent immunity

Serum thymosinlike factors in blood

Diseases of childhood (cancer, infectious diseases)

Birth

10

20

40

30 Age (years)

50

80

Figure 5.18 The relationship of thymus development and function to life span and the incidence of disease. (After Goldstein et al. 1979.)

Figure 5.19 Age-related changes in thymic hormone activity in the plasma. (After Lewis et al. 1978.)

Inductive activity (ng thymopoietin/ml plasma)

20

15

10

5

0 0

186

10

20

40 30 Age (years)

50

60

70

5.6 Age-related Changes in Humans: Detailed Survey

187

Table 5.8 Immunologic Senescence Factor 1. Changes in secretory factors Thymus Thymosin a-1 Thymulin Thymopoeitin Thymic humoral factor 2. Stem cell differentiation Influenced by Bone marrow Thymosin Colony-stimulating factor Interleukin-3

3. Consequences of 1 and 2

Change

Decline in Secretory Factors with Age

Decline in Helper T cell proportion Alloantigen-specific Tk Natural killer cell number Increase in TS cell proportion B cell/T cell ratio Shift in B cell characteristics Antibody production Increase in Tissue graft tolerance Cancer incidence Autoimmune disease Infectious disease

Source: after Sternberg (1994).

number of individuals who were unable to respond (“anergic” individuals) were older than 65 years. In addition, three times as many women as men displayed this defective response. After surgery, the infection and mortality rates for anergic individuals are four- and fivefold higher, respectively, than the corresponding rates for age-matched individuals with a normal immune response (Hausman and Weksler 1985; Phair 1983). Perhaps not all of these defects result from intrinsic changes in the immune system. For example, the decreased proliferation of the Blymphocytes after an antigenic challenge might be at least partly due to a decreased level of mitotic stimulators arising from outside the immune system. At present, the mechanisms that initiate the involution of the thymus are not known. As involution and atrophy are thought to begin during the first year of life, involution may be a hormone-independent intrinsic aging process. But this is admittedly a minority view. In any event,

it has been shown that the age-dependent drop in the immune competence of aged mice can be partly reversed by treatment with thymic hormones (table 5.9). Clinical trials in humans are promising and suggest that the therapeutic use of thymosin enhances the body’s ability to ameliorate immune defects. Nonetheless, although it seems plausible that age-related declines in immune function might contribute to the vulnerability of the elderly to disease and thus increase the risk of mortality, much of the evidence to date is indirect or correlative. Thus, there is abundant evidence that the functional ability of the human immune system changes with age. The underlying cause(s) of this immunosenescence are not known, although the involution of the thymus may have all the characteristics of a true age-related change. The thymus, of course, occupies a key position in the continued functioning of the immune system. However, the very complexity of the immune

188 Chapter 5 Human Aging Table 5.9 Effect of Thymopoietin Administration on the Ability of Old Mice to Respond to a Specific Antigen Age of mice (months)

Thymopoietin treated?

2 24 24

No No Yes

No. of competent cells/spleen 5916 ± 2.3 385 ± 79 977 ± 102

Source: after Hausman and Weksler (1985).

system, coupled with its interactions with the nervous and the endocrine systems, suggest that not all aspects of immune senescence can be traced back to this one root cause. In any event, it should be clear that intrinsic age-related changes in the neural, immunological, and/or endocrine components of the body’s communication system will likely have profound effects on many of their target organs.

5.6.11 Changes in the Reproductive System The reproductive systems of both sexes show age-dependent decrements in function. These changes are most obvious in the female, where they lead to a loss of fertility during midlife. These age-related changes as studied in the female rodent model by Finch and colleagues have led to the development of the neuroendocrine model of aging (see chapter 13). In males, reproductive aging is less dramatic and leads to a decrease in fecundity rather than an absolute loss of fertility. I review reproductive aging primarily from the viewpoint of neuroendocrine control. The reproductive system is strikingly different from other body systems with respect to the mechanisms controlling its development, its acquisition of function, and its loss of function. Furthermore, as discussed in chapter 4, there appears to be a strong relationship between speciesspecific reproductive strategies and speciesspecific longevities. For the human female, this strategy means that she spends substantial portions of her total life span (about 45%) in either

a prereproductive or a postreproductive state. Expression of the strategy depends on the proper functioning of the genetic and neuroendocrine mechanisms that control the reproductive organs. The primitive gonad in utero differs from other body tissues in that it contains the primordia for both ovary and testis and could potentially develop either male or female structures. The developmental decision to choose one alternative or the other is made at about the eighth week of gestation and is the result of an interplay between genetic and hormonal factors. Abnormal sexual development often results from breakdowns in these control processes. In mammals, birth is followed by a period of gonadal quiescence until later activation of the gonads by pituitary gonadotropins. The final maturation of the reproductive system begins upon such activation. This period of growth and maturation is known as adolescence. Puberty defines the maturational state when reproduction is first possible, even if it is not feasible. The neural mechanisms that cause this activation are not yet clearly defined. One theory suggests that a hypothalamic mechanism holds gonadotropin mechanisms in check until puberty. Another theory suggests that small amounts of gonadal steroids inhibit hypothalamic production of the gonadotropins that activate the system and that puberty comes about as a result of the brain’s decreased sensitivity to this inhibition. A third theory suggests that puberty is initiated by the removal of a local inhibitory effect on gonadal activity. Despite the absence of a definitive theory for the onset of puberty, once it is initiated, the neuroendocrine control of reproductive function is quite definitive, particularly in the female. 5.6.11.1 Female Reproductive Aging

Normal Structure and Function. The two ovaries store and alternatively release a mature ovum each month into the uterine tubes for fertilization and transport to the uterus. A woman ovulates perhaps 500 or fewer eggs in her lifetime; this number represents only a tiny portion of the eggs she was born with or that she had at the onset of puberty (table 5.10). This atresia (“wastage”) is a normal compo-

5.6 Age-related Changes in Humans: Detailed Survey

Table 5.10 Effect of Age on the Number of Oocytes in the Human Ovary Age (years) 4-Month-old fetus Birth 4–10 11–17 18–24 25–31 32–38 39–45

Estimated no. of Oocytes 3,500,000 733,000 500,000 390,000 162,000 62,000 80,000 11,000

Source: after Talbert (1977).

nent of follicle development and probably represents a selection mechanism by which only the fastest-growing oocyte is chosen for ovulation. The follicular store is probably exhausted or nearly so at the time of menopause. Women suffering from precocious menopause often have ovaries mostly devoid of follicles. The functional life of the human ovary appears to be proportional to its follicular store and is not simply a matter of chronological age (Wise 1986). The process of follicular destruction (atresia) via apoptosis is under genetic control (see chapter 7). Recent data suggest the possibility that female mice might have active germline stem cell replacement of ova. If this process is operative in humans, then the combined replacement/destruction cycle would lead to a dynamic picture of the processes maintaining the follicular population. The normal sequence of events during the menstrual cycle is shown in figure 5.20 and is dependent on neuroendocrine function.The hypothalamus is the primary regulator of the reproductive system. The hypothalamus contains neurons that synthesize and secrete from their axon endings various protein hormones instead of neurotransmitters. These hormones, usually called gonadotropin-releasing hormones, travel via a special blood capillary network to the anterior pituitary immediately below where they stimulate the pituitary cells to secrete the two primary gonadotropin-releasing hormones: follicle-stimulating hormone (FSH) and luteinizing hormone (LH). These two hormones are crucially involved in the regulation of ovulation.

189

FSH stimulates the development of 10–20 follicles and thereby brings about an elevation in the blood level of estradiol. Only one follicle grows fast enough and is mature enough to be able to respond to the signal next provided by LH. Blood levels of LH increase significantly just before ovulation. LH stimulates the rupture of the follicle and the extrusion of the ripe ovum. This process constitutes ovulation. The ruptured follicle continues to grow and is transformed into a corpus luteum, which actively produces progesterone, as well as estradiol. This action of LH and FSH on the ovaries is blocked by prolactin, a hormone usually produced at high levels only when the mother is nursing. This inhibition underlies the contraceptive effects of nursing. As the levels of the two ovarian hormones, estradiol and progesterone, increase, they act on the hypothalamus to inhibit the production of the gonadotropin-releasing hormones. The decrease in these factors brings about the curtailment of LH and FSH production by the pituitary. As a result of this drop in levels of gonadotropin-releasing hormones, the production of the ovarian hormones drops. If fertilization has occurred, implantation of the fertilized egg in the endometrium stimulates the production of progesterone independent of the hypothalamic–pituitary control axis, thereby maintaining the pregnant state. In the absence of fertilization and implantation, the production of progesterone ceases. Without hormonal support, the endometrium can no longer sustain itself, and a large portion of it is sloughed off in the menstrual fluid. Finally, as a result of the now low level of ovarian hormones, the level of the pituitary gonadotropin-releasing hormones begins to rise again, and the cycle begins anew. The basic logic underlying this control mechanism is negative feedback: An excess of a substance acts to shut down its own production. This is the same logic as is found in the thermostat that regulates operation of a furnace. The apparent complexity of reproductive control arises mostly from the fact that it comprises several interacting and cascading negative feedback cycles. However, there are still some important questions regarding the detailed operation of the cycle that remain unanswered.

190 Chapter 5 Human Aging

(a) Gonadotrophic hormones (from anterior pituitary) Luteinizing hormone (LH)

Follicle-stimulating hormone (FSH) Ovulation (b) Events in ovary Developing follicle

Corpus luteum

Egg

Progesterone

(c) Ovarian hormones Estrogen

(d) Uterine lining

Follicular phase

Menses 0

5

10

Luteal phase 15

20

25

Day of menstrual cycle Figure 5.20 A diagram of the events that take place during the normal menstrual cycle. See text for discussion.

Age-related Changes. The cessation of the menstrual cycle is the major age-related change in this system. Follicular deficiency is the most striking feature of the human ovary after menopause. The ovary is not completely afollicular, for at age 50 it still contains a few hundred or a few thousand follicles. The end of reproductive life is not due to the absolute absence of ova, but due to a failure elsewhere in the system. In humans, this primary event appears to involve the failure to produce ovarian hormones. Clearly, the depletion of follicles does result in a decreased level of ovarian estradiol secretion. This decrease in estradiol,

in turn, might well account for the higher levels of FSH and LH seen in middle-aged premenopausal women. Even after menopause, normal levels of FSH and LH can be induced by exogenous estradiol (Finch and Gosden 1986). The consequences of ovarian estradiol decline are widespread because many other tissues depend on this hormone for their normal maintenance. The affected tissues include not only components of the reproductive system itself (uterus, vagina) and components of secondary sexual characteristics (breasts, external genitalia), but also nonreproductive organs such as skin, skeleton, and

5.6 Age-related Changes in Humans: Detailed Survey

cardiovascular system. Tissue changes probably arise as a result of the cessation of various hormone-induced gene activities in these several tissues. The possibility that concomitant neural changes are taking place, particularly in the hypothalamus, and that these neural changes also act to bring about the menarche, must also be considered. This hypothesis that the decline in estradiol affects various systems is supported by studies of hot flashes in women. Hot flashes occur in the majority (about 85%) of women at menopause. These are episodes of brief (average 2.7 minutes) increases in skin temperature (by an average of 7.5°F) accompanied by increases in pulse rate (by 9–20 beats per minute) and blood flow (Rebar and Spitzer 1987). Hot flashes are associated with a pulsatile release of LH, and they usually can be effectively abolished by estrogen therapy. The neurons of the hypothalamus that regulate the release of LH lie close to the neurons that are involved in thermoregulation. This juxtaposition suggests that some sort of neural change may have taken place in these neurons such that high levels of LH now stimulate the thermoregulatory neurons. Because the proper functioning of the reproductive system depends on multiple feedbacks, a change in either the gonadal or the neural components will rapidly affect other parts of the system and result in a cascading deterioration of reproductive abilities (Wise 1986). Studies on the female mouse have elucidated the nature of hypothalamic involvement in reproductive aging (see chapter 12). Mice and humans differ in enough significant parameters that the same explanation is unlikely to apply in detail to both. Humans use the reproductive system not only for procreation and childbearing but also for sexual pleasure and the expression of love and affection. It has been demonstrated not only that elderly persons have sexual needs and can enjoy sexual relations (Whitbourne 1985), but also that the enjoyment of sexual relations is a positive indicator of longevity (see figure 8.7). The agerelated physical changes that I have described affect the body’s functioning. It is to be expected that these diverse changes would also affect sex-

191

ual physiology. Masters and Johnson (1966) described these changes in the human female and male. There is no physiological reason for women to view themselves as asexual after menopause, and there appear to be important psychological reasons for individuals to continue appropriate sexual interest and activity. 5.6.11.2 Male Reproductive Aging

Normal Structure and Function. Sperm are produced in the testes. Each testis is subdivided into about 250 compartments, and each of these compartments is tightly packed with highly coiled seminiferous tubules. The two testes together contain a total length of about 500 meters (1640 feet) of tubules. Sperm are produced continuously within the tubules. The sperm is a highly differentiated cell specialized for the task of delivering one inactivated haploid set of chromosomes to the ovum. While in the testes, the sperm are immobile. They become partly motile only after they have spent about 18 hours in the epididymis. The sperm become fully motile and mature only after they have entered the female reproductive tract. Sperm are stored in the ductus deferens. The seminal fluid consists of the secretions of the seminal vesicle, the prostate, and the bulbourethral glands. Each ejaculate of 3–6 milliliters usually contains some 300 million to 400 million sperm. Of these, only one can fertilize the ovum. The other sperm cells may play an important, albeit accessory, role, since men who have fewer than 20 million sperm per milliliter of ejaculate are generally sterile. The male reproductive system is under a less obvious form of neuroendocrine control than that of the female reproductive system. In addition to producing sperm, the testis secretes the male sex hormone, testosterone. This hormone is responsible both for the normal functioning of the accessory glands of reproduction and for the development and maintenance of the secondary sexual characteristics. Testosterone is produced by the Leydig cells of the testis. These cells are activated by LH, and the effect is enhanced if FSH is also available. Maintenance of the structure of the seminiferous tubules and of

192 Chapter 5 Human Aging sperm development in the tubules depends on the combined effects of FSH and testosterone. Thus the same hypothalamus–pituitary–gonad axis of control is active in both sexes. Human males exhibit a daily, not monthly, rhythmicity in their testosterone levels. This variability in gonadal hormone secretion is paralleled by the finding that LH levels in normal young men vary in a pulsatile manner. Sleeping and waking states show a difference in the interpulse interval. It does not appear to be fully accurate to state that women have a cyclic reproductive pattern while men do not. It is probably more realistic to say that both sexes have cyclic reproductive patterns, but their cycles are quite different from one another (see Hermann et al. 2000 for further details). Age-related Changes. Following the same logic as mentioned above for females, I discuss the nature of age-related changes and how they affect the reproductive and sexual functions in men. The data of figure 5.21 reveals two major changes. First, mean testosterone level appears to decrease with age. Second, and more striking, the

circadian rhythm that is characteristic of testosterone production in young men is absent in old men. This decrease in testosterone level is accompanied by an increase in LH levels (Bremner et al. 1983) and by a loss of the pulsatile LH secretion characteristic of young men. However, either because not enough LH is produced or because Leydig cells cannot respond to the increased LH, the testosterone level in the blood does not increase. This situation is analogous to that of the postmenopausal female (Whitbourne 1985). However, these age-related changes seem to make little difference in the functional aspects of semen in old men compared to young men. The decrease in normal motile sperm in old men appears to be compensated for by an increase in sperm density. The decreased fertility associated with increasing age appears to be associated more with cultural than with physiological events. In fact, it has been suggested that the drop in testosterone levels may have much to do with a decrease in numbers of Leydig cells arising from a decreasing frequency of sexual encounters. If confirmed, this may well be the primordial instance of “use it or lose it.”

8 Young men

**

** Testosterone (ng/ml)

7

** **

* *

*

*

*

**

* **

6 Old men 5

4

0800

1200 (noon)

1600

2000

2400 (midnight)

0400

0800

Clock time (hours)

Figure 5.21 Hourly serum testosterone levels (mean ± standard error) in normal young and old men. Asterisks indicate statistically significant differences (*<.05; **<.01) between age groups at that point in time. (After Bremner et al. 1983.)

5.6 Age-related Changes in Humans: Detailed Survey

There is no obvious correlation between levels of sexual activity and testosterone levels, suggesting again that eroticism may reside more in the mind than elsewhere. However, some real age-related changes take place in the sexual responsiveness of the human male. Changes in the vascular system and in the signal transduction mechanisms responsible for penile erection make erectile dysfunction a not uncommon age-related loss of function in males. The development of Viagra and related drugs was important, not just for the more normal sex life it allows some men, but for the demonstration that non–lifethreatening losses of function are now viewed as deserving alleviation. This new understanding underlies some of the discussion of anti-aging interventions discussed in chapter 15.

5.6.12 Metabolic and Hormonal Changes Given the kaleidoscope of age-related changes surveyed in this chapter, it is to be expected that these morphological changes are paralleled by myriad biochemical alterations. An enormous number of individual metabolic reactions are involved in the formation and destruction of the literally thousands of chemical compounds necessary for continued vitality. For any given compound, the rule seems to be that different sets of reactions are used for its synthesis and for its degradation. Thus there are multiple points at which any given reaction can be controlled. In this section I briefly review the important changes in major aspects of metabolic activity, with particular reference to humans. 5.6.12.1 Energy Metabolism

Measuring all the variables affecting energy metabolism in a cellular organelle such as a mitochondrion is daunting enough; to measure individually all the variables operative in an entire organism is impossible. However, a set of standard conditions has been developed under which a standard measure of metabolism, the basal metabolic rate (BMR), can be measured. Early cross-sectional studies showed that the

193

BMR declines with advancing age; these results were contradicted by later longitudinal studies that controlled for the changes in body composition with age. It was later shown that the basal consumption of O2 (BMR) per liter of body water is constant (figure 5.22). The observed decline in the BMR unadjusted for water content is presumably due to the observed loss in muscle mass with age in humans (Tzarkoff and Norris 1978). There is no age-related decline in the overall resting metabolic activity of the body’s cells and tissues. This conclusion is supported by the repeated findings that no age difference exists for mouth or axillary temperatures in the range of 20–100 years (Shock 1977). The BMR is a laboratory measurement. In the real world, people expend a higher level of energy than is suggested by the BMR. Daily measurements of actual energy metabolism, unadjusted for changes in body mass, decline (figure 5.23). This decline persists even when one adjusts the data to reflect changes in body mass with age (see figure 5.23d). There appears to be a real age-related decrement in the body’s ability to produce energy above the resting levels reflected by the BMR, at least between the ages of 28 and 60 years. This apparent age-related decrease in maximum energy production in men (and presumably women as well) correlates well with many (but not all) studies showing that in various animal species, the older animals often have fewer mitochondria. Furthermore, these senescent mitochondria are usually less efficient in oxidative phosphorylation and other aspects of energy production (see chapter 11). It is a well-known and verified observation that the maximum life span of any mammalian species is inversely proportional to its specific metabolic rate (calories/gram/ month or year; see chapter 4). It is another wellestablished observation that increasing the metabolic rate decreases the life span (Loeb and Northrop 1917). This observation is particularly well documented in poikilotherms, in which the metabolic rate varies according to the environmental temperature. The positive correlation between the metabolic rate and the rate of aging has led to the development of several theories of

35

30

Basal oxygen consumption (ml/min) ( )

Body water (liters) ( )

40

225

6

205

5

185

4

30–39

40–49

60–69 50–59 Age (years)

70–79

Basal oxygen consumption (ml/min/liter body water) (O)

194 Chapter 5 Human Aging

80–89

(c)

2,800 2,600 2,400 2,200 2,000

Basal calories/day

(b)

3,000

2,200 2,000 1,800 1,600 1,400 1,200

(d) 1,100 900 700 28 40 50 60 70 Age (years)

80

Calories (total–basal)/kg/day

(a)

Calories (total–basal)/day Total dietary calories/day

Figure 5.22 The relationship of age to basal oxygen consumption, total body water, and basal oxygen consumption per unit of body water. (After Gregerman 1967.)

14 12 10 28

40

50 60 70 80 Age (years)

Figure 5.23 Energy expenditure by men. (a) Total caloric intake per day. (b) Basal metabolic rate. (c) Energy expenditure in addition to the basal expenditure. (d) Energy expenditure in addition to the basal expenditure, per kilogram body weight. (After McGandy et al. 1966.)

aging. These classic theories have recently been disproven in their strict sense, but modern versions of them appear to provide us with some good insights into the relationship between aging and energy. I examine these theories in more detail in chapters 6 and 11.

5.6.12.2 Fuel Utilization and Storage

Most of the body’s energy stores are maintained in the form of triglycerides found in the adipose tissue; the remainder are mostly in the form of protein, and a trace (< 1%) is found as glycogen.

5.6 Age-related Changes in Humans: Detailed Survey

Age-related changes in fuel storage and utilization appear to parallel the shifts in energy metabolism. Fat makes up an increasing fraction of total body mass with age, at least in people following the all-too-common pattern of overeating and under excercising characteristic of a sedentary consumer society. Animal studies suggest that aging has little effect on glycogen content or its utilization by the liver or muscles (Masoro 1985). Advancing age is accompanied by a large (31%) loss of body protein content, mostly from the skeletal muscles. Protein stores are not utilized for energy metabolism except under extraordinary (starvation) conditions. Little is known regarding the role of aging in the utilization of proteins for energy metabolism. There does appear to be a change with age in the body’s ability to utilize both fats and carbohydrates—a change that is related to endocrine changes. To utilize fat as an energy source, the triglycerides stored in the adipose cells must be enzymatically converted to free fatty acids. This process is controlled in large part by the hormone glucagon, and there is an age-related decrease in the ability of the adipose cells to respond to that hormone. However, this age-related rate and pattern of the decline can be substantially modulated by diet, particularly by caloric restriction, such that the treated animals retain a youthlike response for a long time (Masoro 1985). I return to this topic again in chapters 7 and 11. It is interesting to speculate on whether this decreased lipolytic response of the individual cells is compensated for by the age-related increased mass of the body composed of fat. 5.6.12.3 Hormonal Changes Associated with Aging

Many metabolic reactions are controlled and/or influenced by a variety of hormones. In general, the different endocrine glands show a similar pattern of morphological changes with age. The endocrine glands lose weight and develop a patchy, atrophic appearance accompanied by vascular changes and fibrosis (Minaker et al. 1985). The basal levels of many hormones are not affected by age, even though the secretion rates of most hor-

195

mones decrease. This situation necessarily implies that the clearance rates must have decreased in a compensatory manner. There appear to be no general systematic changes in the number or quality of receptors. The intracellular response to the hormone often appears to be diminished as a function of age. The foregoing is a general description and does not apply to all cases. In fact, it has been known since the work of J. C. Spence in 1921 that there is an age-related decrement in the ability of the body to maintain carbohydrate homeostasis after glucose challenge. Cross-sectional studies have shown that there is no change with age of the fasting blood sugar level. What changes is the rate with which the blood glucose level returns to normal after a glucose overload. The oral glucose tolerance test, used to quantitate this process, measures the glucose level in the blood immediately after and 2 hours after the patient drinks a sickeningly sweet mixture. Normal individuals show a rapid drop; diabetic individuals do not. The intermediate values characteristic of an impaired glucose tolerance are typically shown both by large numbers of elderly persons and by younger individuals at risk of developing diabetes. This age-related homeostatic impairment is due not to decreased levels of insulin released and/or circulating in the bloodstreambut apparently to a decrease in the sensitivity of peripheral (nonhepatic) tissues to insulin. The maximum response to insulin is the same in both young and old, but with age the dose–response curve is shifted to the right such that older subjects require twice the amount of insulin as do young subjects in order to attain a given level of glucose uptake. The change in rate does not appear to be associated with changes in number or quality of insulin receptors or the cell membrane. Thus, this decrease must involve currently unknown agerelated changes in the intracellular response of the cell to glucose. In addition, factors such as obesity, diet, and exercise affect the extent of the decrement in glucose homeostasis. Glucagon is involved in both carbohydrate and fat metabolism. Glucagon physiology does not change significantly with age, despite the changes occurring in insulin physiology and

196 Chapter 5 Human Aging despite the age-related changes in glucagon-induced lipolysis. The two hormones and the two sets of metabolic reactions they regulate retain their autonomy despite the interconnectedness of metabolic reactions. Diabetes. Given the increased incidence of diabetes in the elderly, one might suppose that serum insulin levels fall with age. This is not the case, however. The increased incidence of the disease predominantly involves non–insulin-dependent diabetes (NIDD) or adult-onset type 2 diabetes. In this condition, it is not the insulin production that is affected, but rather the ability of the target cells (predominantly skeletal muscle cells) to respond appropriately to the hormone. Before eating, both young and old may have comparable blood insulin levels. But after exposure to an oral glucose load (or exercise or infection or hormonal changes), responses often diverge markedly in that the older person’s insulin levels will often stay higher longer as a result of slower insulin clearance and decreased uptake of the hormone by the target cells. In addition, the target cells may have an altered, less sensitive response to insulin. There are both genetic and environmental factors that interact to bring on the disease. Diabetes has been defined as “a lifestyle disorder with the highest prevalence seen in populations that have a heightened genetic susceptibility; environmental factors associated with lifestyle unmask the disease” (Diamond 2003, p. 600). The leading evolutionary theory for the high prevalence of genetic factors for diabetes is the “thrifty gene” hypothesis of Van Neel (1962), which posited the selection of genes (such as those for high insulin levels) allowing our primate and human ancestors to store food as fat during the occasional times of food abundance. When populations with such an evolutionary history develop a food-rich sedentary consumer society, then the result is the current diabetes epidemic, which accounts for 15% of all medical costs in the United States alone. As I discuss in chapter 7, insulin and insulinlike proteins have an inhibitory effect on the expression of stress-resistance mechanisms necessary for extended longevity. Thus, people with subclinical levels of insulin resistance, as well as those with

full-blown type 2 diabetics, have lower levels of antioxidant defense genes that are activated by one particular transcription factor (NRF-1; Patti et al. 2003). Additional transcription factors, some of which are thought to coactivate these genes, are also decreased. The decreased expression of these key control factors likely gives rise to decreased transcription of various metabolic and mitochondrial genes, and these in turn might give rise to the decreased oxidative metabolism, decreased lipid oxidation, hyperlipidemia, and insulin resistance characteristic of diabetes. In summary, the aging endocrine systems show fairly well-defined but incompletely understood changes that may have profound effects on cell function and cell survival. These include both a loss of capacity to respond quickly and fully to stimulation (as in NIDD), as well as declines in the baseline of several hormones such as growth hormone, dehydro epiandrosterone, antidiuretic hormone, and sexual steroids. The clearest age-related changes in function attributed to changes in the baseline level of particular hormones are menopause in women, androgen decline in men, and decreased growth hormone. Other important age-related changes in function stem from alterations in the target cell’s response to the target hormone (NIDD, but also the effects of cortisol on neural function, which I discuss in chapter 13). As a result of the changes in the individual endocrine systems, the body’s long-range communication system becomes less effective, and these subtle changes in endocrine communication must facilitate the further dysfunction of different organs and systems within the body. The sweeping generalizations about falling hormonal levels being responsible for all kinds of senescent phenomena are misleading and often simply false. Endocrine aging is not so much a decline in endocrine levels as it is a decline in endocrine communication leading to a critical degradation of normal endocrine coordination of cellular and tissue functions. Fossell (2004, ms. p. 452) has written: If the job of endocrine systems is to provide communication and coordination, aging undermines both. The central problem in communication failure is not volume, but fidelity of the message. In endocrine aging the prob-

5.6 Age-related Changes in Humans: Detailed Survey

lem is not serum level, but appropriateness of the endocrine response. With aging, messages becomes degraded, responses inappropriate, often independently of endocrine levels. This altered communication gives rise to other changes that contribute to the overall pathology. I discuss certain of these processes in more detail in chapters 7 and 13. Metabolic Syndrome. The concept of the metabolic syndrome was first introduced by Camus (1966), and its importance has been increasingly recognized (Morley 2004). The syndrome consists of several age-related factors that interact with each other to yield hyperinsulemenia, insulin resistance, aetherogenesis, and death from myocardial infarction. The age-related factors commonly involved are diabetes, hypertension,

197

hyperuricemia, lipid abnormalities, and an increased probability of forming aetherosclerotic plaques (see table 5.6). The presence of the syndrome is generally correlated with increased visceral obesity and a sedentary lifestyle, as shown in figure 5.24. The role of insulin and insulin resistance is clear, but other hormones are also involved. Leptin is produced by the adipose cells and would normally act to increase metabolism and decrease visceral obesity. But with increasing obesity and hypertriglyceridemia, resistance to the effects of leptin occurs in the target cells. Hypertension, in the presence of lipid abnormalities such as high triglycerides and low HDL levels, can lead to the production of aetheroscleroic plaques and an increased probability of death from myocardial infarction. Left untreated, the sedentariness and overeating will likely lead to a debilitating and unpleasant elder life, as well as

Elements of the metabolic syndrome and their interactions.

SEDENTARY LIFESTYLE + OVEREATING

VISCERAL OBESITY

Lipid Abnormalties

Plaques

Hyperinsulemia

Leptin & TNH

Insulin Resistance

Hypertension Hyperuricemia Diabetes

Myocardial infarction

DEATH

Figure 5.24 The manner in which lifestyle factors interact with elements of metabolic regulatory circuits so as to give rise to the complex of related pathologies known as the metabolic syndrome. (Redrawn after Morley 2004.)

198 Chapter 5 Human Aging one that is extraordinarily difficult and expensive to medically manage. Much of the current concern about the obesity epidemic in the United States has its roots in both medical and financial concerns about the future. It may be the case that this common clustering of metabolic disorders has a common origin. A mutation in a mitochondrial tRNAIle anticodon is significantly associated with the linked maternal inheritance of hypertension, hypercholeserolemia, and hypomagsesemia (Wilson et al. 2004). Mitochondrial defects are associated with type 2 diabetes and insulin resistance (see chapter 12). Although diabetes, obesity, and insulin resistance were not present in the kindred analyzed by Wilson et al. (2004), all components of the metabolic syndrome have been linked to various losses of mitochondrial function. It is possible that all the features of the metabolic syndrome may arise from the pleiotropic effects of aging on the loss of mitochondrial function. 5.6.12.4 Pharmacological Changes

Associated with Aging Although some drugs are eliminated from the body largely unchanged, most are converted to a wide variety of metabolites before they are excreted. Given the wide-ranging nature of the physiological age-related changes already described, it should come as no surprise that some of these alterations are of pharmacological importance. This extensive topic was reviewed by Vestal and Dawson (1985) and by Beizer and Timiras (1994). The drugs must be absorbed from the site of administration (usually the gastrointestinal tract) into the circulation, distributed via the circulatory system to both the central and the peripheral tissues, metabolized by the tissues, and excreted by the kidneys. Each of these processes may be affected by various factors. Because of the broad physiological variations seen among the elderly, as well as the wide variety of drugs available, it is difficult to offer broad and valid generalizations on this topic. Nonetheless, it should be clear that many of the age-related changes discussed earlier will have profound effects. For example, the age-related decrease in total body

water and increase in body fat will ensure that a constant dose of a water-soluble drug will result in higher blood levels in the elderly, while a fat-soluble drug might result in higher blood levels in the young. The changes in free drug concentrations in the circulation might be exacerbated by the 32% decrease in the weight of the liver from the fourth to the eighth decade of life. If the intrinsic metabolic capacity of the liver to process such drugs were correspondingly reduced, one would expect these drugs to exhibit longer half-lives and a consequent reduction in clearance times. And, of course, agerelated changes in the density and/or binding efficiency of the drug receptors would contribute to an alteration of drug effectiveness in the elderly. These predictions appear to be upheld in various studies. In elderly patients, the drug propranolol exhibited a 28% longer half-life than in younger people, coupled with a clearance rate only 76% that of younger people. One might expect a similar variation in the symptomatic effects of this drug in people of different ages. Similar studies on another model drug, antipyrine, illustrate the large amount of individual variation in drug metabolism. Statistical analysis of these data substantiates the impression gained from a visual inspection—namely, that the individual variation and environmental effects (such as smoking) far exceed the effect of age alone. Nonetheless, there is an age effect on pharmacokinetics. In some cases, the expected increase in adverse reactions with age has been documented. The incidence of such adverse reactions is much increased at the higher doses, a situation in which many older individuals might find themselves as a result of long-standing chronic conditions. Thus we may view these pharmacological alterations as a specialized but increasingly important aspect of the more general metabolic changes seen with aging.

5.7 Interactions between Aging and Disease The preceding description of normal age-related changes in humans should make clear that almost all age-related changes represent a decrement in

5.7 Interactions between Aging and Disease

function. I have tried to distinguish normal agedependent changes in structure or function from age-related pathologies. This distinction is not easy to draw, in large part because normal agerelated changes are believed to be associated with the prevalence of age-related pathologies. I explore this association in terms of the specific framework of the cardiovascular system. Another conceptual area of interest regards the fit between the simple, clean, and rigorous definition of age-related changes as postulated by Strehler (1982; see the CPID criteria in chapter 1) and the complex, overlapping, and fuzzy descriptions I have presented of the actual age-related changes. I explore this fit in the context of agerelated changes in the skeletal system and suggest some possible alterations in these concepts.

5.7.1 Cardiovascular Interactions Some aspects of the association between the normal and the pathological may represent the presence of time-dependent processes (e.g., changes in aortic elasticity) that may act as a predisposing factor for the development of the abnormality. Undoubtedly, the shift from normal to pathological may also represent situations in which the accumulation of many small quantitative changes may bring about a sudden, irreversible, and qualitative alteration. The branch of mathematics known as catastrophe theory might be appropriate for modeling such situations. The dynamic interactions between the aging process and cardiovascular disease can best be described as follows (Vesilinovitch 1986). The aging processes particularly relevant are those involved in (1) arteriocapillary fibrosis, (2) reduced elasticity of the vessel wall, (3) reduced blood flow, (4) enhanced endothelial injury, (5) defective endothelial repair, and (6) elevated blood lipids. Every one of these processes normally occurs in individuals not affected with cardiovascular disease, and thus they cannot be regarded as pathological states by the CPID criteria. However, their presence will predispose and accelerate the course of the disease. In turn, the advancing disease state will enhance the progres-

199

sion of any of these age-related changes. The disease and the aging processes exert positive feedback effects on each other, thus creating a vicious cycle that leads to accelerated aging and increased debilitation. The goal of any effective segmental interaction would be to break, or at least minimize, the positive feedback signals accelerating this deleterious cascade. More recent work has focused on the subcellular mechanisms underlying heart disease. It seems clear that the progressive accumulation of mitochondrial mutations in cardiac cells correlates with the onset of defects in oxidative phosphorylation, suggesting a potential causal relationship (see chapter 11). The positive feedback between mitochondrial mutations and metabolism is thought to set up a cascade resulting in the progressive onset of age-related loss of function. The data to support this is so far more convincing in mice than in humans (Chien and Karsenty, 2005), but it is probably reasonable to search for ways in which these subcellular mechanisms might interact with, and aggravate, the cellular processes described above.

5.7.2 Skeletal System Interactions As already described, there is a wealth of knowledge regarding the age-related phenomenon of bone loss and its role in both normal and abnormal aging. It will be instructive to compare Strehler’s operational definition of aging (see chapter 1 for CPID criteria), which we are using, with this rich database to see how well (or how poorly) the definition fits the facts without confounding the normal and the abnormal. This comparison of principles and observation reveals a close but not complete congruence between the two (table 5.11). It seems that these defining criteria need to be reviewed to allow for the considerable heterogeneity of the human population (see point 2 of table 5.11). I dealt with this definitional problem of heterogeneityin chapter 3. It would also be useful if the deleterious effect (point 4 of table 5.11) of the normal age-related change were viewed as arising from the fact that it serves as a precondition for the possible development of a pathological age-related change.

200 Chapter 5 Human Aging Table 5.11 Comparison between Theoretical Observations and Actual Observations Principle

Observations

Cumulative

The decrease in bone mass is cumulative over time. Significant differences in the rate of progression exist between men and women in general and between any two individuals of the same gender based on heredity and race. Progression rates differ for various dichotomous classes such as spongy vs. compact bone, pre- vs. postmenopausal women, etc. Extent and kinetics of bone loss are significantly modulated by various environmental factors, some of which are extrinsic to the organism (e.g., diet, exercise) and others of which are intrinsic to the organism but extrinsic to the bone tissue (e.g., gonadal hormone status). There is no doubt of the deleterious effects that bone mass loss can have on an individual. But in and of itself, it is not deleterious. Not only do these losses not affect the death of the individual, but the majority of the elderly do not have symptomatic problems of the skeleton that can be traced solely to decreased bone mass.

Progressive

Intrinsic

Deleterious

Source: from Klebba-Goodman (1986).

In the skeletal system example, we would make the following sort of statement: A decrease in bone mass is a deleterious age-related change initially brought about by changes in certain intrinsic factors and strongly modulated by certain extrinsic factors, all of which interact in a complex cascade to reach the end point of this pathogenic pathway—namely, osteoporosis. This approach directs our attention to the necessary age-related changes that are a precondition to the pathology associated with senescence; it clarifies that decreased bone mass and osteoporosis are not the same condition, but rather that the decreased bone mass is a precondition of the end point of osteoporosis. It would probably be useful if we could devise a statistical normalization procedure

that would clearly distinguish between these normal and pathological age-related changes. In addition, these age-related changes can be modulated to some extent, principally and most easily by appropriate exercise and nutrition. In that sense, then, their deleterious effect is not a fixed attribute. Nathan Shock (1983), generally regarded as one of the founders of modern biogerontology, once wrote: Aging and disease are not synonymous. Aging is a normal part of the life span characterized by slowly progressing impairments in the performance of most organ systems. . . . The ability to adapt to stress also diminishes with age. There is mounting evidence to support the premise that aging in humans may be more than simply the summation of the changes that take place at the cellular, tissue or organ level. The essence of survival is the proper integration of responses of different organ systems to adapt to the stresses of daily living. Impaired effectiveness of these coordinating mechanisms may be the primary factor involved in aging. (p. 137) Today we agree with Shock that senescence involves the loss of integration, but we would question the assumption that aging is not a disease. That view made sense as long as there was no evidence that the age-related loss of functions in each of the different organ systems were modifiable or were connected to one another in some mechanistic manner. But today, as the following chapters make clear, we are beginning to understand the nature of the genetic mechanisms and pathways involved in the aging process, we can significantly retard the aging process, and we can reverse parts of it. If each age-related loss of function eventually gives rise to a disease (aetherosclerosis, diabetes, sarcopenia, etc.), and if the progression of each of these losses of function are somehow correlated with the progression of senescence, then isn’t the aging process, which is responsible for each disease, a part of the disease process itself? As Fossell (2004) put it:

5.7 Interactions between Aging and Disease

Where we stand today is where we stood 130 years ago with regard to microbial disease. In the early 1870s, the feeling was that there was a vast collection of diseases that had no relationship to one another—diseases such as malaria, lockjaw, and hosts of other diseases. We now recognize that all of these diseases have a microbial basis, whether that is bacterial, viral, or fungal. They are linked by a single common concept—they are caused by an organism that we cannot see. That realization gave us a new place to intervene. . . . We tend to view age-related diseases as being related only chronologically. You get them when you get older. But they are actually lumped together at a much more fundamental level, based on what happens in cells when cells age . . . once we can understand their

201

common features, we will be able to intervene and affect all of those diseases. (p. 296) In chapter 7 I will present robust empirical evidence showing that aging is a cell-level phenomenon. In chapter 9 I present an empirically based mechanism for the transition of the cell from health to senescence (see figure 9.6). Therefore, the myriad agerelated losses of function I have described in this chapter must eventually be the outcome of the failure of cell function. A discussion of cell senescence is presented in chapters 12 and 13, and that knowledge will eventually need to be correlated with the information set forth in this chapter. An in-depth analysis of the role that cell senescence appears to play in our organismal aging has been put forth by Fossel (2004), and the interested reader is encouraged to explore that text.

6

Altering Aging Interventions That Modify Longevity and Aging

6.1 Introduction The ability to significantly increase longevity or delay aging by manipulating a particular variable has obvious theoretical and practical interest, for it may be possible to develop a segmental intervention that has a significant impact on the overall aging process. Few of us care much for the more deleterious aspects of aging, so there is a large popular market for the latest insights. Until recently, science and the marketplace offered only palliatives or nostrums. But our biological knowledge of the processes leading to loss of function has broadened greatly, and our ability to intervene in the aging process, although limited, has nonetheless increased dramatically. The current combination of high individual expectations, instantly publicized research, and lack of sufficient knowledge is unstable and leads to situations in which market pressures often result in unprovable and extravagant claims. The best course is to ground our analyses in verifiable animal and human studies and cautiously extrapolate past the data. A discussion of anti-aging interventions is sure to arouse interest. But our enthusiasm for intervention should be tempered by the facts. It is important to understand whether a proposed treatment is based on animal studies only or whether it includes human data as well. And if human data are included, is the evidence anecdotal, or is it of the quality associated with a serious clinical trial? Our emotions often lead us to ascribe more weight to a favorite intervention than it may deserve. So in an effort to help sort

202

out the different types of data, I have divided this discussion of anti-aging interventions into two sections: one dealing with laboratory interventions on animals and the other with interventions tested on humans.

6.2 Experimentally Proven Laboratory Interventions 6.2.1 Caloric Restriction as a Dietary Intervention 6.2.1.1 Effects on Longevity

Caloric restriction has been the single most consistent method of extending life span in vertebrate laboratory animals. The seminal experiments of McCay and colleagues (1934, 1935) grew out of the idea that longevity is inversely proportional to developmental rate. This idea was derived partly from the experimental work of Osborne et al. (1917), which suggested, but did not prove, that underfed rats live longer. McCay and colleagues (1935) demonstrated that rats fed a nutritionally complete but calorie-restricted diet from the time of weaning had significantly increased values of mean, median, and maximum life span when compared to animals fed a normal diet conducive to rapid growth. The animals provided with unlimited calories grew and matured normally. In the restricted group, maturation was greatly slowed, although these animals held their weaning weight and suffered from no other nutritional deficiency, since their diet included adequate

6.2 Experimentally Proven Laboratory Interventions

amounts of protein, vitamins, and minerals. Growth and development in the restricted animals resumed only after they were given additional calories at about 2 years of age. The restricted animals never attained a normal body size or body weight; they remained about 15% smaller than their normal controls. These observations have since been confirmed and extended by many other investigators. The results of a particularly well-controlled experiment are shown in figure 6.1. Examination of such data in terms of Gompertz parameters reveals that calorie-restricted animals live longer and have a longer mortality rate doubling time (MRDT; see chapter 2) than do their ad libitum– fed siblings (Yu et al. 1985; see also figure 2.20). In fact, caloric restriction is the only environmental means that has been shown to significantly slow the mortality rate of any mammal. The basic observation has been found to apply to other species, both vertebrate and invertebrate, and is hallmarked by its ease of repeatability. It is a robust intervention. A review of seven different

studies showed that caloric restriction increased the mortality rate doubling time by about 74%, or, in other words, slowed the rate of aging by that degree (see Finch 1990). 6.2.1.2 Effects on Age-related Pathology

Many studies have shown that dietary history has a major effect on the age of onset and on the incidence of the various age-related pathologies in rodents (see Merry and Holehan 1994b; Weindruch and Walford 1988). (Excellent summaries of the effect of caloric restriction on different aspects of aging can be found in the March 2001 special issue of the Journal of Gerontology.) (http://biomed.gerontologyjournals.org/content/ vol56/suppl_1/index.shtml. First, tissue integrity is maintained well into old age (figure 6.2), and the incidence of chronic tissue inflammations (e.g., chronic glomerulonephritis, myocardial fibrosis, atherosclerosis) and of endocrine hyperplasias is significantly reduced. Second, it appears to be a general but not absolute rule of thumb

100

N/R40 N=60 R/R50 N=56

80

Percent surviving

203

60 NP N=49

40

N/N85 N=57

N/R50 lopro N=56

20

N/R50 N=71

0

0

10

20

30 Age (months)

40

50

60

Figure 6.1 The influence of diet on survival. Each symbol represents an individual mouse. Diet groups are as follows: N/N85, purified diet fed after weaning, at about 85 kilocalories (kcal) per week (reference group); NP, nonpurified diet fed ad libitum at about 113 kcal per week; N/R40, diet restricted after weaning to about 40 kcal per week; N/ R50, diet restricted after weaning to 50 kcal per week; N/R50 lopro, diet restricted after weaning to 50 kcal per week, with a decreased protein content; R/R50, diet restricted before and after weaning. (After Weindruch et al. 1986.)

204 Chapter 6 Altering Aging

Figure 6.2 The influence of diet on tissue structure. Top: A longitudinal section through the gastrocnemius muscle of a control male Wistar rat aged 1010 days. Myofibrillar breakdown is significant; only thin, diffuse Z bands remain to support the sparse, degenerated myofibrils. The sarcoplasm contains few mitochondria, vesicles, and fine filamentous remains. Bottom: A longitudinal section through the gastrocnemius muscle of a food-restricted male Wistar rat aged 1248 days. There is no evidence of myofibrillar breakdown or structural abnormalities in mitochondria or T tubules. Abnormal amounts of lipid were not detected. (From Everitt et al. 1985.)

that tumor incidence is reduced, disease progression is slowed, and onset is delayed. Third, the restricted animals have a greater degree of protection against exogenous carcinogens; these rodents showed significantly fewer tumors after exposure to any of several different carcinogens tested. The mechanism underlying this effect is not yet known with any certainty, but it may involve a decrease in inflammatory processes (see chapter 13), specific sorts of alterations in particular DNA repair enzyme activities or, perhaps more generally, a more rapid mobilization of the body’s detoxification and repair processes (see chapter 7).

Related in some way to the effects on tumors must be the effects of caloric restriction on the immune system. The early effects of restriction seem to depend on the strain, but a general response to the restriction of calories seems to be a decrease in antibody production coupled with an enhanced cell-mediated immunity. Certainly the proliferative response of spleen cells to three commonly used antigens was maintained or enhanced in all ages of the restricted groups, whereas it decreased dramatically in the animals fed ad libitum (Fernandes and Venkatraman 1994)—a result consistent with the treatment’s protective effect against the incidence and prevalence of tumors with age. In addition, autoimmune diseases can be significantly postponed in susceptible strains. A large body of data (reviewed by Finch 1990; Masoro 1988a, 1992a, 2002; Weindruch and Walford 1988) shows that caloric restriction, in addition to having an effect on the age-related pathologies, delays or eliminates the onset of many normal age-related physiological changes. Examples of delayed normal changes range from retardation of the loss of crystallin proteins from the lens, to prevention of the decline in the mouse’s learning ability, to delayed reproductive senescence in female rats. An example of a normal trait that is eliminated in restricted animals is the normal increase in the number of fat cells found in particular fat depots in the rat. Not only does caloric restriction eliminate the increase in fat cells, but it brings about a significant decrease in the fat depot mass as a result of a reduction in the number of fat cells (Masoro 1992). Restricting the amount of fat without restricting the total energy intake did not have this effect. Recent studies suggest that the abdominal fat cells lie at the nexus of a complex signaling network affecting important physiological traits. The variety of traits affected in the different species examined suggests that the dietary restriction is affecting the basic aging process(es) and thus may not be a superficial type of segmental intervention. The same conclusion is drawn from observations on the delayed appearance of strain-specific, age-associated pathologies. Although not age-related pathologies in the strict sense, a delayed onset of puberty and/or an

6.2 Experimentally Proven Laboratory Interventions

inability to maintain normal estrous cycles are often shown by calorically restricted animals and humans. These effects seem to arise as a result of neuroendocrine changes induced by caloric restriction, a topic to which I return in some detail later, when I discuss the evolutionary origins of the caloric restriction effect. 6.2.1.3 Physiological Responses to Caloric Restriction

Caloric restriction must affect some fundamental process(es) involved in the regulation of biological aging. At a minimum, one could hypothesize that the critical variable is the amount of body fat, or the total amount of food eaten, or the total amount of calories taken, or the decreased intake of specific (toxic?) food components such as fats or carbohydrates or proteins, or perhaps more subtle effects, such as the lack of exercise in wellfed laboratory animals. The amount of body fat is not what is important. The mice in one genetically obese strain eat more, gain weight very rapidly, live a shorter time than other mice, and have a high percentage of fat in their body weight (table 6.1). Yet when these animals are calorically restricted, they exhibit a median and maximum life span comparable to that of their long-lived, calorically restricted controls, even though they still have about 3.5 times as much body fat as controls. The increased longevity appears to be related to food consumption as such in these animals and not to body composition.

205

Conversely, the enhanced longevity of dietrestricted animals does not seem to be due to leanness per se. A survey of data from various laboratories has shown that the coincidental losses in body weight that often accompany diet restriction are not consistently related to the effect on the life span (Ingram and Reynolds 1987). Ingram and Reynolds concluded that a curvilinear relationship exists between body weight and life span such that different genotypes will react in different ways to this environmental modulation. For example, if male mice of the B6 inbred strain are subjected to caloric restriction, their mean and median life span values drop by about a third (Harrison and Archer 1987). Thinner is not always better. Furthermore, the diet restriction does not appear to work if it consists of the elimination of any single potentially deleterious component of the diet. The individual restriction of any single food component (such as protein, fats, carbohydrates, fibers, or minerals) to the same extent as observed in the complete diet restriction regime does not markedly affect longevity (Iwasaki et al. 1988; Masoro et al. 1989). It is unlikely that diet restriction experiments extend the life span by reducing the intake of a particular single component of the food. This observation suggests that our life span is not shortened as a result of toxic components in our diet, but it does support the idea that longevity is affected by the daily amount of food (calories) eaten. What about the timing of the restriction? In the early experiments, diet was severely restricted

Table 6.1 Effect of Genetic Obesity and Food Restriction in Aging and Longevity of Mice

Treatment Fed obese Fed normal Restricted obese Restricted normal

Food g/day

Body Wt. g

Fat % Wt.

Immune response of old mice, as % of young value

4.2 3.0 2.0 2.0

59 30 28 20

67 22 48 13

13 49 8 50

Source: data from Harrison et al. (1984).

Collagen denaturation value of old mice, min

Renal function % of young Value

80 52 30 35

113 71 75 86

Longevity Med Max 552 799 814 810

890 970 1300 1280

206 Chapter 6 Altering Aging beginning shortly after weaning. Is this the only effective strategy? The complexity of the process is illustrated by the fact that this lifelong diet restriction regime can be mimicked by manipulation of the preweaning nutrition alone. Rat pups were allowed to nurse at normal or enhanced quantities of milk until after weaning, when both were fed ad libitum. The results suggest that infant overfeeding shortens longevity relative to normal controls. They also suggest that the deleterious effects of this preweaning overfeeding can be partly overcome by daily exercise throughout most of the adult life span. Exercise seems to have no enhancing effect on the life span in these particular normally fed animals, which is consistent with the finding that some ad libitum-fed rats kept slender by exercise show no increase in their maximum life span. This conclusion regarding the deleterious effects of early overfeeding is supported by other studies, such as that done by Stuchlikova et al. (1975), which showed that the maximum extension of life span was observed in animals whose diet (caloric intake) was restricted by 50% throughout their first year of life and who were then fed ad libitum thereafter. Animals given the reverse treatment had a somewhat shorter mean life span, although the life span was still significantly greater than that of animals allowed to feed freely throughout their life. Dietary restriction extends life span to the greatest extent when applied throughout life (Yu et al. 1985). The severe growth retardation that results from the imposition of severe caloric restriction regimes initially made this form of segmental intervention unattractive for human applications. It has been found, however, that a milder caloric restriction is still an effective intervention. This observation applies if the regime is one of moderate caloric restriction that is begun in early life or even in midlife. When the regime is begun in early life, the increased life span appears to be the result of a prolongation of the growth period rather than of the adult period. The data of figure 6.3 demonstrate that a gradual caloric restriction to about 70% of ad libitum calories, started in midadult life, is capable of significantly increasing the mean and maximum life spans of mice.

The success of the intervention is enhanced if the caloric restriction is applied gradually. It is rarely too late to start eating sensibly. An alternative to reducing the daily caloric intake of the animals is to feed them every other day (EOD). This feeding regimen was previously shown to increase longevity in spite of very modest changes in total food intake and body weight (Goodrick et al., 1990). More recent results have shown that these EOD animals show significant changes in insulin sensitivity, which may explain the increase in their longevity (see chapter 7) despite the relatively small weight loss brought about by the EOD regime (Masternak et al., 2005). This regime shows that it is not just the total number of calories that count but that altering the timing of their ingestion also has an effect. This last observation hints at more complexity in the mechanisms regulating CR and the data contained in figure 7.28 is consistent with this view. In addition, some individuals may find an EOD regime more palatable than a severe reduction in calories every day. The mechanisms underlying the effectiveness of caloric restriction are under investigation, and the involvement of the insulinlike signaling pathways (among others) are becoming clearer, as will be discussed in chapter 7. As Masoro (2002) has pointed out, various studies have eliminated two hypotheses regarding the mechanism of action of dietary restriction and forced the reconsideration of a third. First was the idea put forth by McCay et al. (1935) that food restriction increases life span by retarding growth and development. Second was the idea that food restriction increases life span by decreasing body fat. The data in table 6.1 show that this is not the case, although we need to keep in mind that restricted normal animals have fewer fat cells (Masoro, 1992a). The third hypothesis was that dietary restriction increases life span by decreasing the metabolic rate. This idea was particularly attractive because it has an obvious theoretical connection to the oxidativedamage theory of aging (see chapter 10). Recent information suggests that this third hypothesis is too simple to be entirely correct, but it is also not entirely wrong. Dietary restriction does affect metabolism, but not in the simple

207

6.2 Experimentally Proven Laboratory Interventions

B10C3F1

B6

44

(a)

(c)

Body weight (g)

40

36

32

28

Control Restricted

24

0 100

(b)

(d)

Percent surviving

80

60

40

20

0

12

20

28

36

44

Age (months)

48

12

20

28 36 Age (months)

44

48

Figure 6.3 The effect of dietary restriction (to about 50% of ad libitum calories) on life span when started at 1 year of age. Plotted here are the body weights (a and c) and survival curves (b and d) of strain B10C3F1 mice (a and b) and of strain B6 mice (c and d) fed on control or restricted diets. Weights are plotted as means ± standard error. Each point on the survival curves represents one mouse. Note the gradual introduction of the restricted diet, as indicated by the gradual loss of body weight. (After Weindruch and Walford 1982.)

manner envisioned by this theory. Data from the National Institute on Aging–National Center for Toxicological Research (NIA–NCTR) joint biomarker study have shown that caloric restriction induces a major metabolic reorganization in animals (Duffy et al. 1989; Feuers et al. 1991, 1995). This reorganization includes a lowered core body

temperature, a shift away from fat synthesis and toward glucose synthesis, a change in motor activity such that it is concentrated about the feeding time, and an alteration in the body’s metabolic rate such that restricted animals have a lower than normal metabolic rate before feeding but a higher-than-normal metabolic rate after feeding.

208 Chapter 6 Altering Aging One result of such a metabolic shift would be the lowering of the organism’s steady-state production of harmful metabolic by-products that result in oxidative stress and damage (Sohal and Weindruch 1996). This last trait is what we would expect, given the discussion in chapter 7 regarding the effects of up- and down-regulation of the insulinlike signaling pathway in a variety of organisms. Calorically restricted animals seem to be metabolically efficient in additional ways. Let’s consider one example of this metabolic efficiency that may give us an idea of the complexity of the mechanisms involved, while also pointing us in the direction of some likely causal mechanisms (Feuers et al. 1995). Pyruvate kinase is an important gatekeeper enzyme that catalyzes an irreversible ATP-generating step in glycolysis. The enzyme is activated by insulin-induced dephosphorylation and inactivated by glucagon-induced phosphorylation. It is thus controlled by the hormones that regulate glucose metabolism. When pyruvate kinase is activated, even low levels of carbohydrate can be metabolized via glycolysis to yield energy; when pyruvate kinase is inactivated, the pathway is shut down, and the organism must rely on gluconeogenesis for energy production. Young animals can activate or inactivate this enzyme with efficiencies approaching 100%. Old ad libitum-fed animals have lost as much as 90% of this ability. Thus they must use glycolysis to make available for metabolism significantly more carbohydrate and more enzyme if any ATP is to be produced. This results in much wasteful synthesis. Old calorie-restricted animals, however, have a much lower loss of this function and maintain at least 60% of their ability to efficiently regulate pyruvate kinase as they age, so they avoid this wasteful synthesis. It is not just the wasted synthesis that may be important. The ability of calorie-restricted animals to satisfy energy requirements with low levels of blood glucose implies that they can minimize the age-related effects of glycosylation. Maintaining an efficient flow of glucose through glycolysis enables calorie-restricted animals to modulate their NADPH pools better. NADPH cofactors play an important role in activating the expres-

sion of certain longevity determination genes (see the discussion of SIR2 in chapter 7), while also maintaining some of the enzyme systems responsible for the detoxification of free radicals. Thus the ability to maintain “youthful” regulation of this enzyme may spare the organism the harmful effects of glycosylation and free-radical, or oxidative, damage, two harmful processes that can interact synergistically in contributing to the degeneration characteristic of old age (Kristal and Yu 1992). Caloric restriction has been shown to reduce the age-dependent accumulation of advanced glycosylation end products (AGEs) in both red blood cells and skin collagen (Cefalu et al. 1995). In addition, calorie-restricted animals have, in some but not all tissues, a higher level of superoxide dismutase enzyme activity and a lower level of superoxide and/or hydroxide radicals throughout their life span (Lee and Yu 1990). The key element in regulating pyruvate kinase enzyme is the insulin–glucagon hormonal system, a fact that should focus our attention on the effects of caloric restriction (and aging) on the endocrine system (see chapters 10 and 12). The same data should also remind us that the endocrine system regulates many enzyme systems, some of which may be important in the aging process. A multitude of other enzyme reactions are affected by diet restriction, including liver enzymes involved in drug metabolism and elimination (Leakey et al. 1989). The complexity of these changes is illustrated by the observation that DNA repair activity increases in diet-restricted rodents (Lipman et al. 1989), while the same treatment simultaneously decreases both normal DNA synthesis and the binding of a chemical carcinogen to DNA in vivo (Chow et al. 1993). The observation that caloric restriction brings about various alterations in brain neurotransmitters suggests neuroendocrine involvement (Kolta et al. 1989). Finally, persuasive evidence suggests that many other agerelated changes are either slowed or reversed by caloric restriction (Weindruch and Walford 1988). A unifying genetic explanation for these diverse effects is presented in chapter 7. One unexpected beneficial outcome of diet restriction is its effect on learning performance in

6.2 Experimentally Proven Laboratory Interventions

mice (Ingram et al. 1987). Both middle-aged and old mice were tested for their learning abilities in a standard maze test. The control and dietrestricted middle-aged adults had comparable learning levels, as indicated by their number of errors per trial. However, the old diet-restricted animals, exhibiting scores comparable to the middle-aged mice, were clearly superior to the old controls. This study is important because it indicates that the delayed growth and maturation characteristic of diet-restricted animals have no deleterious effect on adult learning abilities but instead maintain these abilities well into the aging process. A robust working hypothesis is that caloric restriction acts by preventing metabolism-driven age-related impairments of neuroendocrine– immune system function, which would result in the maintenance of optimal gene expression patterns in the individual cells of the system (Mobbs et al. 2001). Such a mechanism would indirectly couple food restriction with the senescent processes of the various tissues and organs of the body, possibly by coupling systemically circulated hormonal and/or key metabolite signals to celllevel changes in gene expression patterns. The change in gene expression patterns would then shift the metabolic balances within the organism to a more beneficial state, as just described for the pyruvate kinase enzyme system. This hypothesis implies that caloric restriction acts at the level of the individual cell. Evidence to support this hypothesis was obtained by de Cabo et al. (2003), who showed that the sera of calorie-restricted animals had significantly lower levels of triglycerides, insulin, and insulinlike growth factor 1 (IGF-1) than did ad libitum-fed animals. DeCabo et al.’s demonstration that cells taken from an ad libitum-fed animal and exposed in vitro to sera from a calorie-restricted animal will respond (after a short adaptation period) as if they were calorie-restricted cells, and vice versa, provides strong empirical data supporting the idea that caloric restriction acts at the level of the cell. Aging is basically a cell-level phenomenon. This is consistent with the finding that the skeletal muscle mitochondria of aging, calorierestricted rats has a 23% lower rate of proton leakage across the mitochondrial membrane relative

209

to controls (Lal et al. 2001). A detailed discussion of mitochondrial mechanisms is presented in chapter 11, but for the moment note that oxidative damage is the major senescent process, and the mitochondria are the major source of such reactive oxygen species (see chapter 10 for a fuller discussion of this concept). Caloric restriction acts to reduce the oxidative stress levels within each cell. All our interventions operate so as to in some way shift the cell’s gene expression pattern from that characteristic of a senescent pattern to that characteristic of a healthy pattern. It therefore follows that the shift from the health span to the senescence span should be accompanied by preceding shifts in the cell’s signal transduction pathways. Robust data supporting these concepts are presented in chapters 7, 12 and 13. 6.2.1.4 Genetic Responses to Caloric Restriction

The first report that caloric restriction altered gene expression was that of Richardson and collaborators (1987), who showed that the protein levels, mRNA levels, and nuclear transcription rate of a specific protein in rats are significantly enhanced in restricted animals relative to the agematched ad libitum-fed controls. Because protein synthesis and gene expression often decrease with age, the finding that caloric restriction might reverse this process was of special interest. This topic is discussed in more detail in chapter 7. 6.2.1.5 Evolutionary Origins of Caloric Restriction

Clearly, caloric restriction works (Gerhard 2001). But why should almost all species tested, vertebrate and invertebrate alike, come equipped with a mechanism that enables them to live long if they stay hungry? What is the evolutionary sense behind this concept? One proposal suggests that caloric restriction is best viewed as a special application of the disposable-soma theory (see chapter 4), which is based on the premise that an organism can devote its excess calories, beyond the amount needed for basic and essential functions, to reproduction and/or somatic maintenance. In

210 Chapter 6 Altering Aging this view, caloric restriction evolved as the set of mechanisms by which an organism adjusts its reproductive strategy to the conditions of its environment by shifting from rapid reproduction over a short time period to a reduced rate of reproduction over a longer life span (Holliday 1989; Kirk, 2001; Richardson and Pahlavani 1994). Computer modeling of “virtual mice” suggests that apportioning energy in this manner is a viable strategy (Shanley and Kirkwood 2000), as does the mathematical modeling of real flies (figure 4.5). How might an organism accomplish this change? The data in table 6.2 are instructive. Caloric restriction has a major effect on the endocrine system, causing rapid and significant decreases in the plasma levels of most hormones. Many of the changes recorded in table 6.2 are consistent with the idea that the affected animal is shifting resources away from growth and reproduction and toward survival. The brain is generally believed to have a central role in regulating both appetite and reproduction, probably via certain integrative centers in the hypothalamus and limbic system (Finch 1990). Therefore, the effects of caloric restriction on life span and reproduction likely begin via an unknown sensory input to the integrative centers of the hypothalamus and are probably transduced into hypothalamic–pituitary–endocrine organ trophic signals or changes in gene expression, which manifest in hormonal alterations such as presented in table 6.2. These hormonal alterations

then probably bring about coordinated alterations in gene expression in affected target cells. From this point of view, we might consider the integrative centers of the brain to be acting as if they were the pacemaker regions for senescence and longevity in vertebrates (Finch 1990). In fact, the genetic data presented in chapter 7 not only support this concept, but also reveal the identity of the genetic pathway that transduces nutritional changes into gene expression changes enhancing either somatic maintenance or reproduction (see table 7.12 for summary). Masoro and Austad (1996) presented a scenario for the evolution of the caloric restriction effect based on data, suggesting that short-term unpredictable food shortages bring about changes in the glucocorticoid system and, through it, the systemic alterations characteristic of caloric restriction. Elevations in the glucocorticoid levels are usually indicative of stress and may have deleterious effects on the organism, as I discuss in chapter 13. However, it is well known that exposing animals to transient or low levels of certain environmental stresses can have a beneficial effect on the organism’s age-specific mortality rate (qx) and/or longevity. Masoro (1998, 2002) hypothesized that the caloric restriction effect may well depend, in part at least, on a hormesis effect. Hormesis refers to situations where an organism exhibits a maximum fitness as a consequence of exposure to some low level of chemi-

Table 6.2 Effects of Caloric Restriction (CR) Versus ad Libitum feeding (AL) on Hormone Changes in Rhesus Monkeys Hormone

Effect of CR vs. AL

Comments

Growth hormone Insulinlike growth factor-1

Normal age-related decline Normal age-related decline

Thyroid hormone

CR CR CR CR

Follicle-stimulating hormone Leutinizing hormone Progesterone Estradiol Testosterone Dihydroepiandrosterine Melatonin

CR = AL CR = AL CR = AL CR = AL CR = AL CR > AL CR > AL

= < = <

Source: Data from Mattison et al. (2001).

AL AL young animals AL middle aged and older AL

Normal age-related difficulty in T3 to T4 conversion Normal age-related increase Normal age-related increase Normal age-related increase Normal age-related decline Normal age-related decline Normal age-related decline Normal age-related decline

6.2 Experimentally Proven Laboratory Interventions

cal or environmental stress that would be harmful or even lethal at high doses. There is no evidence of which I am aware linking hormesis to the caloric restriction effect in invertebrates. This does not invalidate Masoro’s (1998) view, for it might be the case that mild levels of glucocorticoid stimulation assist the serum factors (insulin, IGF-1, etc.) that modulate the caloric restriction effect at the cell level. The greater complexity of the mammal relative to the fly makes it likely that multiple complementary signals are involved. If this general view of the process is correct, one might predict that the only species that will show an increase in their life span as a result of caloric restriction are those that show a shift in their reproductive strategies in response to a decrease in calories. There are no firm data on this point for humans, but it is interesting to note that well-nourished women who engage in hard physical exercise often stop menstruating. In rural China, where caloric intake is often restricted, the age at menarche is approximately 18 years (Willett 1994). This time of onset represents a significant delay when compared to the age of about 13 years in U.S. women. Invertebrates such as the nematode and Drosophila also respond to caloric restriction. A most striking demonstration of this phenomenon was done by Mair et al. (2003). They first established that, as expected, wild type Drosophila had different and characteristic life spans and patterns of age-specific mortality rates (qx) on normal and restricted diets. Mair et al. then transferred adults from one dietary regime to the other at different ages. (Unlike the mice of figure 6.3, the flies do not have to be gently transferred from one regime to the other; perhaps this simply reflects the greater complexity of mammals relative to flies.) They found that within a few days after the dietary shift, the flies would express the qx characteristic of their new nutritional regime. The qx, in other words, would flip from the normal state to the restricted state, or vice versa. That flip, in turn, probably represents the outcome of the gene interaction network shifting from a nonoptimal to an optimal state, or vice versa. There is no cell memory of its prior

211

aging state. Why, then, do the animals transferred to a caloric restriction regime at an older age not show the same life span extension as do young animals? I suppose this may have to do with the accumulation of prior unrepaired damage, which prevents the gene interaction network from fully expressing the optimal expression patterns. In any event, this finding does show that it is never too late to adopt a longevityextending intervention. This caloric restriction effect is not limited to flies. Austad (1989) showed that restricting the number of fruit flies that certain spiders were allowed to eat significantly increased their life span. Taken together, then, the results obtained from vertebrates and invertebrates strongly suggest that diet restriction is an effective intervention across a wide variety of animal species. This conclusion is supported by the highly conserved nature of the gene pathways responsible for the nutritional effect, as noted in chapter 7. 6.2.1.6 Toxicity Studies and Caloric Restriction

Chemicals that have the potential to affect human health are routinely tested for their toxicity in rodents, and such tests play a key role in the regulatory process for new medicines, food additives, and other chemicals designed for human consumption. The animals used in these tests must be in good health at the beginning of the tests. Over the years, the rodents used in such tests have usually been allowed to feed ad libitum; moreover, they have been selected for fast growth and reproduction. The application of these evolutionary principles has led during the past several decades to the expected outcome; namely, the rodent models used for testing have shown a progressive increase in their mortality and morbidity. These alterations have brought about a possible confounding of toxicity data with the effects of diet on longevity. Investigators recognizing this problem have incorporated the control of dietary intake into the experimental design and conduct of animal studies, and they now consider the role of caloric restriction in carcinogenicity, toxicity, and pharmacology studies (Hart et al. 1995).

212 Chapter 6 Altering Aging

6.2.1.7 Caloric Restriction and Reproduction

Severe or poorly controlled caloric restriction of a pregnant animal leads to malnutrition of the fetus, with stage-specific deleterious effects on its longevity and other traits (see figure 3.14). But how does caloric restriction affect the reproductive behavior and performance of nonpregnant animals? The answer appears to be speciesspecific (Schwetz 1995). No significant effect on reproductive performance was noted in SpragueDawley rats under 10–20% restriction, and only minor effects were noted at 30% restriction. Swiss mice are another matter. Significant effects, including a reduction in the number of live pups per litter, were observed in both genders and at all levels of restriction. This strain specificity implies that caloric restriction, even though it affects almost all species tested, cannot be viewed as a universal intervention. Within any one species, individuals with certain genotypes may react negatively to this intervention. 6.2.1.8 Caloric Restriction Mimetics

Serious efforts are now underway to develop pharmaceutical interventions that mimic the effects of caloric restriction on aging. A recent review of the topic (Masoro 2002) listed a number of different compounds claimed to mimic caloric restriction either in whole or in part (table 6.3). These are only the compounds available to the public and do not include any of the several candidates undergoing testing by various biotech and pharmaceutical firms. The antioxidants have use-

ful properties, but they mostly appear to be at best only segmental mimics in that they do not bring about the profound metabolic reorganization characteristic of caloric restriction. The glucoregulatory drugs seem to be a better bet (Mattson et al. 2001a). Indeed, high doses of metformin are reported to induce gene expression changes in the rat liver characteristic of those induced by caloric restriction and to also induce about a 20% increase in median life span (Spindler 2003). That finding, if verified, is proof that pharmaceutical interventions can act as a caloric restriction mimetic and significantly extend mammalian life span. Given the extensive physiological data suggesting that glucoregulatory processes lie at the heart of the caloric restriction effect, and given the robust genetic data identifying the insulinlike signaling pathway as a major longevity-determinant mechanism, it seems reasonable that manipulating the glucoregulatory apparatus will be a characteristic of any effective future mimetic. In the meantime, we should monitor our blood sugar and insulin levels by other means.

6.2.2 Manipulations of Metabolic Rate in Laboratory Animals The data presented in chapter 4 suggest that one reason for the interspecific differences in life spans might involve differences in basal metabolic rates; mammalian species with high metabolic rates tend to have short life spans and vice versa. Within any single species, however, there is often

Table 6.3 Compounds Thought to Decrease Cell Damage Due to Aging Antioxidants

Glucoregulatory agents

Deprenyl Vitamin E Glutathione Spinach or blueberry extracts a-Lipoic acid Acetyl-L-carnitine Coenzyme Q10 Melatonin Phenylbutylnitrone

2-Deoxyglucose Phenformin Metformin Chromium picolate Dihydroepiardrosterone (?) Antiglycation agents Aminoguanidine

6.2 Experimentally Proven Laboratory Interventions

140

Females

Males

120

Maximum life span (LT90) (days)

extensive individual variation in life span. It would be interesting to know how much, if any, of this individual variation is due to metabolic differences. The usual experimental procedure is to alter the metabolic rate of the animals and then to measure life span. Altering metabolic rate in homeotherms is a difficult task because their body physiology is designed to yield a body temperature and basal metabolic rate that is relatively independent of the environment. Temperature manipulation as a way of altering life span is much more easier in a poikilothermic animal, whose body temperature is not physiologically preset, than in a homeotherm. Loeb and Northrop (1917) were the first to demonstrate that life span is markedly increased in Drosophila raised at low temperatures (about 16°C) but markedly decreased in Drosophila raised at higher temperatures (28°C). This observation has been confirmed many times since Loeb and Northrop’s investigation. The data in figure 6.4 show that flies responded in the expected manner to temperature, living significantly longer at low temperatures than at high temperatures, but these data also show that temperature manipulations of the normal-lived strain cannot overcome the inherent genetic factors responsible for the extended longevity of the long-lived strain. Temperature can modulate but cannot override the effect of those genetic pathways summarized in table 7.12. The finding that the metabolic rate of a poikilotherm depends on the ambient temperature became one of the key observations underlying Raymond Pearl’s “rate of living” theory, which he put forth in 1928. This concept originally stated that there is an inverse relationship between metabolic rate and aging (see chapter 11 for discussion). The problem with this approach is that much data suggest that the effect of temperature on metabolism involves more than a simple alteration of metabolic rate. The L and R strains depicted in figure 6.4 have the same metabolic rates but very different longevities (see figure 7.14), a finding that is generally true of other strains as well (Van Voorhies et al. 2004). Another problem with this rate of living concept is that the animals raised at lower tempera-

213

100

80

60 IL

IL EL IR ER

40

18

22

25

28 18 Temperature (ºC)

EL IR ER

22

25

28

Figure 6.4 The effects of ambient temperature on the maximum (LT90) adult longevity of a long-lived (L) strain and a normal-lived control (R) strain of Drosophila. The strain-specific longevity is plotted against the particular temperature at which each population was raised. Each population was further subdivided according to what portion of the life span was spent at the indicated temperature: E individuals spent both their developmental and adult life at the indicated temperature; I individuals were allowed to develop at 25°C before being shifted to the indicated temperature. Note that (1) developmental temperature has little effect on the life span within each strain; (2) there is not much difference between males (left side of figure) and females (right side); (3) lower adult temperatures greatly increase life span within each strain; and (4) this environmentally mediated extension of life span does not overcome the genetically mediated superiority of the L strain. (After Arking 1988.)

tures usually show an extension of every phase of their life cycle, which is why this treatment has been called the rubber band mechanism of life extension. Thus, although the animals’ chronological age was greatly extended, it is not at all intuitively clear that the animals’ biological age would also be extended. This ambiguity is simply another reflection of the fact that chronological time is a poor measure of biological age. Finally, Lints (1989) reviewed the data supporting the rate-of-living theory and concluded that

214 Chapter 6 Altering Aging it rests on a weak theoretical basis and is not supported by several thorough studies. Perhaps because of the abstraction of data, it is easier to draw conclusions when comparing metabolic rates across species than within species. Cutler (1982) plotted the maximum (LT90) life span and the daily metabolic rate for 77 mammalian species (figure 6.5). In general, these interspecific comparisons support the idea of an inverse relationship between LT 90 and daily metabolic rate, as first stated by Pearl (1928). Cutler (1984) interpreted the data as showing that every species has a lifetime energy potential (LEP), which is a measure of the amount of energy that individuals of that species will, on average, expend per gram of body weight during their life span. In other words, each species has a fixed amount of energy to expend on living. Cutler’s statistical analysis shows that these 77 species can be statistically divided into three groups with different LEP

Maximum life span (LT90) (years)

AB C

values. But there are obvious exceptions. Humans, for example, are way off the scale, having a maximum life span that is twice as high as other animals with a comparable LEP (e.g., orangutans and red deer). Note that the data of figure 6.5 yield three identifiable hyperbolas. Because hyperbolas approach their limits asymptotically, at low metabolic rates a small difference in the metabolic rate may be associated with a large difference in life span. In the same manner, a small difference in longevity may be associated with a large difference in metabolic rate. Despite these caveats, the data in figure 6.5 show that, in general, mammals show an inverse relationship between life span and metabolic rate. Does this prove to be the case? The relationship may be real, but it need not be causal. Our review of the Drosophila data (see figure 11.1) illustrates that the maximum life span can be significantly altered without any measurable change in metabolic rate. Thus, this species at

Human Primates Artiodactyla Carnivora Perissodactyla Proboscidea Hydrocoidea Lagomorpha Rodentia Insectivora Marsupialia

African elephant

80

Hippopotamus Orangutan Horse Gorilla Red deer Capuchin

60

40

20 Kinkajou Genet

Rhesus monkey Lemur Gibbon Cat Owl monkey Marmoset Tree shrew Tarsier

Field mouse

Pygmy shrew

Gerbil

Rat

C B A

Opossum 0

0

40

80

120

160

200

240

280

Specific metabolic rate (cal/g body weight/day)

Figure 6.5 The total amount of energy estimated to be used by each of 77 mammalian species during their lifetime, based on data taken from the literature (see Cutler 1984 for references). Each symbol represents the intersection of the estimated maximum life span (LS) and specific metabolic rate (SMR). There is a strong inverse correlation between these two variables. Note that the data can be grouped into at least three similar but different categories according to lifetime energy potential (LEP), which is calculated as LS ÷ SMR. The LEP values for the three groups shown here are as follows: 220 kcal/g for most nonprimate mammals (curve A), 458 kcal/g for most primate species (curve B), and 781 kcal/g for humans, capuchins, and lemurs (curve C). (After Cutler 1983a.)

6.2 Experimentally Proven Laboratory Interventions

least either has no fixed LEP, or the LEP value must be very plastic. If the LEP were fixed, we would have a difficult time explaining the decrease in mortality rates with increasing age in the medfly or in humans (see chapter 2). After all, to claim that the LEP is fixed is tantamount to claiming the existence of a fixed species-specific maximum life span. Much of the data presented in chapters 2 and 3 argue against such a conclusion. Austad and Fischer (1991) suggest that the LEP theory is fundamentally flawed and that the explanation lies elsewhere. The fact that larger animals live longer than smaller animals may be a consequence of the survival value of increased body size, not an effect of decreased metabolic rate. This explanation allows us to rationalize the very long life spans of birds and bats (see chapter 4), which have very high metabolic rates per unit mass, and understand them as arising as a consequence in part of the survival value of their lifestyles. The high longevity of large terrestial mammals can then be understood as a consequence of reduced environmental vulnerability rather than decreased metabolic rate per unit mass. Not many predators want to irritate an elephant. Finally, the difference in longevity between island and mainland varieties of the opossum can be best explained in terms of their difference in body size (Austad 1993). In effect, Austad and Fisher argue that we have been looking at the wrong variable. Austad and Fischer’s (1991) explanation lets us resolve the apparent contradictions in the metabolism data simply by shifting our approach to an evolutionary point of view and, in so doing, rediscover that an animals’ life-history strategy plays an important role in determining its life span. But this does not mean that metabolism plays no role. It is simply not possible to understand the long life of birds, for example, without also taking into consideration the fact that they exhibit a decreased amount of oxidative damage because of their low rates of mitochondrial free-radical production (Herrero and Barja 1997b; Sohal and Weindruch 1996). Clearly, although we do not yet fully understand the role of metabolic rate in regulating life span on either an interspecific or intraspecific physiological basis, we can nonetheless conclude that its role is perhaps best understood if we real-

215

ize that the metabolic rate is linked to the levels of oxidative stress suffered by the organism over time, and that a full explanation of the organism’s life span is likely to involve factors other than metabolic rate.

6.2.3 Exercise in Laboratory Animals Numerous studies have investigated the effects of exercise on laboratory rodents. The initial impetus for these investigations was to test the prediction of the rate-of-living theory that an intervention such as exercise will use up calories, accelerate aging, and decrease longevity. Early studies (Slonaker 1912) showed that exercised rats died younger than sedentary controls, but more recent studies do not show such an effect. It is believed that the earlier studies did not use pathogen-free rats and that what was observed was that exercise stress worsens chronic infectious diseases of these rats and hastens their death (Holloszy and Kohrt 1995). One complication in conducting exercise studies on rats is that exercised rats tend to lose weight, and their growth may be stunted. Thus, the experiment must be very carefully controlled to ensure that the effects of exercise are not being confounded with the effects of caloric restriction. When these conditions are met, the data show that lifelong exercise improves the mean life span but has no effect on the maximum longevity, at least in some strains (Holloszy 1993). In these experiments, rats subjected to both caloric restriction (by about 30%) and exercise showed the expected increase in maximum longevity ascribed solely to the dietary intervention. In other words, exercise by itself does not increase the maximum life span, but it doesn’t interfere with the one intervention that does. Other studies, such as that by Goodrick (1980), suggest that both mean and maximum life spans may be extended by exercise. The reason for the disagreement is not known but may reflect the properties of the different mouse strains used. Goodrick’s study also showed that when a regular exercise regime is begun early in life, it appears to increase both the mean and the maximum life span. Exercise regimes begun late

216 Chapter 6 Altering Aging in life have variable effects on life span values, suggesting the existence of some sort of threshold age (Edington et al. 1972). An alternative hypothesis, given that older rats do not exercise as hard as younger rats, is that there is a level of exercise below which no beneficial effect may be detected. Another study addressed the effect of reduced physical activity on the longevity of female mice (Mlekusch et al. 1996b). The inactive animals were housed in small cages constructed to permit little physical activity. These mice were compared to littermates housed in large cages equipped with running wheels and other aids to physical activity. The inactive group showed an 11% reduction in longevity, despite a significant voluntary decrease in food intake. The excess mortality was found mostly in the senescent phase (≥LT 40), which suggests that the lack of exercise accelerated the senescent processes. However, the same group also reported that the life-shortening effects of reduced physical activity were abolished by feeding the mice a fat-rich diet (Mlekusch et al. 1998). This regime resulted in a decreased mortality throughout the life span, suggesting that diet affected both longevity determination as well as senescent mechanisms. The effect was attributed to the sparing effect of lipids on cortisol secretion by the adrenal gland. Stress mediated through cortisol secretion is known to reduce longevity in rodents (see chapter 13). In contrast to the fat-rich diet, a high-glucose diet decreased longevity throughout the life span (Mlekusch et al. 1996a). Moderate caloric restriction in the absence of physical activity appears not to be fully beneficial. Another study examined the effects of lifetime exercise on the thermal stability of the rat’s tail tendon (Viidik et al. 1996). The results are consistent: Exercised animals have thermal-stability values characteristic of chronologically younger animals. If begun early enough, exercise also seems to have beneficial effects on the normal age-related decrements in the physiological functioning of various organ systems in the mouse, such as a slowing of collagen cross-linking in the tail tendon, improved cardiac contractability, and pre-

vention of decline in the energy stores of skeletal muscle. Perhaps the most fascinating benefit of exercise is the effect on the neurotrophic factors of the brain. Neurotrophic factors protect neurons from damage, promote their growth and function, and generally serve in maintaining the long-term plasticity of the nervous system. Brain-derived neurotrophic factor (BDNF), nerve growth factor (NGF), and fibroblast growth factor are common neurotrophic factors. BDNF and NGF are expressed at high levels in the cerebral cortex and hippocampus, areas important to memory and higher mental functions. In rats, even short-term exercise significantly increases BDNF mRNA in these brain areas (Cotman and Neeper 1996). In fact, there is a tight correlation between the distance the animal runs and the level of BDNF mRNA detected in its brain. This is not an isolated effect; similar stimulatory effects of exercise have been noted in other neural characters (see Cotman and Neeper 1996). As Cotman and Neeper point out, the exercise-dependent increase in these neural factors may provide a fundamental molecular mechanism for use-dependent changes in the cellular structure of the brain, and the resulting enhanced neurons may strengthen the initiating exercise behaviors. The increased resistance to insulin often associated with advancing age can lead to deleterious effects such as insulin resistance and hyperinsulinemia. But this progression is not inevitable and may be altered by diet and exercise. Exercise has been shown by many studies to maintain the insulin sensitivity of skeletal muscle and thus enhance insulin sensitivity, but just how this was done was not clear. The quantitative changes in four key proteins of the insulin signaling system were measured in young and old trained and untrained rats (Arias et al. 2001). The four proteins involved were insulin receptor (IR), insulin receptor substrate (IRS) proteins, phosphatidylinositol 3-kinase, and protein kinase B. These proteins are believed to be present in approximately equal concentrations for optimal functioning of the system. IR normally increases with age, whereas IRS normally decreases with age. This increased disparity between

6.2 Experimentally Proven Laboratory Interventions

their relative levels may play some role in bringing about insulin resistance. However, exercise increases IR in young but not old rats, while exercise increases IRS in old but not young rats. The net effect of exercise appears to be to return the system to an optimal functional state in both young and old animals, thus enhancing insulin sensitivity in an age-appropriate manner. Exercise seems to have no effect on several other age-dependent factors, most notably the number or synaptic activity of neuromuscular junctions in skeletal muscle. The reduced capability of old animals to adapt to vigorous exercise may be due to this unrelieved age-related decline. Other animal studies seem to indicate that exercise can have both positive and negative effects, but that long-term training begun no later than midlife maximizes the beneficial effects of exercise.

6.2.4 Vitamin and Antioxidant Treatment There is little, if any, evidence showing that life span can be extended by vitamin or antioxidant supplementation. There is, however, evidence showing that morbidity may be significantly decreased or delayed as a result of such treatment (Meydani 2001; Packer 1991). The liver mitochondria of rats fed a vitamin E-deficient diet have a structure and functional level that closely resembles those of older animals (Armeni et al. 2003). Vitamin E supplementation reduces the oxidative stress markers in the plasma of old rats and reduces the time required for complete wound healing (Khodr et al. 2003). Vitamin E is a lipophilic molecule and preferentially accumulates in membranes. Its accumulation in mitochondrial membranes probably accounts for its ability to significantly reduce reactive oxygen species in heart mitochondria in rats given large doses of the vitamin over a short-term (2 months) period. These antioxidant effects likely arise from the role of vitamin E as a chain breaker in the membrane, and this effect seems to affect more processes than one might first think. Inflammation is now known to be a major senescent process in that the associated high levels of oxidative

217

stress trigger different deleterious processes leading to cell death (chapter 13). The antioxidant effect of vitamin E protects the cardiac and immune systems from these effects of oxidative stress. Work with animal models of neurodegeneration suggests that reactive oxygen species are involved at several key points in the apoptotic cascade and that vitamin E can protect rat brain cells in culture from toxin-induced apoptosis (Gonzalez-Polo et al. 2003).

6.2.5 Genetic Manipulations in Laboratory Animals The selection and mutagenesis experiments done on fungi, nematodes, and fruit flies and described in chapter 7 have conclusively demonstrated that longevity is under genetic control. This finding opens an avenue for potential interventions. Manipulation of the genome produces statistically significant increases and decreases in the life span of these laboratory organisms. Will genetic interventions really become practical or widespread anti-aging interventions? I do not believe they will. No one is suggesting that genetic alteration of our germ cells would have any effect on our aging, so that is not on the table. Genetic alteration of our somatic cells has been offered as being a viable intervention. I doubt it. Although somatic gene engineering may have limited usefulness in alleviating those genetic disease due to the loss of function of one specific gene in a specific tissue, there is no reason to believe that such a procedure will have any effect on the aging process. Nor could it, if aging is a cell-level phenomenon as noted by Mair et al. (2003) that is simply triggered by neuroendocrine signals initiated by our diets. Unless one can envision some way to change the relevant gene in every cell of the body, and to do this without inducing any deleterious side effects, then there is no theoretical way to use gene therapy to increase longevity in normal individuals. The real value of the genetic studies being done on laboratory models is not to pioneer new methods of engineering human genes, but rather to allow investigators to efficiently dissect the pathways and figure out

218 Chapter 6 Altering Aging how they work. Only when armed with that knowledge will it be possible to develop effective pharmaceutical interventions that can extend longevity. Several examples of such pharmaceutical interventions are presented in chapter 7, and they serve as the proof of concept for this line of reasoning. I discuss the whole idea of anti-aging interventions in chapter 15.

6.2.6 Transplantation of Tissues and Organs in Laboratory Animals Transplantation of aging tissues from a young organism to an older organism was once touted as an aging intervention for humans; we now classify it in the category of “unlikely interventions” (which I discuss at the end of this chapter). 6.2.6.1 Stem Cells and Therapeutic Cloning

Much has been written in the media about stem cells and cloning, but much of that coverage is inaccurate at best. Cloning of plants was first done in 1947 by Steward (see Steward, 1970), and of frogs in 1952 by Briggs and King, but the lack of public interest in these non-mammalian experiments ensured that only developmental biologists were passionate about the data. However, the cloning of Dolly the sheep in 1997 grabbed the public interest, and the issue is here to stay. Cloning of large mammals has agricultural and pharmaceutical implications (Heyman 2005). The safety and ethical problems inherent in the procedure indicate that it has little if any use as a reproductive protocol for humans. However, what does have medical potential is obtaining stem cells from human embryos or adults for what is called therapeutic cloning. Embryonic stem (ES) cells can be obtained from human embryos left over from in vitro fertilization procedures which the parents do not want and which are presently under indefinite storage in the freezer. The hope is that these stem cells can be grown in culture and induced to form large quantities of young differentiated cells, which can then be used as cell transplants for ill adults. People with degenerative brain disorders such as Alzheimer’s disease or Parkinson’s disease

could have new neurons transplanted into the appropriate areas of their brain so that these new cells could replace the ill cells and restore function. Other types of stem cell derivatives could be used for other diseases that arise from defective or senescent cells. Hwang et al. (2005) have demonstrated that it is possible to remove the nucleus from a donated human ova, replace it by transplanting a nucleus from the cell of an particular individual, and thus create a line of embryonic stem cells genetically identical to the nuclear donor. Such cells could be used to create genetically identical replacement tissues which could be transplanted into ill individuals without the need to use immune-suppressing drugs. Laboratory studies support the validity of this concept. Adult stem cells might be obtained from multipotent cells present in most organs and these could, in theory, be manipulated in culture to generate young cells of the particular tissue type needed by the donor. Because the cells would usually be transplanted into the donor, there would be no need for immune suppression drugs, which would be a major advantage with this protocol. However, adult stem cells can be obtained only by difficult and not always successful invasive procedures. But the point is that use of ES or adult stem cells is a scientifically rational procedure of using young cell transplants to alleviate or cure a varied number of age-related diseases. Because stem cells of either type are not available for therepeutic purposes, clinical researchers have turned their sights toward using differentiated cells, usually from some other part of the patient’s body, as cell transplants for the diseased organ (Couzin et al. 2004). Hundreds of patients are enrolled in the current cell-therapy studies now underway around the world. Both beneficial and harmful responses have been obtained. The former likely verify the validity of the basic concept that age-related diseases may best be treated by young cells; while the latter likely suggest that much more work needs to be done on this topic. The perceived correctness of the concept will continue to drive this field of research. I discuss anti-aging interventions in chapter 15.

6.3 Useful Methods of Modulating Aging Processes in Humans

6.3 Useful Methods of Modulating Aging Processes in Humans Manipulating the life spans of mice and rats and flies is all very interesting, but what, you may ask, are real people to do in the here and now? Are there any segmental interventions to the aging process that an ordinary person can practice without withdrawing from the modern world or without becoming a guinea pig for an unproven regime? Strategies that fulfill these criteria do exist. In one sense, they are based on recently confirmed scientific investigations; in another sense, they date from the origins of Western civilization. As the Greeks put it: “a sound mind in a sound body,” and “all things in moderation.” There are no magic bullets that can stop aging in its tracks, but there is also no biological necessity to become decrepit.

6.3.1 Avoiding Premature Death Though this suggestion may seem obvious, the first thing to do to age successfully is to avoid dying prematurely. Engaging in activities that carry a high risk of catastrophic injury or death, as is the wont of teenage boys, decreases the prob-

219

ability of surviving to middle age (see figure 8.3). More important but not as obvious is the necessity of avoiding behaviors that will bring on progressive disability and lead to death in middle age or early old age such as smoking, drug abuse, HIV infection, and so on. Consider the data presented in figure 6.6, which tabulates the factors contributing to the causes of death in the U.S. population. Tobacco is implicated as a causative agent in about 40% of all deaths. Its role in bringing about heart disease and cancer, which together account for 57% of all deaths, is well documented. Tobacco use is an addictive but otherwise wholly avoidable highrisk behavior. The conclusion seems obvious: If you smoke, you are significantly increasing your risk of having a premature and unpleasant death. In fact, the single most important life extension step is to eliminate the use of tobacco. We will all die of something, but delaying the age of onset of major diseases will inevitably lead to an increase in the mean life span (see figure 2.14), to say nothing of improving the quality of life. But smoking is not the only culprit. Note that poor diet and/or lack of exercise are listed as a contributory cause of death in almost 30% of the deaths (figure 6.6). I discuss the interventions of restricting diet and of exercising in the next section of this chapter. The point to be garnered

Tobacco

400

Diet/activity patterns

300 100

Alcohol Microbial agents

Figure 6.6 Contributory causes of death in the United States, 1990 estimates. (From G. T. Baker III and J. Frozard, Unpublished data, Gerontology Research Center, National Institute on Aging.)

90 60

Toxic agents Firearms

35

Sexual behavior

30

Motor vehicles

25

Illicit use of drugs

20 0

100 200 300 Number of deaths (thousands)

400

220 Chapter 6 Altering Aging from these data is that you are one of the most powerful influences on your own rate of aging. I am nonplussed when friends and students ask me about anti-aging interventions while holding a cigarette or drinking a high-sugar soda. What’s the point of hoping to live a long life if your own behaviors make it likely that you will die prematurely?

6.3.2 Physiological Interventions Suitable for Individuals 6.3.2.1 Nutrition Data in Humans Animal studies indicate that caloric restriction is the single most effective intervention known. Although rigorous proof of the efficacy of diet restriction in humans is not yet available, enough evidence has been gathered to strongly suggest that dietary restriction would have similar effects in humans. No formal, well-controlled long-term studies dealing with the effects of caloric restriction on humans has yet been completed. The severe malnutrition too often practiced on prisoners and refugees in time of war clearly has devastating short- and long-term effects on the health of these people (Mohs 1994a), but such data cannot be used as evidence one way or the other in this question. There is, however, some persuasive empirical and anecdotal evidence supporting the benefits of caloric restriction. First, in the past, the caloric intake of much of the population of Okinawa was much lower than the norm in Japan, but the nutrition of the Okinawans was otherwise adequate. Okinawa has a high incidence of centenarians: 2–40 times as many as on any other Japanese island. Other anecdotal evidence suggests that few, if any, centenarians or other long-lived people have been obese. Second, the seven people who voluntarily entered Biosphere 2 for 2 years and reduced their caloric intake while there are reported to have shown physiological changes similar to those observed in calorically restricted rodents (Walford et al. 2002).

Third, although it has now been established that the populations of Soviet Georgia, Ecuador, and Pakistan have no super-long-lived individuals who live past the maximum human life span of 120 years (as discussed in chapter 8), the fact remains that some of these populations contain relatively large numbers of physically active individuals of advanced age. Analyses of the traditional diets of these populations suggests that they are generally low to moderate in calories and that complex carbohydrates are the primary source of energy (Schlenker 1984). The intake of proteins and saturated fats varies considerably, from very high and based on animal foods (meats and dairy products) among the Georgians, to very low and based on vegetable foods among the Hunzas (in Pakistan) and the Vilacambians (in Ecuador).The latter two groups do not have an absolutely assured food supply and are probably slightly underfed every winter. Aged individuals in all three societies are usually slim. The results of studies on the food-restricted animals lead us to suspect that their moderate caloric intake, continued physical activity, and avoidance of weight gain may play major roles in the enhanced longevity and delayed aging observed in these populations. Evidence for this position is now available. Fourth, the “Mediterranean paradox” refers to the relatively long life spans of the Greeks and Italians, despite their high total fat intake and high rates of smoking. The explanation of this paradox is believed to be the Mediterranean diet, which is high in fruits, vegetables, cereals, and legumes, but low in animal fats, and which enables the people who eat it to avoid cardiovascular disease. Red wine, which is a common drink in these societies, contains a chemical (revastrol) that activates the Sir2 gene and the genetic pathways enhancing longevity (see chapter 7). Retrospective studies of elderly rural Greeks suggest that those who followed the whole traditional diet were only half as likely to die as were those individuals who adhered to only a portion of the traditional diet. Thus each of these societies may have found their own versions of a healthy, lifeenhancing diet. Fifth, an 11-year follow-up study of a group of German vegetarians revealed that the mortal-

6.3 Useful Methods of Modulating Aging Processes in Humans

of oxidative damage. The individuals involved in this study were on caloric restriction for 3–15 years; an analysis of their medical records suggest it takes at least 1 year for the calorie-restriction regime to exert its effects. The observed changes are in the same direction as predicted by the earlier rodent and monkey studies, thereby indicating that humans are also susceptible to the effects of caloric restriction (and presumably to the effects of any future caloric restriction mimetic drugs that may appear). It is reasonable to conclude that the animal studies are more or less indicative of the longevity-determinant mechanisms operative in the human body. A relatively short-term National Institutes of Health study in which human volunteers eat a nourishing, palatable but calorically restricted diet under controlled conditions is now under way (Heilbronn and Ravussin 2003). Periodic assays will be made to determine if the physiological process known to be affected in the rhesus monkeys by caloric restriction are also altered in the same manner in humans. An entirely different line of inquiry has to do with the effects of nutritional deficiencies on health and longevity. Such studies may be regarded as the obverse of the calorie-restriction studies, and it might be expected that poor nutrition or overnutrition would lead to a decreased health span and overall longevity. Poor eating habits, poor appetite, difficulty in chewing or

ity from all causes was reduced by one-half compared with the general population (ChangClaude et al. 1992). The lowest mortality was obtained for cardiovascular diseases, but significant declines were also noted in deaths from cancer and from respiratory and digestive diseases. These data are comparable to those noted elsewhere, for example among Seventh Day Adventists, Mormons, and other groups whose traditions promote the practice of good nutrition and health habits. It has been suggested that the health benefits of the dietary practices of each of the several groups discussed above are sufficiently well documented as to be an important public health recommendation, even if it adversely affects the beef industry (McMichael 1992). Sixth are the data obtained from people who have decided to put themselves on a calorierestricted diet. The Caloric Restriction Society is a private organization committed to educating and supporting people interested in living longer via caloric restriction. An evaluation of the effects of long-term caloric restriction in 18 human volunteers has been done, and a summary of the results is presented in table 6.4 (Fontana et al. 2004). The data show that, compared to matched controls on a normal diet, caloric restriction significantly alters the body build, various indices of cardiovascular health, key indices of the glucoregulatory system, and one measure

Table 6.4 Effects of Caloric Restriction (CR) on Humans CR group (n = 18)

Control group (n = 18)

Significant?

Age (years)

50.3

50.3

No

BMI (men) Total body fat (%) Lean body mass (%)

19.6 6.7 93.3

25.9 22.4 76.8

Yes Yes Yes

Total cholesterol (mg/dl) LDL (mg/dl) HDL (mg/dl) Triglycerides (mg/dl) Fasting glucose (mg/dl) Fasting insulin (mlU/ml) Cross reactive protein (mg/ml) Source: from Fontana et al. (2004).

221

158 86 63 48 81 1.4 0.3

205 127 48 147 95 5.1 1.6

Yes Yes Yes Yes Yes Yes Yes

222 Chapter 6 Altering Aging swallowing food, loneliness, and hospitalization are all risk factors for malnutrition. Proteincalorie malnutrition has been called the most undiagnosed nutritional disorder in the world today (Gambert and Kassur 1994); it develops in cases where a prolonged diet is very low in protein and high in carbohydrates. This condition is often found in the elderly, for reasons such as those just described. Supplemental nutrition may be justified by physiological changes, which may account in part for a decreased efficiency of nutrient absorption or waste product excretion, as well as by nonphysiological factors such as a change in income or increased consumption of sweets and decreased intake of protein. A great deal of evidence suggests that aging is statistically associated with the development of glucose intolerance, by which is meant the inefficient uptake of glucose from the blood by the peripheral tissues (see reviews by Ruhe and McDonald 1994; Halter 1995). Glucose homeostasis is maintained largely by the insulinmediated uptake of glucose by the peripheral tissues. When this process becomes inefficient, the result is a high blood sugar level. In severe cases, the afflicted person may be said to have developed non–insulin-dependent diabetes mellitus (NIDD), a pathological syndrome that may have severe and adverse effects on the aging process. Obesity and lack of physical activity may be the most important factors in the development of glucose intolerance in the aged, as nonobese individuals who exercise show no decrease in insulin sensitivity (Kahn et al. 1990). Overnutrition is a major factor in modern society and may be one reason the incidence and prevalence of glucose intolerance is so widespread. Obesity is a known risk factor for shortened life. It has been estimated that almost 45% of Americans between the ages of 65 and 74 have severe or impaired glucose tolerance (Halter 1995). In addition, the average annual incidence of NIDD triples between the ages of 50 and 70, compared to only a 25% increase between the ages of 30 and 50 (Ruhe and McDonald 1994). NIDD is an agerelated disease that does and will continue to affect a large number of individuals. In addition, there is a variable association between glucose in-

tolerance and hypertension in some, but not all, groups of people (Halter 1995). Obesity may be one confounding variable, and ethnicity may be another; the association can be detected in some Caucasian populations and is absent in other racial or ethnic groups. The molecular biology of obesity, although far from clear, is gradually being characterized (Flier 2004). The central nervous system controls our appetite and the manner in which we use and partition our energy. These efferent outputs, which obviously play an important role in the genesis of obesity, are controlled by long-term afferent signals from the abdominal fat cells (e.g., leptin), pancreas (e.g., insulin), and by meal-related afferent signals from the stomach (e.g., ghrelin) and small intestine (e.g., peptide YY, glucagon-like peptide-1 cholecystokinin). There are complex relationships between the various afferent inputs and the efferent outputs, so it is difficult to disentangle the mechanisms. Altering the dietary carbohydrate intake can modulate the development of glucose intolerance. Healthy elderly people demonstate an improved glucose tolerance, an enhanced tissue sensitivity to insulin, and improved pancreatic beta-cell function when placed on a restricted diet in which their carbohydrate intake was limited to 85% of the prior ad libitum diet (Chen et al. 1987). The effects of aging on glucose and insulin action were reduced but still persisted as a result of this treatment. The fact that improvement was observed after a mere 3–5 days of carbohydrate restriction suggests that a longer term dietary restriction might prove at least as beneficial. Note that this is fully consistent with the results of the several studies showing that aging is a cell-level phenomenon and that the cells can switch from an optimal to a senescent pattern of gene expression within a few days after exposure to the new signals. It may be instructive to recall the effects of dietary restriction on the rodent’s ability to maintain efficient regulation of its pyruvate kinase enzyme, as discussed earlier, and its subsequent effects on the animal’s ability to age in a healthy manner. It is never too late for flies or people to adopt a healthy diet. Along the same lines, other studies (cited in Schlenker 1984) indicate that animals fed diets

6.3 Useful Methods of Modulating Aging Processes in Humans

consisting of highly processed foods also have a higher incidence of degenerative diseases. Comparable longitudinal studies of people who are fed mostly fast foods are not yet complete, but the high-calorie, high fat, high protein, and high salt content of the typical fast-food meal does not bode well for the continued metabolic efficiency of the people who eat them. If you routinely partake of such fare, you are a member of the experimental group. Good luck. If the animal studies are any guide, you will need it. The current obesity epidemic in the United States may be regarded as the outcome of this unintended human experiment on nutrition and longevity. Adjusting your diet as you age is clearly important. Many people go on a diet to (mostly unsuccessfully) remove the excess weight they put on as they grow older. Most popular diets offer between 1700 and 2100 calories per day, and that is why one can lose weight if one stringently follows the particular diet’s guidelines. In addition to losing weight, however, the most effective intervention would be to manipulate your diet as a young or middle-aged adult to delay or minimize the deleterious effects of aging. In other words, if we don’t practice some mild form of caloric restriction, we should at least ingest a lowfat, nutrient-dense diet. Guidelines for such a diet, released by the National Research Council in 1989, are summarized in table 6.5. These recommendations, if followed, are expected to reduce the incidence of cardiovascular disease,

223

cancer, and obesity. More people are becoming increasingly aware of the linkage between diet and health, and the U.S. food industry has responded to this heightened concern with an array of new nonfat products. However, many of these products have substituted the fat with sugar, diglycerides (which do not count as fat on food labels), artificial sweeteners, and sucrose polyesters (Willett 1994). It is not clear that these substitutions are in fact healthful, and it is entirely possible that some people may be misled by the labels. It is clear, however, that optimum health can be achieved from a diet that emphasizes generous helpings of vegetables and fruit. It is likely that the dietary guidelines of table 6.5 were revised by the government in 2005. The new edition of the guidelines are intended for use by both the public as well as policy makers, nutritionists, and health educators. The report and its new recommendations may be viewed online (http://www.healthierus.gov/dietaryguidelines/.). The guidelines are based on a report by the Dietary Guidelines Advisory Committee, a panel of nutrition experts who reviewed the relevant literature, and this report is also available online at http://www.health.gov/dietaryguidelines/ dga2005/report/.). The new guidelines do provide more detailed caloric intake guides as modulated by gender, age, weight, height, and activity level. In addition, the new guidelines place more emphasis on physical activity as a means of maintaining caloric balance. Finally, the guidelines

Table 6.5 Summary of National Research Council Recommendations for a Healthy Diet 1. Reduce total fat intake to 30% or less of calories. Reduce saturated fatty acid intake to less than 10% of calories and cholesterol intake to less than 300 mg daily. 2. Every day eat five or more servings of a combination of fruits and vegetables, especially green and yellow vegetables and citrus fruits. Increase starches and other complex carbohydrates by eating six or more daily servings of a combination of breads, cereals, and legumes. 3. Maintain protein intake at moderate levels. 4. Balance food intake and physical activitiy to maintain appropriate body weight. 5. Limit alcohol consumption to the equivalent of 1 ounce of pure alcohol per day. 6. Limit total daily salt intake to 6 g or less. 7. Maintain adequate calcium intake. 8. Avoid taking dietary supplements in excess of the recommended daily allowance in any one day. 9. Maintain an optimal intake of fluoride, particulary during the years of primary and secondary tooth formation and growth. Source: after Willett (1994).

224 Chapter 6 Altering Aging recommend eating more complex, low glycemicindex carbohydrates as well as more mono- and polyunsaturated fats. The glycemic index is a measure of the speed with which different foods increase the blood glucose level, and therefore insulin secretion. Pure glucose, or foods made from highly refined starches which are readily digestible (e.g., white bread), are examples of high glycemic-index foods, whereas high-fiber carbohydrates (e.g., whole oats, wheat bran) are examples of low-glycemic foods. The basic idea is that rapid changes in blood glucose will result in correspondingly rapid changes in insulin levels, and these rapid falls in insulin levels will stimulate the appetite even though one has recently eaten a high-glycemic index food. Eating in response to this mistaken physiological signal will inevitably increase the caloric intake and, eventually, one’s weight and body mass index (BMI). Thus, eating low glycemic-index foods will maintain the blood glucose and insulin levels at a more even level, will not stimulate appetite, and will allow one to adhere to a nutritionally complete caloric intake that allows weight loss and maintains the lower BMI without supernatural heroics. Calories are important, but the types of food we eat and the physical activities we engage in are equally important. I can testify from personal experience that the high-protein/low-glycemic foods included in the South Beach diet do effectively control appetite and control weight, at least in the short term (Foster et al. 2003; Samaha et al. 2003). I am no longer counting calories, but I will probably stay on this healthy and satisfying food regime for the rest of my life. Others have had similar experiences (Lewis 2004). Roy Walford, an internationally respected immunologist and gerontologist, described in practical terms a high-nutrient, low-calorie diet in his book How to Double Your Vital Years: The 120 Year Diet (1986). The diet contains a variety of foods, is relatively practical to implement in today’s society, and is based on the principles elucidated by past decades of research on animals and humans. Readers interested in pursuing this effective intervention for themselves or others may wish to read this or another similar text. The National Institutes of Health is now undertaking

a small-scale 1 year study of this diet on ordinary people. The participants will be assayed periodically for a variety of standard physiological traits, including many of those known to be altered in calorie-restricted animals. Should the study show that people react physiologically similar to calorie-restricted animals, this would confirm the limited data suggesting the benefits of such dietary intervention (Walford et al. 2002). 6.3.2.2 Vitamins and Minerals

Studies of U.S. and British populations (cited in Schlenker 1984) show that diets deficient in various vitamins and minerals are associated with a high rate of mortality that often involves degenerative diseases. The need for micronutrients is currently under discussion and represents an evolution in our thinking on the topic. For example, the minimum recommended daily allowance (RDA) for vitamin C (ascorbic acid) was originally based on the knowledge that the prevention of scurvy requires at least 10 milligrams per day. It is now known that our bodies require higher doses for the maintenance of good health and for the prevention of various pathologies. Linus Pauling strenuously championed the preventive health benefits of vitamin C megadoses of as much as 10 grams (10,000 milligrams) per day. Studies revealed no ill side effects in healthy individuals taking up to 5 grams of ascorbic acid per day (Cohen et al. 1994). High levels of plasma ascorbic acid have been associated in some but not all studies with a lower incidence of hypercholesterolemia, cancer, and certain other agerelated pathologies (Cohen et al. 1994). Perhaps most germane to the normal aging process are the antioxidant effects of vitamin C on glycosylation and oxidative damage; I discuss these topics in more detail in chapter 10. Vitamin E is an essential fat-soluble compound with antioxidant activity that is found in biological membranes and is considered to protect against lipid peroxidation. Persons having conditions that interfere with the normal processes of fat digestion, absorption, or transport often have low serum vitamin E levels (< 0.5 milligrams per deciliter). Severe and chronic vitamin

6.3 Useful Methods of Modulating Aging Processes in Humans

E deficiency can result in various neurological symptoms. The minimum daily intake of vitamin E necessary to avoid these deficiency syndromes was set by the Food and Drug Administration (FDA) at 15 international units (IU). But perhaps more important to our purposes are the results of several epidemiological studies that generally indicate an inverse relationship between the incidence of various cancers and the level of serum vitamin E (and/or of other antioxidants as well). An increasing amount of evidence, discussed in chapter 10, suggests that these beneficial effects may be modulated via the known antioxidant effects of vitamin E and its ability to assist in countering free-radical promulgation and damage. For example, healthy adults taking a daily supplementation of 1000 IU of vitamin E for 10 days had a significantly decreased level of pentanes in their breath, which is a reliable index of lipid peroxidation taking place in situ. A review of 19 studies done in different countries on different populations suggests that significant increases in the risk of heart disease are associated with lower blood levels of vitamin E and/or other antioxidants. Although some foods are rich in vitamin E (e.g., vegetable oils, whole grains, wheat germ), it is not usually feasible to obtain from the diet alone the high values of vitamin E needed for disease prevention, which are commonly in the range of 300 to 1200 IU per day. Oral vitamin E is well tolerated by the body and is relatively nontoxic in healthy people. Animal studies suggest that huge doses over a long period of time, equivalent to the consumption of 35,000 milligrams per day for 2 years by a man weighing 70 kilograms, are necessary before chronic toxicity can be detected (VERIS 1991). Human studies show that high daily doses (100–800 IU) of vitamin E taken for a 3-year period yielded no observed side effects (Farrell and Bieri 1975). In fact, the upper recommended limit of vitamin E was recently increased to 1200 IU/day for adults. Epidemiological studies show that high blood concentrations of vitamin E are associated with a decreased risk of cardiovascular diseases and certain cancers. A recent metanalysis concluded that daily doses >400 units/day might lead to a small but significant increased mortality in hu-

225

mans (E. R. Miller et al. 2004). The discrepancy between this study and the others cited here is not clear, but it may have to do with the fact that 16 of the 19 populations examined in that study were composed of sick or unscreened individuals. These initial and apparently contradictory findings were confirmed by a large scale analysis of the effects of vitamin E on healthy people as well as those with diagnosed cardiovascular and/ or diabetic symptoms (Lonn et al. 2005). In these two studies, the mortality risk relative to healthy age-matched controls is significantly increased at doses ≥ 400 I.U./day. The mechanisms for this effect are not clear but they were speculated to involve deleterious effects of vitamin E on the oxidation of low density lipoproteins (LDL). The net result of these investigations is that higher doses of vitamin E may be neutral or beneficial in healthy people but harmful in individuals with cardiovascular and/or diabetes. It is not clear whether the diseases caused the metabolic differences between these two groups of people, or whether the metabolic differences preceeded and somehow brought about the diseases. In any event, there is a major disagreement between a host of epidemiological studies and these several clinical trials. A cautious skepticism is a necessary trait when considering the use of unregulated supplements. It may be possible to construct mixes of vitamin E and certain dietary phenolic compounds which would permit the beneficial effects of vitamin E at doses below that level when deleterious effects take place (Frank 2005). Other vitamins besides ascorbic acid and vitamin E are needed, including thiamine (vitamin B1), folic acid, and vitamins B12, D, A, and K. All of these micronutrients are available in a wellplanned natural diet. Some healthy people might be persuaded to take larger-than-recommended doses of one or two of them for preventive means, and an increasing number of studies indicate that such supplementation is of value. Deciding to supplement your diet with vitamin(s) means that you are banking on a mechanism of aging that hasn’t been proven yet, so you could be wrong. But, of course, by the time this approach is proven efficacious, it might be too late for you to do anything about it. So you must accept responsiblity

226 Chapter 6 Altering Aging

1.5 1.0 0.5

Chi for trend = 9.87 p < 0.001

Body mass index

32

0.0

2.5

3.2

3.4

2.0 1.5 1.0 0.5

Chi for trend = 4.67 p < 0.001

0.0

32

2.0

<1 9. 0 0– 21 .9 22 .0 –2 25 4. .0 9 – 27 26. .0 9 –2 8. 9 29 .0 –3 1. 9

(b)

2.5

<1 19 9.0 .0 –2 1. 9 22 .0 –2 25 4. .0 9 – 27 26. .0 9 –2 8. 9 29 .0 –3 1. 9

(a)

Multivariate relative risk

We live in a society that glorifies thinness. This attitude is certainly reinforced by our very real concern with avoiding pathologies such as cardiovascular diseases, diabetes, and the like—conditions for which excess weight and obesity are considered to be risk factors. It was once the case that, in order to determine if your body weight

19 .

6.3.2.3 Weight Control and Caloric Restriction in Humans

was within normal guidelines, you had to consult a chart which showed body weight as a function of gender and height. It was all too easy for someone to dismiss being overweight with the remark that one was “big-boned” or some equivalent excuse. No more. Those cumbersome charts have been replaced by a single number that takes all of the above factors into consideration: the body mass index, or BMI. The basic formula is weight (in kilograms) divided by height (in meters2). For those not using the metric system, you can calculate the metric equivalents by dividing your weight in pounds by 2.2. The metric equivalent for your height can be calculated as your height in inches multiplied by 2.54, the result divided by 100, and that result then multiplied by itself. The calculated BMI should lie somewhere between 18 and 34. Values approximating 23–25 are considered optimal; values over 28 suggest that one is in the heaviest 15% of the U.S. population. If you don’t smoke, you can obtain your relative risk of death by inspecting figure 6.7 or 6.8. Several studies that examined the relation between weight and mortality produced diverse findings, ranging from no association to a J-shaped or U-shaped relation to a direct association to an inverse relationship. One retrospective study used data obtained from the life insurance industry (Andres 1984). When mortality was plotted as a function of weight, the surprising result was a

Multivariate relative risk

for your own treatment. Remember, though, that there is no evidence that taking megadoses of all of these micronutrients is of value, and some recent studies have shown that supplementation of certain of them (vitamin A, Vitamin E, betacarotene) can be detrimental in certain populations. For example, the current RDA for vitamin A is 1700 IU. Consumption by women in early pregnancy of more than 10,000 IU per day was found to be associated with a high level of birth defects (Lammer et al., 1985). There may be a fine line between enough and too much. A simplistic approach to nutrition—popping a pill while continuing to eat a cheeseburger, smoke a cigarette, and sit on your duff—will simply not work. Even the ultimate bible of consumers, Consumer Reports, agrees with this pronouncement in its report on practical methods of living longer (see the January 1992 issue).

Body mass index

Figure 6.7 The relative risk of death from all causes, 1976 through 1992, according to body mass index for (a) all women (4726 deaths) and (b) women who never smoked and had stable weights (531 deaths). All relative risks have been adjusted for age in 5-year categories. The reference category is that of women with a BMI less than 19.0. Note that the curve changes from J-shaped to linear as the classification becomes more stringent. (After Manson et al. 1995.)

6.3 Useful Methods of Modulating Aging Processes in Humans

85 yrs

2 1

75–84 yrs

2

Relative risk of death from all causes

1 65–74 yrs

2 1

55–64 yrs

2 1

45–54 yrs

2 1

30–44 yrs

3

2

29

.0

–3

32

9 1.

9 8.

9 27

.0

–2

6.

9 25

.0

–2

–2 4. .0 22

19

.0

–2

<1

1.

9.

9

0

1

Body mass index

Figure 6.8 The relative risk of death from any cause according to age group and body mass index (BMI) category among healthy white men and women who have never smoked. All relative risks were adjusted for age, education, physical activity level, and alcohol consumption. The reference category was made up of subjects with BMIs of 19.0–21.9. The bars represent 95% confidence limits. Relative risk estimates are not shown for age-BMI groups with fewer than five deaths. (After Stevens et al. 1998.)

U-shaped curve instead of the continuous linear relationship implied by the popular expectation that thinner people always live longer than plumper people. A later prospective study investigated body weight and mortality among 115,195 U.S. women enrolled in the prospective Nurses’ Health Study (Manson et al. 1995). The relationship between weight and mortality changed as the sub-

227

jects were sorted according to various health parameters. A J-shaped relationship was observed for all women before sorting (figure 6.7a), but this relationship became direct when smoking and/or a history of unstable body weight were taken into consideration (figure 6.7b). It seems that the presence of smokers in the study population accounted for the initially J-shaped relationship: many people smoke to keep their weight down, and so the low BMI group contains a number of people who will suffer the undesirable side effects of smoking. Removing them from the study reveals the direct relationship between BMI and mortality risk. A third study, involving above 325,000 nonsmoking people 30 years of age or older and enrolled in the American Cancer Society’s Cancer Prevention Study, also detected a significant association between greater BMI and mortality (Stevens et al. 1998). The relative risk associated with increased BMI decreased with age (see table 3.3); although it is also interesting to note that there are apparently no obese men older than 85 (figure 6.8). All three studies, when considered against the background of the animal studies involving caloric restriction, can be interpreted as showing that there is, quite literally, no free lunch. The anecdotal data obtained from interviews with centenarians suggests that such individuals generally follow rational and moderate dietary practices. The most obese individuals have a mortality risk 30–100% higher than that of thinner individuals. Approximately one-third of the U.S. adult population is overweight (defined as more than 20% above the desirable weight), and the studies suggest that these people are at a significant risk of premature mortality. If so, then the long-term trend of increase in the U.S. life expectancy may slow or even decline. The probable negative effect of this trend on longevity has been mathematically modeled (Fontaine et al. 2003). The prediction is that maximum longevity might be decreased by as much as 22% for young people who are severely obese (BMI >45), with proportionally smaller decreases for older individuals or for young people with a lower BMI. This might translate into as much as a 13-year decrease in one’s life span. The current societal disapproval

228 Chapter 6 Altering Aging of overeating and obesity stems from the belated awareness on the part of health, business, and political leaders of the huge economic and social costs that must be borne if our society continues to encourage such unhealthy practices. Even McDonald’s has downsized their super size. According to an old folk saying, after the age of 40, every person is either a fool or a physician. That saying might be a useful guide to the behaviors we should adopt. Many people diet. A few people have actually adopted and practice a calorie restriction diet/ nutrition regime. The Calorie Restriction Society (www.calorierestriction.org), for example, has at least 130 members who claim to adhere to a 25– 40% calorie reduction from their prior dietary regime (about 1800 calories for men, about 1500 for women). Most are middle aged, married with children, non-vegans, and do not fast. About half have been on caloric restriction for more than 1 year, and their self-reported BMIs decreased from 25 to 21 for men and from 25 to 22 for women. Their self-reported physiological values are consistent with low body fat and good health (K. Hashmi, personal communication and http: //www.cron-web.org; also see table 6.4) and are consistent with the idea that the caloric restriction response is present in humans. I discuss this possibility and its scientific, social, and ethical underpinnings in chapter 15. 6.3.2.4 Exercise in Humans

The effectiveness of exercise as an anti-aging intervention has been the focus of many studies.

Few of these investigations have been longitudinal studies, so the interpretations are sometimes open to question. However, the story told by the various studies is reasonably consistent. The data suggest that the maximum oxygen uptake (VO2 max) declines at a rate of about 1% per year. This measure of respiratory function is a good indicator of aerobic capacity and often has predictive capability (see chapter 3). The reasons underlying this age-related reduction are not known, although the decreased cardiac output and the changes in the connective tissue of the lungs and respiratory muscles (see chapter 5) must play a part. This decline is not inexorable, however; the VO2 max value increases as a result of habitual strenuous exercise. In addition, healthy older men who were given a year of endurance training exhibited statistically significant increases in their VO2 max values (table 6.6), a result that confirmed earlier reports in the literature regarding the beneficial effects of regular, moderate exercise (Buskirk 1985). The exercise must be regular or deconditioning will set in, and all the hard-won gains will disappear within a few months, as table 6.7 shows. It has been observed that adaptation to exercise is an age-dependent phenomenon, for a given effort will produce greater effects at younger ages than at older ages. The reasons underlying this change probably involve many factors, two of which must be the intrinsic structural and metabolic changes taking place in the cardiovascular system, as well as in the skeletal muscle fibers and the nerves that innervate them (see chapters 5 and 11). Successful exercise regimes for older individuals take these changes into ac-

Table 6.6 Effects of Habitual Physical Activity on Selected Physical Variables Variable Age, years Weight, kg Triceps skin fold Distance run/year, km VO2 max, ml/kg/min HDL/total cholesterol

Marathon runners

Joggers

Sedentary

48.2 72.4 8.4 2644 57.2 0.36

48.7 76.2 9.5 1488 49.2 0.31

48.3 86.0 12.5 0 42.5 0.18

Source: Data from Hartung et al. (1981).

6.3 Useful Methods of Modulating Aging Processes in Humans

Table 6.7 Percentage Changes in Selected Variables As a Result of Short-term Physical Conditioning and Deconditioning Variable VO2 max VE2 max Maximum heart rate

With conditioning

With deconditioning

11.7% 15.1% 15.1%

– 7.2% – 3.5% –13.9%

Source: data from Miyashita et al. (1978). Note: VE2

max

= maximum volume of expired air.

count by altering the type and the pace of the prescribed physical activities (Buskirk 1985). Loss of skeletal muscle mass with age is well documented, and the process of sarcopenia is described in chapter 11. This loss decreases the strength available for the elderly to do various tasks of daily living. Because the lean body mass represents the most metabolically active compartment of the body, its loss must certainly contribute to the decreases in the VO2 max and in the resting metabolic rate as well as to the increase in insulin resistance. Animal studies show that exercise enhances insulin sensitivity by increasing the levels of specific proteins involved in the insulin-signaling pathway, restoring the equimolar concentration of these proteins, and thus restoring the pathway to its optimal functional state in both young and old rats (Arias et al. 2001). Elderly individuals who engaged in resistance training increased their muscle mass, increased their resting metabolic rate, decreased their fat body mass, and increased their caloric intake (D. Evans 1995). I pointed out in chapter 5 that the bones respond to the stresses placed on them by muscle use. The expectation that exercise benefits the skeleton is borne out by the data. Strength training has a beneficial effect on bone density: A group of elderly women who exercised for a year increased their bone mineral density, while the nonexercising control group showed a significant decline (D. Evans 1995). I have also pointed out the association of glucose intolerance and its consequent effects with age. Exercise has been shown to improve glucose tolerance in previously sedentary subjects and to prevent the

229

onset of NIDD (Holloszy and Kohrt 1995). It has been shown that the improvements in carbohydrate metabolism occur primarily in skeletal muscle, brought about in part by increased muscle mass and in part by physiological changes that improve the muscles’ uptake of glucose in response to insulin (D. Evans 1995). Finally, exercise induces significant alterations in plasma hormone concentrations in all age groups, without regard to the subject’s accustomed activity levels beforehand (table 6.8; Silverman and Mazzeo 1996). Certain of these hormones (e.g., growth hormone) have significant metabolic effects, as we’ll see in the next section. The observation that the old, trained men demonstrated a greater hormonal response than did the young, sedentary men suggests that the neuroendocrine mechanisms underlying the hormonal responses are enhanced with endurance training. The data further suggest that the normal age-related diminution in these neuroendocrine mechanisms can be delayed or attenuated by exercise. Weindruch (1995) has pointed out that skeletal muscle accounts for a large share of the body’s oxygen consumption at rest and at work. This fact, coupled with the low repair capacities of muscle, suggest that some of the age-related changes in muscle reflect oxidative damage to the muscle cell (Luhtala et al. 1994). This correlation raises the possibility that caloric restriction coupled with moderate exercise is an effective intervention. Such an effect has been demonstrated in rats: A 10% caloric restriction combined with a mild daily exercise regime led to a significant extension of maximum longevity, which was, however, less than that seen with a standard 40% caloric restriction without exercise (McCarter et al. 1996). In these animals, the age-retarding effects of caloric restriction were mobilized by the combination of two low-level interventions. This combined intervention regime might serve as a model for people. In fact, more people are successful in losing weight when they combine a moderate diet with a moderate exercise regime than those who use only one or the other of these interventions. The effects of exercise in humans has been well studied (DiPietro, 2001; Poehlman et al., 2001; Rejeski and Mihalko, 2001; Westerterp

230 Chapter 6 Altering Aging Table 6.8 The Effect of a Maximal Exercise Test on Plasma Hormone Concentrations Hormone Lactate (mM)a Norepinephrine (ng/ml) Epinephrine (ng/ml) Growth hormone Cortisol (mg.dl)

Young sedentary

Young trained

Middle-aged sedentary

Middle-aged trained

Old sedentary

Old trained

4.8 + 0.3 2.0 + 0.4

6.0 + 0.4 7.5 + 1.3*

3.9 + 0.5 3.1+ 0.3

5.2 + 0.4 8.6 + 2.5*

3.7 + 0.4 4.1 + 1.0

3.8 + 0.6 6.0 +1.1*

0.4 + 0.04

1.2 + 0.3*

0.4 + 0.12

1.3 + 0.5*

0.3 + 0.05

0.7 +0.3*

5.6 + 1.8 19.8 + 2.3

21.9 + 4.0* 26.8 + 2.2*

8.6 + 3.2 14.9 + 0.9

19.0 + 4.4* 21.6 + 1.8*

2.8 + 1.0 16.3 + 2.4

11.3 + 3.5* 19.5 + 2.4

Source: data from Silverman and Mazzeo (1996, table 4). a

Lactate levels indicate metabolic effort

*Significant training effect.

and Meijer, 2001). Physical exercise in humans reduces both morbidity and mortality. A prospective study of 121,700 women in the Nurses’ Health Study were examined for their exercise habits and then assayed over the next 8 years for their incidence of coronary events (Manson et al. 1995). Their exercise habits were converted into weekly metabolic-equivalent (MET) scores for total physical activity, and the statistical correlation of their morbidity (coronary events) with their MET scores was carried out. The data, shown in table 6.9, show that even moderate walking can noticeably decrease one’s morbidity. If morbidity is decreased, then it is reasonable to expect that mortality will also decline (up to a point, at any rate). The data of table 6.10 show that the 5-year follow-up mortality of older men and women was also significantly correlated with their exercise habits. The Harvard Alumni Study used self-reported data in a longitudinal study and found that vigorous exercise (>2000 kcal/week) was associated with an estimated 2 more years of life than that observed in the more sedentary (<500 kcal/week) subjects. The beneficial effects of exercise seem to be tied to an increased physical fitness, objectively measured as the VO2 max (see figure 3.16) as well as improvements in glucoregulatory function. Body composition is differentially affected by the type of exercise; resistance exercise reduces fat mass and increases fat-free (i.e., skeletal muscle) mass, and aerobic exercise reduces fat mass only.

Molecular characterization of the expression of about 4727 genes in the skeletal muscle of endurance athletes and sedentary men showed that the trained men had significantly higher expression of genes involved in molecular chaperones, muscle repair, and the oxidative pathway, but a lower expression of genes involved in the glyconeogenesis and glycolysis pathways (Yoshioka et al. 2003). These findings are consistent with earlier data showing that endurance exercise training increases muscle mitochondrial density and oxidative enzyme activity. In addition, endurance athletes had a higher expression of a particular myosin light-chain isoform (MLC2), which is the major isoform in slow-twitch skeletal muscle fibers. Thus, the overexpression of this particular isoform may contribute to the greater power-generating capabilities of the endurance-trained subjects. Resistance exercise has been shown to directly regulate the circadian clock genes of skeletal muscle independent of the suprachiasmatic nucleus of the brain, and thus to appropriately regulate a number of genes involved in exercise-induced muscle hypertrophy (Zambon et al. 2003). But fat and muscle are not the only tissues affected. Aerobic exercise is reported to reduce the age-related loss of both gray and white matter (Colcombe et al. 2003). The mechanisms underlying this observation are not known with surety, yet it would be useful if such a low-cost lifestyle change could decrease the morbidities and unhappiness associated with brain dysfunction.

6.3 Useful Methods of Modulating Aging Processes in Humans

231

Table 6.9 Relative Risk of Morbidity (Coronary Events) According to Quintile Groups for Physical Activity Quintile group for total physical activity

MET-hour/week (median)a No. of coronary events Relative risk (multivariate)

1

2

3

4

5

0.8 178 1.0

3.2 153 0.88

7.7 124 0.81

15.4 101 0.74

35.4 89 0.66

Source: adapted from Manson et al. (1999, table 2). aMET = metabolic equivalent. As a reference, walking done at an easy or casual walking pace (<2 mph), an average pace (2.0–2.9 mph), a brisk pace (3.0–3.9 mph), or a very brisk pace (>4.0 mph) has a score of 2.5–4.5 MET/hour, depending on the pace.

Table 6.10 Effect of Weekly Physical Exercise on 5-year Mortality in Older Men and Women Exercise (kcal/week)

Relative risk of mortality

<67.5

67.5–472.5

472.6–980.0

980.1–1890.0

>1890.0

1.0

0.78

0.81

0.72

0.56

Source: adapted from data of Fried et al. (1998), as presented in DiPietro (2001).

Physical frailty arising from severely impaired strength, mobility, balance, and endurance significantly diminishes the ability of many older persons to conduct the activities of daily living. It also significantly increases the risk of falls and their associated trauma (e.g., hip fracture), as well as the risk of mortality. Between 10 and 62% of the elderly, depending on their sex and age, need help with the activities of daily living and cannot live independently (Guralnick and Simonsick 1993). Physical frailty is a negative predictor of active life expectancy, with all the personal and public costs thus implied. Although physical frailty was once thought to be an inevitable part of aging, it now appears that it can be reversed by exercises designed for endurance training and for muscle strengthening. The success of exercise intervention in the laboratory appears to depend on the intensity of the exercise stimulus. The high-repetition, lowintensity stimulus of aerobic training leads to only minimal strength gain when compared to the lowrepetition, high-intensity stimulus of resistance training (Fiatarone et al. 1993). Nonetheless, any exercise is better than none. A community study showed that elderly people who walked 30 min-

utes or more every day performed significantly better in both the laboratory and the community than did those who did not walk (Svanborg 1993). In fact, regular walking in physically capable older men is not only associated with a lower overall mortality rate than nonwalkers, but the distance walked is inversely related to mortality (Hakim et al. 1998). The musculoskeletal system seems to retain its plasticity at all ages. The fact that our reserve capacity goes down with age means that the need for adopting preventive or postponing measures becomes more obvious. A mouse study showed that life-long exercise prevented many of the age-related alterations in gene expression in the hearts of middle-aged and old animals (Bronikowski et al. 2003). Congestive heart failure is a common age-related condition of humans, and it is characterized by a particular pattern of gene expression changes. These genes have not been characterized fully, but it will be interesting to find out if they include any of the genes whose expression is modulated by exercise in the mouse. There seems to be no clear relationship between athletic competition done as a college youth and subsequent longevity (Schneider

232 Chapter 6 Altering Aging 1985), although many former athletes can do aerobic walking later in life with less effort and strain when compared to age-matched nonathletes (Buskirk 1985). However, when men habitually active in endurance-type recreational sports (such as cross-country skiing or jogging) were compared to habitually sedentary men, the active men scored better on several (but not all) physiological parameters than did the inactive men (see table 6.7). As with any other intervention, carrying exercise to an extreme may be detrimental. When 20 men engaged in vigorous exercise (10 hours per day for 30 days), the rate of oxidative DNA damage products detected in their urine increased by 33% (Poulsen et al. 1996). Human cells exposed to extreme oxidative stress in vitro are not capable of repairing the DNA damage that occurs in their mitochondria, although they are capable of repairing nuclear damage (Yakes and Van Houten 1997). Thus, extreme exercise may preferentially expose one’s mtDNA to increased oxidative damage, thereby potentially damaging health instead of improving it. The conclusion of Buskirk (1985, p. 924) seems appropriate: “Thus, functional alterations wrought by regular exercise blunt the downward trends commonly associated with aging.” Properly used, exercise can be an effective intervention. 6.3.2.5 Hormonal Interventions in Humans

Growth Hormone. The ability of hormones to stop or reverse the aging process is well established in the popular mind as a result of the indiscriminate comingling of modern hormone replacement therapy for postmenopausal women with the older attempts of Brown-Sequard (1889) to rejuvenate older men with testicular extracts from dogs and guinea pigs. But what of the facts? Rudman et al. (1990) treated 21 healthy men over 60 years of age, each of whom was deficient in growth hormone, with doses of recombinant human growth hormone designed to restore their IGF-1 levels to those of young adults. The treatments lasted 6 months and showed no adverse side effects. The men showed a threefold rise in

plasma IGF-1 levels, a 9% increase in lean body mass, a 14% decrease in fat tissue mass, a 1.6% increase in lumbar vertebral bone density, and a 7% increase in skin thickness. These effects disappeared within 3 months after the treatments were stopped. A larger follow-up study was carried out for 12 months (Cohn et al. 1993) and found that significant numbers of the participants developed potentially serious side effects ranging from carpal tunnel syndrome to hyperglycemia. Other investigators discovered the same phenomenon (Schwartz 1995). Many of the subjects dropped out of the studies. This finding lent importance to interventions based on indirect stimulation of the growth hormone levels. Both Iovino et al. (1989) and Corpas et al. (1992) found that administration of growth hormone-releasing hormone increased levels of growth hormone (GH) and IGF-1 in older individuals, giving hope that alternative routes would provide an effective intervention strategy without side effects. This concern cannot be minimized, given the seriousness of the side effects so far observed and the pleiotropic effects of the growth hormone on many tissues. A comprehensive review by Wolfe (1998) of the role of GH in slowing aging processes emphasizes the variable outcomes of GH therapy and the high possibility of deleterious side effects arising from its unregulated use. People who suffer from hypopituitarism can benefit from GH replacement therapy, but GH supplementation of normal elderly people yielded either no functional change or adverse side effects after extended use. Elderly men who exercised vigorously showed no noticeable effect of GH on body mass or muscle size, presumably because exercise induced more GH than the supplementation provided (table 6.8). GH is now known to be regulated by an entirely new signaling pathway involving hexapeptides when administered orally or nasally (see Wolfe 1998 for review and references). Such GH-releasing peptides have been demonstrated to induce the normal pulsatile nature of GH secretion. The effect of increased GH, however induced, on the aging process still needs to be rigorously confirmed or disproved. Wolfe suggests that the popular enthusiasm for GH and

6.3 Useful Methods of Modulating Aging Processes in Humans

other anti-aging compounds will lead to unregulated self-administration and all sorts of potential damage. In addition, a chronic excess of GH in young adults is associated with eventual diabetes, osteoarthritis, hypertension, and possibly cancer—the very problems that characterize normal aging, which the use of exogenous GH is intended to avoid (Rudman and Shetty 1994; Wolfe 1998). Thus, at the moment, the use of trophic factors to improve muscle mass and glucose metabolism must be viewed as a research tool only. Important questions remain to be answered. DHEA. Dehydroepiandrosterone (DHEA) is the adrenal steroid hormone found in the highest concentration in human plasma. Its serum levels decrease as a function of age in humans. DHEA levels begin to rise at puberty, peak in the late teens (teenaged males have about 50% more DHEA than teenaged females), and then decline in a linear fashion at a rate of about 10% per decade until low and equivalent levels are reached in both sexes at age 75 (Orentreich et al. 1984). The levels in normal human serum can be modified by various metabolic, physiological, and behavioral factors. DHEA is believed to have no specific effect of its own; rather it is transformed within the body’s cells into a variety of active androgens and/or estrogens, and this presumed steroid interconvertibility is postulated to underlie its varied physiological and pharmacological effects (Orenteich et al. 1984). DHEA is also thought to have an antioxidant effect. The hormone can be bought in health food stores and is written about from time to time in the popular press. Some evidence suggests that administration of DHEA suppresses the incidence of tumors in rodents, but these results are confounded by questions about whether caloric restriction was the operative mechanism in these experiments. Evidence suggests that DHEA acts as a physiological regulator of interleukin-2 biosynthesis (Yu 1995). If so, its ability to regulate such cytokines may be a plausible mechanism by which the molecule exerts its antitumor effects and putative anti-aging action. In any event, life-

233

time feeding of DHEA to healthy mice has no effect on their median or maximum life span (Orenteich et al. 1984). There is some evidence suggesting that DHEA may exert an inhibitory effect on corticosteroid secretion, and thereby indirectly act to stimulate the immune system. Some evidence suggests that short-term administration of DHEA to older humans improves their feelings of well-being in both objective and subjective terms (Orenteich et al. 1984). Hormone Induction via Exercise. An inspection of table 6.8 indicates that exercise can induce a fourfold increase in the levels of endogenous growth hormone, even in elderly individuals. This increase is comparable to the threefold increase in IGF-1 levels noted when older adults were given growth hormone directly. The beneficial and anti-aging effects of growth hormone have already been enumerated. It stands to reason, therefore, that the simplest and cheapest way to induce growth hormone levels to the point where they exert anti-aging effects is to exercise. Perhaps an exercise-induced hormonal effect underlies the observation that individuals who engage in aerobic exercise and are cardiovascularly fit have a less loss of brain tissue than do individuals who are not fit (Colcombe et al. 2003). At the very least, we can exercise now while we wait and see if a safe and effective trophic response is developed. Estrogen. Estrogen has been shown to decrease the mitochondrial production of free radicals in females relative to males (see table 8.5). One might expect estrogen replacement therapy in menopausal women to have a beneficial effect. Indeed, among women who generally led a healthy life (Seventh Day Adventists), those who had used estrogen replacement therapy had an expected age at death of 89.6 years compared to 79.8 years for their cohorts who did not use hormone replacement therapy (Fraser and Shavlik 2001). This 10-year difference was not confirmed in large-scale studies of the general population. In fact, one study using a combination of estrogen and progesterone replacement therapy was canceled due to concern over a higher breast

234 Chapter 6 Altering Aging cancer incidence, and another study using only estrogens did not show any preliminary effect on longevity (Turgeon et al. 2004). Older women who were long-term users of estrogen replacement therapy exhibited no protective effects of the hormone on their muscle aging: they had the same incidence of sarcopenia as did the control group (Kenny et al. 2003). Perhaps the apparent discrepancy between these various studies may be explained by assuming that high- and low-risk factors interact in a nonlinear manner such that the beneficial effects of estrogen can only be expressed in the absence of any high-risk behaviors—a condition which is not often met in the general population. 6.3.2.6 Antioxidant Supplementation in Humans

Chapter 10 points out that dietary supplementation of antioxidants to various laboratory animals is only occasionally effective in prolonging life, although such treatments are believed to be effective in reducing morbidity. Certainly, individuals with deficiencies in serum levels of various antioxidants are at a significantly increased risk for cardiovascular disease, cancer, and other age-related pathologies. Recent epidemiological studies showing that supplementation with specific types of antioxidants offers the potential for slowing the onset of some age-associated diseases and for retarding physiological declines. As discussed earlier, several studies have shown that individuals who consume high levels of antioxidant vitamins, primarily vitamin E, either through their normal diet or by supplementation, are at reduced risk for cardiovascular disease (Meydani, 2001; Rimm et al. 1993; Stampfer et al. 1993). There is also evidence that vitamin C provides some protection against non–hormone-dependent cancers (Block 1991) and that it may lower blood pressure of borderline hypertensive subjects (Troute 1991). These studies are encouraging but not definitive. It is possible for a compound to have more than one effect in vivo. Ascorbic acid is usually a free-radical scavenger, but in the presence of excess iron it becomes a free-radical generator. To-

copherol (vitamin E) may act as a procoagulant, and has differential effects in healthy vs. ill people, as described above. Thus, predicting the in vivo results of pharmacological doses of these natural products is difficult. Data from a double-blind, placebo-controlled prevention trial of the effect of daily supplementation of alpha-tocopherol, betacarotene, or both on almost 30,000 Finnish smokers over a period of some 5 years underscore this caution. This study reported that participants who received alpha-tocopherol showed an 18% increase in the incidence of cancer as compared to participants who did not receive this supplement (Alpha-tocopherol, Beta Carotene Cancer Prevention Study Group 1994). The disappointing findings of the Finnish study were reinforced by two discontinued large-scale studies in the United States: the Beta-Carotene Retinol Efficacy Trial (CARET) and the Physicians’ Health Study. In one study involving more than 18,000 people, people who smoked and people who were exposed to high levels of asbestos were given daily doses of beta-carotene, vitamin A, or a placebo for approximately 8 years. The individuals taking betacarotene experienced a 28% increase in the incidence of cancer and had a 17% greater rate of mortality as compared to the control group. There were no differences between the groups in the incidence of cardiovascular disease. Although one may raise questions regarding the experimental protocol of the Finnish and CARET studies, such as the dubious inclusion of only individuals already at high risk, chosen from a population that has one of the highest mortality rates in Europe, these results, combined with the negative findings on betacarotene in the Physicians’ Health Study, clearly demonstrate that not all antioxidants are equally useful, and some judgment must be exercised. Consider the increasing incidence of insulin resistance in elderly people. Type 2 diabetes is the most common metabolic disease in the United States, and its incidence is increasing in part because of its association with obesity. It is generally accepted that the onset of diabetes coincides with an impaired glucose metabolism in the skeletal muscles and liver such that these cells no longer respond effectively to the insulin signal. A strong positive correlation has been observed

6.3 Useful Methods of Modulating Aging Processes in Humans

between intramuscular fat content and the insulin resistance. Even healthy lean elderly individuals show a 45% increase in their intramyocellular lipid content, and this is accompanied by a 40% decrease in their rates of mitochondrial oxidative and phosphorylation activities when compared to the young controls with comparable body composition values (Petersen et al. 2003). Defective mitochondrial activity might be related to the increased lipid content. The beta-oxidation of fatty acids takes place in the mitochondria. The fatty acids must be transported into the inner membrane of the mitochondria by the carnitine cycle. However, the carnitine content of the mitochondria decreases with age, so one might expect to find an alteration of intracellular lipid content with age if the transport mechanism is slowing down. This is correlative evidence and does not specify causation, but it does suggest that the age-related incidence of diabetes may stem in part from age-related alteration in fatty acid transport and metabolism. In that case, supplementation with carnitine might alleviate diabetic symptoms to some extent, and such an effect is noted in animal studies (Ames and Liu, 2004). Based on those studies, a food supplement containing both alpha-carnitine and alpha-lipoic acid is now being marketed. Its use is, again, a personal decision which each individual must make for themselves, but its use seems to be based on good science. The FDA estimates that there are 29,000 dietary supplements on the market today, with sales of more than $18 billion a year. Some of these supplements are known to be unsafe, such as ephedra, which the FDA has now banned. A few of these supplements are known to have some efficacy (e.g., gingko is reported to have mild antioxidant activity in the lab; carnitine/lipoic acid mixes decrease production and damage of reactive oxygen species in normal and diabetic rats). But the safety and efficacy of most of the supplements are unknown. The situation is complicated because most people take more than one supplement or combine it with a prescription medicine, and so the possible number of drug interactions could be astronomical. Given this background, one should have a sound scientific

235

basis for deciding to take a dietary supplement, and one should always inform their physician of such self-medication. 6.3.2.7 Lifestyle Factors

It is likely that there is no silver bullet, no one pill or procedure that will significantly slow down senescent processes. Maximal protection against senescent morbidity and mortality will probably arise from the interaction of a number of different interventions. We can view these interventions as a lifestyle that minimizes certain risk factors. Several studies have identified a common set of risk factors that, when adopted, significantly extend longevity or compress morbidity for those who practice that lifestyle compared to those who don’t. These risk factors include a low glycemic diet, regular resistance and aerobic exercise, a low and steady BMI, avoidance of smoking, moderate alcohol consumption, and certain types of hormone replacement therapy, all of which are commonly offered as part of a good preventive medicine program. A health survey of Seventh Day Adventists regarding their incorporation of these risk factors was correlated with state mortality data (Fraser and Shavlik 2001). Those Adventists with lower risk factors (i.e., greater incorporation of healthy practices) had a life expectancy at birth of 80.2 years (men) or 84.8 years (women), which was significantly greater than that of the general U.S. population (73.0 and 79.7 years for men and women, respectively). Individuals with very low risk-factor levels had an expected age at death of 85.1 years (men) or 89.6 years (women). The mortality difference occurs in the 60+ ages, suggesting that the low risk factors are slowing down the rate of senescence. The very small difference in the male–female life spans in the low-risk group relative to the general population suggests that men benefit more than women from this lifestyle. It is likely that these factors interact to significantly reduce the mitochondrial production of free radicals and otherwise slow down the senescent processes, as discussed earlier. This choice of lifestyle can result in an extra 7–10 years of life. To put this in perspective, realize that

236 Chapter 6 Altering Aging demographers have predicted that for life expectancy to increase by this much in the general population will take five or six decades. Yet we already know how to achieve that dream by making a conscious lifestyle choice. It is an interesting phenomenon that most of us will not make that choice to live longer and healthier. Perhaps this book will spur you to make a pro-longevity choice. A similar type of study done with people who had been university students in 1939–1944 period showed that the rate of morbidity and functional decline was highly correlated with the individual’s risk factor exposure (Hubert et al.

2002). The moderate- and high-risk individuals had death rates that were 1.3 and 1.9 times higher than the reference low-risk group. There was no acceleration of functional disabilities in the lowrisk group in the years immediately preceding death, but this acceleration was obvious in the other two, and it began earlier in the high-risk group. Both this study and the Adventist study suggest that the rate of senescence is accelerated in individuals whose lifestyle exposes them to certain high risks, and this accelerated senescence results in greater loss of function preceding death and a significantly shorter life span (10 years less).

7

Genetic Determinants of Longevity in Animal Models

7.1 Introduction It has long been a popular axiom that our inheritance plays a major role in determining our length of life. Indeed, an apocryphal saying prescribes that, if one wishes to live a long and happy life, one should first arrange to have parents who are long-lived and wealthy. But note that this truism immediately mixes together both genes and environment, for who would doubt that wealth alters the context within which our genes interact with each other and with the environment. Successful identification of the genetic determinants of longevity thus depends to a large extent on our ability to minimize environmental effects while maximizing genetic effects. Only then can we identify which, if any, genes play a major role in significantly extending (or decreasing) longevity. The need to minimize the environmental factors means that we must deal with simple environments. We humans have constructed, live in, and interact with an incredibly rich mix of complex environments; our study must start not with us but with experimental animals in which we can rigorously control the environment so as to highlight the genetic effects. The fact that many studies on the biology of aging cannot, for obvious ethical, legal, and practical reasons, be done with human beings reinforces this emphasis. The fact that our experimental procedures necessarily focus on accentuating genetic factors affecting longevity while simultaneously minimizing environmental ones means that only a simplistic misreading of this chapter would support

an uncritical acceptance of genetic determinism (Lewontin 2000). However, what genetic analysis will unambiguously identify are the genetic and metabolic pathways involved in longevity regulation and the environmental conditions necessary for their expression. These dual and complementary sets of information are not only much more useful than the identification of any one gene, but will be absolutely critical to accurately understanding the process. We will also want to know whether any particular aging mechanism or senescent process is one which is, so to speak, either “public” (i.e., broadly conserved across phylogenetic lines including mammals and thus of general importance to the understanding of human aging) or “private” (i.e., restricted to species other than mammals and, while interesting, not likely to directly assist us in understanding human aging; Martin et al. 1996).

7.2 The Need for a Model Organism, and the Cost Only a few species are well suited for genetic investigations, in the sense that there are both a substantial body of knowledge regarding the organism’s heredity, chromosome structure, life history, development, maturation, adult physiology, and reproduction and a stock center that maintains all the various wild-type and mutant strains. The knowledge and the mutants constitute the basic tools for a genetic investigation into aging processes. The four species that best meet

237

238 Chapter 7 Genetic Determinants of Longevity in Animal Models these requirements include the single-celled yeast Saccharomyces cerevisiae, the small nematode worm Caenorhabditis elegans, the fruit fly Drosophila melanogaster, and the laboratory mouse Mus musculus. Other species have been used, as was pointed out in chapter 4, but the focusing of efforts on these four has led to knowledge so detailed that we can now recognize deep similarities in their public aging mechanisms. These four species superficially appear to be very dissimilar to humans. The obvious biological differences between insects and vertebrates do not necessarily constitute a major disadvantage to using the fruit fly, for example, as a model for the genetic analysis of aging. The experimentally verified phenotypic expressions of aging in Drosophila listed in table 4.6 bear a striking resemblance to the age-related changes that take place in humans, as described in chapter 5. This probably reflects the fact that biological mechanisms fundamentally important in one species are generally important in most species, a fact that even the medical journals now acknowledge (Hariharan and Haber 2003). The major gene systems regulating aging are highly conserved public mechanisms; their characterization required the integration of data from all four of the species listed above. This sort of molecular homology validates the usefulness of comparative animal research on aging. That being said, it must be noted that all four of these model organisms are intrinsically short lived. It may well be that studying species that are intrinsically long lived would yield insights into mechanisms of slow aging that do not exist in our short-lived models. In fact, I note in chapter 10 that the study of long-lived birds reveals the existence of such adaptations for long life in their mitochondria. Although the study of certain long-lived organisms is increasing (the July 2003 issue of Experimental Gerontology is devoted to this topic) , they are simply too rare or costly or too inconvenient to use to describe basic aging mechanisms. The evolving strategy is to characterize the basic mechanisms in great detail using model organisms, and then use one or more of the long-lived species to investigate if and how specific details differ between the two.

The laboratory experiments done over the past few decades have allowed the robust identification of four different but intertwined genetic and physiological pathways that regulate longevity in all four of our model systems. These four longevity processes may be summarized as follows: (1) metabolic control, (2) stress resistance, (3) genetic stability, and (4) reproductive effects. There are three subcategories under the rubric of metabolic control: caloric restriction, insulinlike signaling system, and nuclear-cytoplasmic interactions. The first two are conceptually cohesive; the last one is very heterogeneous and includes a variety of mechanisms united primarily by a longevity-related interaction between nuclear and mitochondrial genomes. Be aware that organizing the data into four separate categories is a pedagogical tactic adopted to make it easier to understand this messy pile of data; it should not be interpreted as indicating that there are four separate longevity processes that do not interact with one another. Perhaps it would be best to view the four categories as being rather porous, with significant exchange of molecules among them. Finally, information regarding different patterns of senescence is still sketchy, but I deal with it later in this chapter. To provide some coherence and ready comparison to our complicated tale, I structure our discussion of each model organism around the evidence indicating the effects of each of the above four processes on the longevity of that organism. Table 7.12 summarizes the main results of this discourse. Note that the organization of this chapter allows an evolutionary or comparative view of any one mechanism by reading about the same longevity process in all four groups, while reading about all processes in one species allows a comprehensive overview of all the major longevity-determination mechanisms in one organism. Reviewing table 7.12 at various times during your reading of this chapter may be a useful step in grasping the broad outlines. My conclusions are necessarily an overly simple approximation of the genetic events that take place in complex environments. The genetic mechanisms described in the following pages do not represent the complete story of longevity

7.3 Saccharomyces cerevisiae

regulation in each of these species. What they do represent are the substantial, although still incomplete, characterization of mechanisms that I endeavor to pull together in an integrated theory of aging in chapter 14.

7.3 Saccharomyces cerevisiae 7.3.1 Background Information The budding yeast Saccharomyces cerevisiae is probably the fungus with which we are most familiar. Each cell has a finite, age-dependent life span and undergoes a series of characteristic agerelated changes (see figure 4.9). There are two methods of measuring aging in this organism: replicative aging, in which the life span is measured by the number of new buds (offspring) that the mother cell produces by mitotic divisions (Mortimer and Johnson 1959); and chronological aging, in which the life span is measured by the length of time that the nonbudding cell can survive in a stationary-phase culture (Longo et al. 1996). There are advantages and disadvantages of both methods, and so different labs use one or the other metric depending on the particular question being asked. Both methods have been used to uncover different aspects of the longevity mechanisms at work in this yeast, and the combined data have led to some interesting conclusions. The fact that the replicative mean and maximum life spans are characteristic features of any given yeast strain, even though these parameters may vary considerably from one strain to another, suggests that the yeast life span has a strong genetic component (I. Muller et al. 1980). This fact, as well as the observation that the senescence phenotype is dominant and is determined by soluble cytoplasmic factors (Egilmez and Jazwinski 1989), suggests that there are differences in the expression of specific genes during the yeast life span and that the expression of such genes late in life may bring about the production of the hypothesized senescence factors. This interesting genetic hypothesis was tested directly by means of molecular

239

techniques. Egilmez et al. (1989) used a differential hybridization procedure to identify yeast genes that are preferentially expressed in either young or old cells. Most of the genes are expressed at the same level in both young and old cells; six were not and appear to be age specific. Five of these were found preferentially in young cells, and one was expressed preferentially in old cells. Since these investigations, 14 genes that display differential patterns of expression have been isolated; I will discuss the role that three of these genes appear to play in regulating yeast longevity. One of the first genes analyzed was the LAG1 gene, or longevity assurance gene 1 (D’Mello et al. 1994). LAG1 is expressed predominantly in younger cells and shows a marked decrease in its expression as aging proceeds, finally disappearing at about generation 18 (figure 7.1A). The gene was mapped to the right arm of chromosome VIII. Sequencing revealed it to be a unique gene without significant sequence homology to any other known yeast genes. It codes for a transmembrane protein that facilitates the transport of GPI (glycosylphosphatidylinositol)-anchored proteins (Barz and Walter 1999), and it is localized in the endoplasmic reticulum. Humans and nematodes each have a homologous gene that can functionally substitute for a missing LAG1 gene (Jiang et al. 1998), which suggests that the gene and its function are broadly conserved. The human gene is expressed in brain, skeletal muscle, and testis; its precise function is not yet known, but it may have something to do with cell survival and/or signal transduction. Whatever its function, null mutants of the gene have no observable effect on cell growth or metabolism, but they have a dramatic effect on yeast longevity such that the mutants live significantly longer than the wild type (figure 7.1B). This experiment shows that some genes affecting aging may be detected simply by means of their asymmetric time of expression. The other two genes analyzed by Jazwinski (1993) are the two yeast homologues of the mammalian proto-oncogene c-ras, called RAS1 and RAS2. Knowing that the RAS gene plays a key role in regulating many important cell and metabolic processes in both mammals and yeasts, Chen et al. (1990) transformed yeast cells with

240 Chapter 7 Genetic Determinants of Longevity in Animal Models

Relative mRNA levels

A

and in stress resistance, and I discuss them further below. The importance of these earlier studies is that they demonstrated that one could use standard techniques to measure and manipulate the life span of individual yeast cells, thereby opening up the study of aging in these unicellular organisms.

5 4 3 2 1 0

5

0

15

10

20

Age (generations)

Percent surviving

B

100

LAG1 deletion 50 Wild type

0

0

10

30 20 Age (generations)

40

Figure 7.1 (A) The mRNA transcript levels of the longevity assurance gene 1 (LAG1). When measured in yeast of different ages, these data show that the gene has an age-related pattern of expression. The expression of this particular gene peaks in young yeast cells and then declines. (After D’Mello et al. 1994.) (B) The life span of yeast cells with a deletion in the LAG1 gene. These yeast cells live longer (a mean of 25 generations) than do normal controls with a complete LAG1 gene (mean of 17 generations), suggesting that the normal gene is acting as a negative regulator of longevity. (After D’Mello et al. 1994.)

the Harvey murine sarcoma virus gene, v-Ha-ras, and found that both the mean and maximum life span of the cells almost doubled when the transformed genes were expressed at a moderate level. Two different conserved genes, RAS1 and RAS2, were identified as affecting longevity but in different manners. Extended longevity in the two mutants depends on underexpression of RAS1 but overexpression of RAS2 (Sun et al. 1994). Both genes have pleiotropic effects and affect a large number of different cellular processes. Later studies showed that these two genes play important roles in nuclear-mitochondrial interaction

7.3.2 Metabolic Control via Caloric Restriction Yeast respond to caloric restriction (CR) by extending their life span. This is shown in figure 7.2A, which shows that wild-type yeast raised in a media containing 0.5% glucose have a higher mean and maximum life span than those raised in 2% glucose. Similiar responses were observed by other investigators using other strains (e.g., Jiang et al. 2002), suggesting that this CR response is a widespread property of yeast. A similar effect on longevity is obtained if one grows a hexokinase-null mutant on 2% glucose (figure 7.2B) or if one simply deletes the glucose-sensing genes of the cell (figure 7.2C). Hexokinase is an enzyme essential for the initial utilization of glucose by the cell; mutating it has the effect of reducing the cell’s food supply. Rendering the cell insensitive to glucose has the same effect. Thus, any mechanism that blocks the uptake and metabolism of glucose has the potential to act as a CR mimic. Glucose is an essential metabolite for energy production, and high levels activate many important biochemical pathways. Without additional evidence, it would be difficult if not impossible to decide which of these pathways were actually playing a role in bringing about the CR effect on replicative longevity. One successful way to obtain the necessary evidence is to test the longevity of yeast containing one null mutant in a candidate gene pathway. If the pathway is not essential to the CRdependent longevity increase, then there should be no effect on life span. But if the candidate pathway plays an important positive or negative role in the CR-dependent longevity effect, one should see a significant alteration of life span even in the presence of high levels of glucose.

7.3 Saccharomyces cerevisiae

A

1

Fraction viable

WT, 2% Glu cdc25-10, 2% Glu 0.75

WT, 0.5% Glu cdc25-10, 0.5% Glu

0.5

0.25

0 1

Fraction viable

0 B

0.75

10

20

30

40

50

WT hxk2

0.5

0.25

0 1

Fraction viable

0 C

0.75

10

20

30

40

50

WT gpr1 gpa2

0.5

0.25

0 10

20

30

40

50

Figure 7.2 Limitation of glucose availability or attenuation of glucose signaling extends life span. (A) Longevity is increased by growth on low-glucose medium or by attenuation of cAMP-protein kinaseA signaling. (B) Deletion of hexokinase gene HXK2 extends life span. (C) Deletion of the glucose-sensing genes GPR1 or GPA2 extends life span. (After Lin et al. 2000.)

Several important pathways have been assayed in this manner, and the results are instructive. The yeast cAMP protein kinase A (PKA) pathway is activated by high levels of glucose. If one mutationally inactivates any single component of the

241

PKA pathway, PKA mutants that decrease PKA activity also result in a significant increase in longevity relative to wild type, whereas the few mutants that increase PKA activity result in a significant decrease in longevity (Lin et al. 2000). This result not only tells us that the PKA pathway regulates life span, but it also gives us a general principle: nutritional or mutational conditions can alter PKA activity in a positive or negative direction, and this will then result in an inverse effect on the yeast cell’s replicative longevity. A more complex set of results were obtained when a similar pathway analysis was done to determine if any members of the SIR (i.e., silent information regulator) gene family played a role in the CR effect (Guarente, 1999). Although initially found in yeast, homologous genes have been found in all model species, and so they have been grouped together as members of the sirtuin (i.e., sir-like) family. The sirtuin genes are a group of related nicotinamide adenine dinucleotide (NAD+)-dependent histone deacetylase enzymes that play an important role in altering gene expression, and they have been generally implicated in longevity extension. As their name implies, they remove acetyl groups that have been posttranslationally attached to particular amino acid (lysine) residues in certain histone proteins. The effect of removing the acetyl group varies with the function of the particular protein involved. The most common (but not the only) effect takes place in nucleosomes, where removing the bulky and charged acetyl group allows the DNA to recondense about the nucleosome structure. This condensation makes it difficult for polymerases and other necessary factors to bind to the gene promoters and effectively transcriptionally inactivates (i.e., silences) any genes in that region of the chromatin. However, some (nonhistone) enzymes are inactivated by acetylation and activated by deacetylation. As a general rule, then, deacetylation may be viewed as a biological switch that allows various proteins to shift in a controlled manner from one functional state to another. The requirement for NAD+ is unusual but is thought to allow proteins to change their functional state in coordination with the overall metabolic state of the cell. This comes about because NAD+/

242 Chapter 7 Genetic Determinants of Longevity in Animal Models appears incapable of expressing the CR response of extended longevity in response to low glucose levels if it contains a null mutant in one of the sirtuin genes (SIR2; figure 7.3A; Lin et al. 2002). This implies that the SIR2 gene is a necessary component part of the CR pathway. However, in a different yeast strain, Jiang et al. (2002) found that a null SIR2 mutant did not prevent life extension when fed low levels of glucose (figure 7.3B), although the mutation did significantly reduce the mean life span on both high- and low-

NADH are thought to be important metabolic cofactors, and the deactylase enzyme can be active only when the metabolic state of the cell permits the transient accumulation of NAD+. Although this hypothesis is still not fully accepted, there is some independent evidence to support it (Araki et al. 2004). There appear to be significant strain differences regarding the role of the histone deacetylase enzymes in bringing about the expression of the CR pathway in yeast. One strain of yeast cells

A

1

WT, 2% glu WT, 0.5% glu sir2

fob1

, 2% glu

sir2

fob1

, 0.5% glu

0.5

0.25

0 0

10

20

30

40

50

60

Generations B

100 SIR2 (2% glucose) SIR2 (0.1% glucose) sir2 (2% glucose)

Survival (% live)

Figure 7.3 (A) Caloric restriction (CR) extends lifespan and enhances rDNA silencing in a Sir2-dependent manner. Wildtype cells grown under CR conditions live long, and this CR response is not expressed if the cells contain a mutant sir2 gene. (After Lin et al. 2002.) (B) The same type of experiment done by a different lab using a different wild-type strain yields a slightly different effect. CR still enhances life span in normal (SIR2) strains, but its effect is additive with that of a sir2 mutation such that the longest life span is seen in sir2 mutants raised under CR conditions. (After Jiang et al. 2002.)

Fraction viable

0.75

sir2 (0.1% glucose) 50

0 0

10

20 Age (generations)

30

40

7.3 Saccharomyces cerevisiae

glucose media. This implies that the SIR2 gene is not a necessary part of the CR pathway but that it does affect longevity in some manner. Both sets of data show that SIR2 activity is required for maximum longevity. Furthermore, whatever the SIR2 gene is doing in these strains, it must be affecting an important public aging mechanism because its homologue in nematodes (sir2.1), as well in the fruit flies and mammals, is essential for the operation of the insulinlike signaling system, a most important pathway (see next section). Thus, different strains of yeast cells have different expressions of a common longevity pathway. Perhaps the two strains have different alleles of the SIR2 gene with different properties, or perhaps there are allelic differences in their target genes. Taking all the data together, it is reasonable to conclude that the SIR2 gene plays a major role in modulating the cell’s response to CR, and so it is regarded as a public (i.e., conserved) mechanism in that regard. However, there are also strain-specific differences in that modulation, which can be regarded as private (i.e., nonconserved) variations on a general mechanism. Such genetic heterogeneity complicates analysis but should not surprise us. In fact, it may provide us with a potential mechanism for better understanding the origin of individual differences in human aging. Let us examine the role of SIR2 in the CR response of the strain that requires its presence. The wild-type SIR2 gene is thought to transcriptionally silence the chromosomal region containing the rDNA loci. In the presence of a mutated SIR2 gene, the yeast cell’s nucleolus underwent a progressive enlargement and fragmentation due to the accumulation of extrachromosomal rDNA circles (Sinclair and Guarente 1997). The existence of these circles was interpreted to mean that the stability of the yeast cell’s genome was adversely affected, leading to the death of the cell. Factors accelerating the accumulation of such circular molecules result in a shorter life span, whereas factors delaying rDNA circle accumulation yield an increased life span. This process was thought to be under SIR2 control and constitute the major anti-aging effect of this gene (Guarente 1999). However, this interpretation is controver-

243

sial because not all yeast strains exhibit rDNA circles with aging. Even though Guarente drew a plausible connection between rDNA circles, SIR2, and longevity by postulating that genetic stability of the rDNA genes is important for longevity, its restricted occurrence suggests that this rDNA circles–histone deacetylase–longevity relationship, while real, may be a private pleiotropic effect peculiar to some strains of yeast (Sinclair 2002). Another possible hypothesis about the SIR2– CR response is suggested by recent findings at the DNA repair and metabolic level. The first case involves the mammalian Sir2 (SIRT1) gene, which is another member of this same NADdependent deacetylase family. One of its target genes is the Ku70 DNA repair factor. When tightly bound to the Bax protein in the mitochondria, the deacetylated Ku70 factors prevent apoptosis (programmed cell death; see chapter 11) from occurring. If the Ku70 factors are acetylated as a result of cell damage, then the Bax protein is released, apoptosis takes place, and the cell dies. But this can be prevented if SIRT1 deacetylates the acetylated Ku70 factors, for this inhibits apoptosis and thus promotes mammalian cell survival (Cohen et al. 2004; Wood et al. 2004). The net effect of this is to extend longevity in yeast as well as in metazoans. The second case involves acetyl-coenzyme A (CoA), which is a key metabolite linking the Emden-Myerhoff pathway to the Krebs cycle. Biochemical studies have shown that the synthesis of CoA is accomplished by an enzyme called acyl-CoA synthetase. The activity of this enzyme depends on whether it has been post-translationally acetylated at the lysine-609 position, for such an acetylated form of the enzyme is inactive (Starai et al. 2002). If the acyl-CoA synthetase is acetylated and thus inactive, then CoA cannot be made in sufficient quantities, and the flow of nutrients through the Krebs cycle will suffer. This presumably inefficient production of ATP molecules from glucose might explain the inability of SIR2 mtuants to respond to CR, for only cells with the ability to deacetylate and thus activate acylCoA synthetases are capable of expressing a CRdependent extended longevity.

244 Chapter 7 Genetic Determinants of Longevity in Animal Models There are likely to be other cases where sirtuindependent deacetlyation plays an important role. For example, another human SIR2 homologue (SIRT3) is an NAD-dependent deacetylase located in the mitochondria of metabolically active tissues (Onyango et al. 2002). Its target genes are unknown, but its existence indicates the possibly important role of this gene family in connecting metabolism and longevity at some basic level in the cell organelle that plays the key role in metabolic control. It may be pertinent that the Krebs cycle takes place in the mitochondrion and this is the entry point of the CoA metabolite into the cycle. Given all the above information, it may be that many key metabolic enzymes are variously regulated by deacetylase enzymes. These overlapping functions may account for some of the apparent differences between yeast strains. This assumption of mitochondrial involvement is partially supported by the finding of Lin et al. (2002) that high levels of respiration (i.e., the flow of nutrients through the Kreb’s cycle and into the electron transport chain) are essential for the expression of CR-dependent extended longevity. Neither yeast with mutated (and thus inactive) SIR2 proteins, nor yeast with an active SIR2 protein but a mutationally inactivated electron transport chain were capable of engaging in respiration nor in expressing an extended longevity (Lin et al. 2002). Both components are essential for the expression of the CR phenotype in this strain. In addition, a gene localized in the peroxisome and involved in the biosynthesis of nicotinamide, PNC1, functions in this same pathway with SIR2 (Anderson et al. 2003). As nicotinamide both acts as an inhibitor of SIR2 activity and also functions in the NAD salvage pathway, then PNC1 probably plays an indirect role by its ability to remove the inhibitor and thus activate the pathway. Microarray analysis of the yeast genes differentially affected by nutritional or mutational treatments showed that CR changes the gene expression patterns of 133 genes relative to the normal situation (Lin et al. 2002). This genebased metabolic shift results both in an increased yield of ATP per input glucose molecule as well

as in an increased NAD+/NADH ratio of the mitochondria. The resulting high levels of NAD+ could regulate the NAD-dependent SIR2 proteins, as described above (Araki et al. 2004). Lin et al. (2002) observed that the CR-dependent extended longevity did not appear to confer any increased resistance to oxidative stress arising from either paraquat or hydrogen peroxide exposure to the long-lived cells. This finding is rather different from the description of CR effects noted in other yeast strains and in other organisms (see below). However, resistance to oxidative stress is enhanced in the postreproductive life span of stationary-phase yeast, as well as in the extended longevity phenotypes of both the nematode and the fruit fly (see below), both of which have postmitotic adults. Perhaps the difference in the yeast extended-longevity phenotype arises from the fact that it is a reproductive cell in which we are measuring replicative life span in which metabolic control is more effective in extending longevity, whereas stress resistance is more effective in extending longevity of postmitotic cells. An examination of the effects of three different histone deacetylase enzymes showed they had distinguishable but overlapping effects on CR-induced longevity extension (Jiang et al. 2002). The enzymes tested each represent one of the three known classes of histone deacetylase enzymes, each of which has its own specific patterns of transcriptional repression and yield diverse gene expression patterns and physiological outcomes (Chang and Min 2002). Deletion of the RPD3 (class I) gene extended life span, and there was no additive effect of CR on this mutationally induced longevity. This implies that the RPD3 gene may be a part of the CR pathway. Deletion of the HDA1 (class II) gene had no effect on longevity, but it acted synergistically with CR to increase life span. This implies that the HDA1 gene acts as a nonessential but useful cofactor in the CR pathway. Deletion of the SIR2 (class III) gene shortened life span but did not prevent the CR effect. This implies that the SIR2 gene acts in a pathway different from, but partially overlapping, the CR pathway. Thus, in this yeast strain, the histone deacetylase en-

7.3 Saccharomyces cerevisiae

zymes play supporting but not starring roles in the CR pathway. Examining the evidence presented above leads to the conclusion that, in this particular yeast strain, the pathways involved in CR-dependent extended longevity must begin at the cell membrane with the recognition of glucose, involve the PKA pathway, be mediated somehow by SIR2 and/or PNC1, involve other histone deacetylases such as RPD3 or HDA1, and terminate in the nucleus with the up- or down-regulation of a large number of specific genes, which then significantly affect the metabolic and physiologi-

245

cal profile of the cell so as to allow it to live longer under conditions of energy scarcity. The mitochondrion is clearly involved in this regulatory process; understanding its precise regulatory role will be a difficult but informative task (see figure 7.4). The extension of longevity by means of caloric restriction is clearly a universal aspect of all model systems. After we discuss the mechanisms underlying the increase in life span brought about by an increased stress resistance (see below), then we shall combine that information with the CR data described above to arrive at the basic cellular longevity regulation mechanism (see figure 7.7).

Strain 2

Strain 1 Low Glucose gls ON

?

RAS2

RAS2

?

?

SIR 2 (III) Non-CR Longevity Effect

RPD3 (I)

SIR 2 (III)

?

HDA1 (II) ?

?

?

CR Induced Extended Longevity

?

Figure 7.4 A synthesis of the initially contradictory data into a possible model of the yeast CR pathway with putative strain-specific substitutions of one histone deacetylase enzyme for another. Roman numerals designate the HDA class.

246 Chapter 7 Genetic Determinants of Longevity in Animal Models to the bacterial phosphatases that hydrolyze the transcriptional regulators ppGpp and pppGpp, suggesting the possibility of an evolutionary history for this protein (see chapter 4). The upstream involvement of a transcription factor ensures that many other nuclear genes are likely to be affected by this nuclear–mitochondrial interaction. And, in fact, biochemical studies clearly showed not only that the enzyme activities of citrate synthetase and acetyl-CoA synthetase were significantly lower than normal, but that there were existed numerous defects in the cell’s metabolic abilities when either RTG1 or RTG2 was deleted, leaving an inactive transcription factor (Small et al. 1995). All the evidence indicates that some nuclear genes are affected by the functional state of the mitochondria. This then implies that there must be ongoing communication between the different cellular organelles so as to allow coordination of their separate but interconnected metabolic processes. A role of this communication pathway in regulating longevity was suggested when it was shown that yeast without any functional mitochondria (traditionally called petites) live longer than do normal yeast with a full complement of functional mitochondria (figure 7.5A). Even more interesting was the fact that both RAS2 (figure 7.5B) and RTG2 (figure 7.5C) gene expression were required if the nuclear– mitochondrial interaction was to affect longev-

7.3.3 Metabolic Control via Nuclear–Mitochondrial Interaction Some time ago, an intracellular signaling system from the mitochondrion to the nucleus was discovered in yeast and named retrograde regulation (Parikh et al. 1987). This system appears to be a mechanism by which the cell can adjust to changes in mitochondrial function by readjusting the transcription of nuclear genes. It appears as if mitochondrial–nuclear communication is a widespread phenomenon but one which the different model organisms have solved each in their own way. It thus may be viewed as a public aging mechanism with species-specific solutions. In view of this, I refer to this mechanism “nuclear-mitochondrial interaction” to indicate the nature of the process in generally applicable terms. It was first shown in yeast that the expression of nuclear genes RTG1, RTG2 are dependent on the functional state of the mitochondria (Liao and Butow 1993). The effect is not localized to these two genes, for the RTG2-encoded protein (Rtg2) is thought to facilitate the formation of an active heterodimer of the Rtg1 and Rtg3 proteins, which together form a transcription factor that induces gene expression from the retrograde response element found in the promoters of certain nuclear genes (Rothermel et al. 1997). The Rtg2 protein has domain similarities

A

Survival (% live)

YPK9

0

10

20 Age (generations)

30

40

7.3 Saccharomyces cerevisiae

B

247

100 grande

Survival (% live)

petite grande ras2 petite ras2

50

0 0

10

20

30

40

50

Age (generations)

Age (generations)

C

100

50 YPK9 RTG2 rtg2

0 0

10

20

30

40

Age (generations)

Figure 7.5 (A) A petite yeast strain (open symbols) has a significantly extended longevity (mean = 26.8) relative to the same strain without the petite mutation (mean = 19.8). (After Kirchman et al. 1999.) (B) RAS2 expression is required if nuclear–mitochondrial communication is to increase the life span of petite yeast relative to wild-type (grands) yeast. The life span of the petite strain YSK365 (mean 29.9) is reduced (P < .001) when a disruption of RAS2 is introduced into the strain (mean 16.5). The life span of the ras2 petite does not differ (P = .8) from the coisogenic grande strain (YPK9) containing the RAS2 disruption (mean 16.3). Disruption of RAS2 shortens life span (P < 0.01) relative to the coisogenic control YPK9 (mean 22.8). (after Kirchman et al., 1999.) (C) If activation of nuclear–mitochondrial communication were responsible for the extension of life span, then eliminating this response should also eliminate life span extension. Life span of YPK9 rtg2D (closed symbols; mean 19.0) and its coisogenic petite YSK365 RTG2 (open symbols; mean 17.7) on glucose. The two RTG2 deletion strains did not differ in life span (P = .2). Deletion of RTG2 completely eliminated the extension of life span seen in petites, confirming that nuclear–mitochondrial communication is necessary for the extension. (After Kirchman et al. 1999).

248 Chapter 7 Genetic Determinants of Longevity in Animal Models ity. The nuclear genes controlled by the RTG genes encode proteins found in the mitochondria, cytoplasm, and peroxisomes, and their activation somehow extends longevity. RAS2 interacts with this pathway somewhere downstream of the mitochondrial signal and may well determine life span by other processes. The physiological effect of induction of the retrograde response is an adaptation to survival on reduced energy and carbon sources (Jazwinski 2000a,b). What this means is a shift from the Krebs cycle to the glyoxylate cycle, a switch from the use of glucose to the use of acetate, and an increase in gluconeogenic activity. Rather than simply being a reduction in calories, the retrograde response represents a shift in the cell’s metabolism as a result of switching from using a high caloric content energy source (glucose) to one of lower caloric content (acetate; Jazwinski 1999). This nuclear–mitochondrial interaction is not the same as CR, even though the two phenomena send signals between similar organelles. The fact that yeast longevity depends on the proper functioning of several complex intracellular signaling systems may be the most important point of this discussion, for it implies that organelle-level regulatory decay within the cell

may underlie the loss of function characteristic of senescence and aging. I return to this topic in chapter 13. The diagram in figure 7.6 shows one possible interpretation of the data.

7.3.4 Stress Resistance There is a general correlation between stress resistance and extended longevity in yeast and other organisms. High levels of glucose activate the yeast RAS1 genes and the PKA pathway, leading to increased growth and a shorter life span. Low nutrient levels or inactivation of the RAS1 gene leads to decreased PKA activity and a concommitant increase in stress resistance and life span, particularly in stationary-phase (i.e., nonreplicating) yeast (Longo et al. 1996). However, RAS2 seems to play a major role in rapidly activating important stressresponse genes after the application of chronic environmental stress to the cell (Jazwinski 2000a). In fact, overexpression of RAS2 in the absence of any chronic stress extends longevity (see figure 7.5b), suggesting that stress gene activation may play a role in preventing oxidative damage and thus in delaying the onset of senescence. Protecting the cell against transient stresses seems to re-

CIT Rtg2p FUM

MKS1p

ACO Rtg1, Rtg3

Krebs (TCA) Cycle

RAS

Glutamate

MDH

ISOCIT CIT SDH

KETOGL Glyoxalate Cycle succinate

Figure 7.6 The original case of nuclear–mitochondrial communication in yeast which uses glutamate levels (a reflection of the level of Krebs cycle activity) to permit the activation of the glyoxalate cycle to supply succinate to the mitochondria if the Krebs cycle is not sufficiently active. (Adapted from Liu and Butow 1999.)

7.3 Saccharomyces cerevisiae

quire the combined effects of both the RAS1 and RAS2 genes, indicating that the two genes may modulate different regulatory pathways. There is some data suggesting that the RAS1 genes activate the phosphoinositol pathway. The ability of the two RAS genes to modulate a wide variety of vital processes led Jazwinski (2000a) to describe them as constituting a homeostatic device for yeast longevity. The fluctuating environment of the yeast cell continually alters the opposed activities of the RAS1 and RAS2 genes, thus allowing these opposed genes to set a dynamic equilibrium allowing the functional integration of the many different cell processes under their positive or negative control. If so, this reinforces the importance of regulatory signals to the cell’s ability to withstand stress and live long. Yeast transferred to water and maintained in stationary phase do not divide but can remain alive for periods of time, ranging from a few days up to a month or more (Longo et al. 1996). Such yeast cells are in a hypometabolic state, having decreased their metabolic rate and macromolecular synthesis by more than 100-fold on entry into stationary phase (Werner-Washburne et al. 1996). Even in a hypometabolic state, damage occurs. What defensive mechanisms underlie the long chronological life span of these stationary-phase yeast? Resistance to oxidative damage appears to be the main protective mechanism, for mutants deficient in either the cytosolic form of superoxide dismutase (CuZnSOD or SOD1) or the mitochondrial form (MnSOD or SOD2) die prematurely compared to wild type, particularly in well-aerated cultures where the high oxygen levels would likely accelerate the formation of reactive oxygen species (see chapter 10; Longo et al. 1996). Overexpression of either form of SOD extends survival by about 30% in some strains, possibly because it protects the mitochondrial aconitase enzyme (and other oxidative targets) from age-related inactivation (Fabrizio et al. 2003). This enzyme is a key component of the tricarboxylic acid cycle and is believed to be very sensitive to superoxide; thus knocking the enzyme out cripples oxidative metabolism. However, the overexpression of SOD2 increased the chronological life span but decreased the replicative life span (Harris et al. 2003). This

249

reduction in replicative life span occurred independently of the SIR2 gene effects on genetic stability; rather, it appears to be the result of accumulated damage in the mitochondria of old mother cells. Conversely, underexpression of the prohibitin genes also leads to differential effects on the replicative and chronological life spans (Piper et al. 2002). The prohibitin proteins influence the maturation and turnover of the respiratory chain components, and their expression usually decreases in senescing yeast. Absence of the proteins leads to an altered mitochondrial morphology and defective mitochondrial segregation. Thus, similar phenotypes can be generated by different gene alterations. When stationary yeast are exposed to oxidative stress, their MnSOD, CuZnSOD, and glutathione reductase mRNA levels, enzyme levels, and enzyme activities do not show a simple increase, but they change in complex ways (Cyrne et al. 2003). This is to be expected. Oxidative stress induces an increased turnover of all proteins, and the increased input of new proteins (including the antioxidant enzymes mentioned above) may be compensating for this increased proteolysis. This maintenance of steady-state protein levels may be essential for the maintenance of cell viability when it is under such stress and may preclude any large increases in activity of the antioxidant enzymes despite their increased mRNA levels. This then puts a cap on the amount of protection such enzymes can provide the cell, and thus the cell will eventually succumb to a prolonged and severe degree of oxidative stress. All these data are consistent with a role for superoxide-dependent mitochondrial damage in yeast aging and death. But lowering the glucose level in the media or deleting the RAS1 gene leads to about a 100% increase in longevity. This implies that pathways other than oxidative stress resistance contribute to the life span of the stationary-phase yeast. Mutational analysis has shown that the RAS2 functions at an early stage in a pathway involving the SCH9 gene, which negatively affects two transcription factors (Msn2 and Msn4) that repress multiple stress resistance genes and promote senescence (Fabrizio et al. 2003). Many of these different stress genes deal with factors other than oxidative stress, such as

250 Chapter 7 Genetic Determinants of Longevity in Animal Models heat shock and DNA repair (Maclean et al. 2003). One interesting aspect of this study was the finding that cells with only one defective DNA repair gene showed either no effect or only a mild decrease in longevity. However, cells with two or three defective DNA repair genes showed a large, nonadditive decrease in longevity. This synergistic effect suggests that the several DNA repair genes cooperate with each other to enable the attainment of a full chronological life span. The apparent redundancy of DNA repair genes has a functional purpose. The same may hold true for the other stress response genes. Thus, the stationary-phase yeast appears to depend for its long chronological life span on the fact that low glucose levels repress the pro-growth and pro-senescence pathways, and this repression relieves the normal inhibition of expression of a variety of stress-resistance genes. If one takes the information from the CR experiments discussed earlier and combines them with the data regarding the mechanism of stress resistance, then one can construct the canonical longevity regulation mechanism, as depicted in figure 7.7. The six-step pathway relating the level of glucose in the extracellular environment to the eventual activation of the stress-resistance genes and proteins is a fundamental mechanism of all eukaryotic cells. This statement is based not only on the fact that the comparable pathway in our own cells uses the same functional types of molecules, but also on the fact that most of the genes involved have significant sequence similarity to those of yeast. The only way to account for these two indisputable facts is to conclude that this pathway was selected for when all eukaryotes were single cells and has been maintained since then as a fundamental aspect of all animal cells. When I later discuss the insulin-like signaling pathway in worms, flies, and mammals, be aware that I am discussing the conserved pathway which first appeared in the yeast cell.

7.3.5 Genetic Stability During the foregoing description of yeast longevity pathways, I pointed out that certain of the prolongevity class III histone deacetylase enzymes

(SIR2 in particular) seem to have the normal function of transcriptionally silencing the rDNA genes and repressing chromosome breakage via homologous recombination in this highly redundant chromosomal region. Although there is reason to believe that the SIR2-dependent phenomena discussed above is a private or strain-specific aging mechanism, genetic stability is certainly an important longevity-assurance mechanism. At the chromosomal level, the prevention of abnormal chromosome breakage and reunion prevents the destruction or disruption of the genome at the chromosome level. At the genic level, it is obvious that the proper functioning of our cells depends on the genome being able to faithfully express a particular suite of genes and to faithfully repress some other suite. Stochastic alteration of the transcriptosome will likely lead to an increased loss of function. Neither genome disruption nor gene dysregulation are conducive to the maintenance of the cell’s function. An overall view of the organization and interactions of the genes and gene products that maintain the chromosomal or nuclear genomelevel integrity of the yeast cell are summarized in figure 7.8. These proteins regulate the passage of the cell through the different phases (G1, S, G2, and M) of the cell cycle. They do this by somehow sensing whether the cell has accumulated any DNA or mitotic spindle damage. The presence of such damage activates gene-based pathways that promote cell cycle delay or arrest, thereby giving the cell time to repair the damage (Kolodner et al. 2002). It seems reasonable to conclude that disruption of this quality control mechanism would lead to affected cells either dying or else behaving abnormally as a result of the damage. The evidence overwhelmingly supports this conclusion. For example, mutations that disrupt any one of the three S-phase checkpoints indicated in figure 7.8 significantly increase the rate of genome rearrangements, suggesting that these genes may play a major role in maintaining genome stability. Simultaneous inactivation of these three checkpoints leads to a massive (12,000-fold) increase in the rate of genome rearrangements. This finding suggests that numerous errors occur in normal S phase that could

7.3 Saccharomyces cerevisiae

Yeast Specific Gene & Metabolite Names

Functional Category of the Yeast Component

Ligand

Glucose

Gpr1

Ras2

Cyr1 (cAMP)

Sch9

251

PKA

Msn2, Msn4

MnSOD, Catalase, heat shock proteins, others.

LONGEVITY EXTENSION

Receptor

G-proteins

Second messengers

Serine-threonine kinases

Stress resistance transcription factors

Stress resistance proteins

Physiological Outcome of Altered Cell State

Figure 7.7 Combining the caloric restriction and stress response mechanisms in yeast allows the formation of this combined multistep pathway, which is one of the major longevity regulating pathways in yeast. The yeast genes specified by the various experiments discussed in the text are used to construct the pathway. The general functional property of each member of the pathway is listed in the right-hand column. This pathway seems to be generally conserved from yeast to mammals. (After Longo and Finch 2003.)

potentially result in deleterious effects but that the high level of redundancy within and between the several checkpoints detects and corrects almost all of them. This is the result one would expect from a system composed of numerous components, none of which is 100% perfect, but the combined effort of which affords the entire circuit an extraordinarily high effectiveness. The redundancy is an integral part of the system. In fact, theory suggests that one major difference between machine systems and living systems is that the former is characterized by a low redun-

dancy of high-reliability components, while the latter is distinguished by a high redundancy of lower reliability components. The replicative life span of the yeast cell depends on the proper functioning of this system. As one might expect, this cell-level system is conserved in other organisms. In humans, for example, some of these homologous genes are involved in inherited cancer susceptibility when mutated, and thus their normal function in humans, as in yeast, would seem to be to promote longevity by maintaining genomic stability.

252 Chapter 7 Genetic Determinants of Longevity in Animal Models

Cell Cycle Stages

Function

Effect

Proteins

G

S

G

DNA damage checkpoint

Intra-S Replication checkpoint 1 + (replication) checkpoints 2&3

DNA damage checkpoint

Mitotic checkpoints 1 & 2

G1 arrest in response to DNA damage

S phase arrest in response to replication blocks

Slowing of replication in response to DNA damage in the S phase

G2 arrest in response to DNA damage

Metaanaphase arrest in response to spindle damage

Block exit from mitosis

RAD17 RAD24 MEC3

RFC5 POL2 DRC1

RAD17 RAD24 MEC3 SGS1

RAD17 RAD24 MEC3

BUB1, 3 MAD1, 2, 3

BUB2 BRA1/BYR4

1

M

2

Figure 7.8 A brief summary of the yeast cell’s mechanisms to detect and repair DNA damage at different points in the cell cycle. There are six checkpoints distributed over the four cell cycle stages. Detection of DNA or related damage by the proteins listed triggers the effect that characterizes each checkpoint. Many of the genes and proteins have human homologs. (After Kolodnov 2002.)

High levels of oxidative stress are known to generate high levels of DNA damage in the nuclear genome of yeast, and different types of oxidative stressors induce different types of DNA damage. An interesting outcome of these investigations is the finding that preexisting unrepaired DNA damage interferes with the cell’s response to oxidative stress and so causes an increase in cellular levels of reactive oxygen species (Salmon et al. 2004). Given the conservation of DNA repair mechanisms, then we might well expect to find that levels of reactive oxygen species and DNA damage levels are linked in a positivefeedback damage-cascade in all living cells. The stability of the mitochondrial genome needs to be maintained if longevity is to be maintained, yet our knowledge of this organelle’s repair systems is very incomplete (Mandavilli et al. 2002). One of the genes coding for a base-excisionrepair enzyme, ntg2, is directed at both the nucleus and the mitochondria. Yet inactivation of this gene does not result in higher rates of mutagenesis, indicating the probable presence of some unknown back-up system. In contrast,

inactivation of the glycosylase OGG1 gene does lead to increased mutation rates, suggesting that some mitochondrial repair functions are nonredundant. This uneven redundancy may account for part of the observed lower efficiency of mitochondrial DNA repair relative to that of the nucleus. I point out in chapter 10 the key role that the mitochondria play in the cell, including the evidence that mitochondrial dysfunction plays an important role in cell aging. The higher incidence of DNA damage in the mitochondria relative to the nucleus may have very important consequences, and it likely constitutes a form of mitochondrial–nuclear interaction. We may view the many DNA repair genes as being specialized for different functions, each of which is essential for the maintenance of genetic and/or genome stability and hence of extended longevity. None of the mechanisms known to extend longevity will likely work well in the absence of a fully effective DNA repair system. In this sense, fully effective DNA repair systems are a necessary component of all longevity extension mechanisms.

7.4 Caenorhabditis elegans

7.3.6 Reproductive Effects Jaswinski (1993) proposed that the nature of the cell division in yeast might play a role in determining their longevity pattern. He noted that stem cells reproduce via an asymmetric cell division, but cells with a limited life span reproduce via symmetric or binary fission. This same dichotomy was observed in Volvox (see chapter 4). The relevance of this observation to yeast has to do with its effect on the replicative life span. The mother cell undergoes an asymmetric division to give rise to a small bud that eventually separates itself and becomes the smaller daughter cell. In addition to the obvious size difference (see figure 4.9), there is also an unequal distribution of cellular potency and components. The normal daughter cell does not, for example, inherit the same functional age as its mother, but rather starts anew as a young cell with a whole and unshortened life span. Clearly, an asymmetric distribution of physiological potential has occurred. What is the mechanism that underlies this asymmetry? The process is under genetic control, for there is a yeast mutation that causes the daughters to be born old and to die at the same time as their mothers (Lai et al. 2002). The ATP2 mutation affects a particular mitochondrial protein (F1, F0-ATPase) that alters various mitochondrial properties, including their asymmetric retention in the mother cell. Mutant mother cells pass on dysfunctional mitochondria to their daughters, which consequently cause slow growth and accelerated aging. As another illustration of how asymmetry and aging are intertwined, yeast mother cells undergo increasing levels of oxidative stress as they grow older (Laun et al. 2001), and thus they accumulate increasingly higher levels of oxidatively damaged (carbonylated) proteins as a function of replicative age (Aguilaniu et al. 2003). These oxidatively damaged proteins are not distributed to the daughters but are retained in the mother cell. The daughter cells thus inherit only undamaged proteins from their mother and are able to start their life span at a level of maximum function. This process depends on the proper functioning of the SIR2 gene because

253

mutations in that gene permitted daughter cells to inherit high loads of oxidatively damaged proteins. There is some reason to believe that this damage may arise as a result of a defective distribution of F-actin during cytokinesis, and so the actin skeleton may be required for proper asymmetric inheritance. The fact that SIR2 affects the F-actin distribution broadens the known effects of this gene and should be considered in any overall review of its effects. Other examples of asymmetric inheritance can easily be found (e.g., the effects of MnSOD overexpression [Harris et al. 2003] or prohibitin underexpression [Piper et al., 2002]). The essential point, however, is that the universal fact that offspring are always younger than their parents absolutely depends on the operation of these asymmetry-generating mechanims. Even though the yeast cell can live for extraordinarily long periods of time in a nonreplicative mode, once the cell begins to reproduce, the asymmetric inheritance built into its reproductive process to protect the young daughter cells begins the accelerated accumulation of damage products within the mother cell. Senescence is an integral part of reproduction in single-cell organisms such as yeast, in colonial forms such as Volvox (chapter 4), and in multicellular organisms as well.

7.4 Caenorhabditis elegans 7.4.1 Background Information Caenorhabditis elegans is a harmless, small, soildwelling, free-living nematode that was deliberately chosen by Brenner (1974) as an organism of genetic study. Nematodes are no stranger to research on aging; I have already discussed their usefulness in uncovering the role of altered proteins (see chapter 4). Klass (1977) did the first descriptive analysis of aging in this organism and began the tradition of using it to critically test various theories of aging. Following his lead, several investigators searched for and obtained genetic variants that shorten life span and others that extend life span in C. elegans.

254 Chapter 7 Genetic Determinants of Longevity in Animal Models life cycle of the normal animal contains a developmental period, an adult reproductive period, an adult postreproductive period, and a senescent period that culminates in death. Friedman and Johnson (1988a,b) found that the extended life span in the long-lived strain is due to a lengthened postreproductive period only, which occurs with no change in the length of the developmental and reproductive periods or any alteration in the organism’s fertility. The gene controlling this phenomenon was called the age-1 locus, and it was mapped to chromosome II (Friedman and Johnson 1988a,b). Age-1 later turned out to be homologous to the daf-23 gene, but the common usage in the field is to refer to it as the age-1 gene. Because the recessive mutant brought about extended longevity, it seemed reasonable to con-

7.4.2 Metabolic Control of Longevity: Insulinlike Signaling Pathway The genetic analysis of aging in C. elegans was begun in 1982 by Johnson and Wood with the generation and description of recombinant-inbred strains. As illustrated in figure 7.9, Johnson (1987) later showed that these recombinant-inbred strains show as much as a 70% increase in their mean and maximum life spans when compared to their wild-type progenitor strains. Furthermore, the Gompertz plot (figure 7.9D) suggests that the increase in life span is brought about by a decrease in the rate of aging of the long-lived strain relative to the controls (figure 7.9C). However, the increase in life span does not occur as a result of stretching out the entire life cycle like a rubber band. The

Progenitors

Recombinant inbred strains

Fraction surviving (%)

A

B

100

N2 Berg

80

100 80

60

60

40

40

20

20

0

0 0

10

20

30

40

50

60

0

C Log mortality rate

TJ119 TJ135 TJ143

10

20

30

40

50

60

D

0

0

-1

-1 N2 Berg

-2 -3

TJ119 TJ135 TJ143

-2 -3

0

10

20 30 40 Age (days)

50

60

0

10

20 30 40 Age (days)

50

60

Figure 7.9 Alteration of longevity in the nematode Caenorhabditis elegans. Survival data (A, C) and Gompertz mortality data (B, D) for the wild-type parent strains (A, B) and for three different recombinant inbred strains (C, D). The strain designated TJ143 is a long-lived strain in which the age-1 mutant was later characterized. The differences in its survival and mortality curves when compared with the other strains are obvious. (After T. E. Johnson 1987.)

7.4 Caenorhabditis elegans

clude that the normal allele repressed longevity and thus that the mutants are loss-of-function alleles. But the nature of the processes was not clear until two different labs uncovered the relationship between extended longevity and dauer larvae. The nematode responds to conditions of overcrowding or inappropriate temperature and/ or limited food by arresting its development early in the life cycle and molting into a nonfeeding, stress-resistant, developmentally arrested, and sexually immature “dauer” larva stage. The dauer larva can survive adverse conditions for months and then, when conditions improve, resume development and become a normal adult. The dauer larva is the nematode’s way of riding out bad times. The genetic analysis of dauer larva formation had been performed some time earlier (Riddle et al. 1981), and the animal’s ability to enter the dauer larva stage was known to be under the coordinate control of a complex genetic pathway. It was also known that null mutants for certain genes (daf-2, daf-23) force the animal to enter the dauer larva stage. When the activity levels of these genes are only partly lowered, the animals become not dauer larvae, but instead become adults with extended life spans (Kenyon et al. 1993; Larsen et al. 1995). Dauer larva genes are not just involved in the development of the organism, but are also active in the mature adult. The age-1 and the daf-2 genes function in the same pathway, are dependent on the same two downstream genes (daf-16 and daf18), and generally appear to act in the same manner (Dorman et al. 1995). This finding led to the identification of a subset of mutants that had significant effects on adult longevity and a larger set of daf mutants that affect larval development, allowing researchers to focus their efforts on just the relevant mutants. Molecular characterization of the age-1 gene revealed that it codes for a conserved enzyme called phosphatidylinositol-3hydroxyl kinase (PI3K; Morris et al. 1996). All these findings led to the realization that the daf-2 pathway in worms was probably homologous with the insulin/insulinlike growth factor signaling pathway (ISP) in mammals. It was immediately recognized that this homology, if true, meant that one could use the knowledge of

255

the mammalian ISP to tentatively assign functional roles to the daf-type aging genes and quickly attain insight into the mechanisms regulating the aging process. (It was later realized that the ISP was homologous to the yeast caloric restriction/stress response pathway [figure. 7.7], thus completing the evolutionary links.) A large body of work was conducted to confirm and extend this insight (for reviews, see Braeckman et al. 2001; Guarente and Kenyon 2000; Hekimi et al. 2001; T. E. Johnson 2002). It was soon established that the daf-2 gene coded for a protein homologous to the mammalian insulin/insulinlike growth factor-1 receptor (IGF-1R; Kimura et al. 1997). It was also established that the daf-16 gene is a member of the Forkhead class of transcription regulatory proteins and is a close homolog of several mammalian transcription factors (Lin et al. 1997; Ogg et al. 1997). These findings allowed the functional rebirth of the daf-2 pathway as an important intracellular signaling system (figure 7.10), whose strong similarities to the longevity regulation mechanism of yeast belie its evolutionary relationship. But this transference of the daf gene terminology to the ISP did not really tell us how the system worked in the C. elegans. What tissues were involved? What was the signal? What did it mean to the animal? Answers to these questions were rapidly attained. The involvement of the nervous system was shown by two groups using two different techniques (Apfeld and Kenyon 1998; Wolkow et al. 2000). Apfeld and Kenyon used genetic mosaics to identify neuroectoderm as the site of daf-2 activity that affected life span. Wolkow et al. used tissue-specific promoters to drive the expression of a wild-type daf-2 gene in different tissues of an otherwise mutant animal and found that neuronal expression was sufficient to restore the normal life span to the long-lived mutant. In each case, the main question being tested was whether the ISP mutants had to be expressed in only some cells or in all cells of the worm for it to express a mutant longevity. In both studies, the longevity of the animal matched that of the gene expressed in the neural tissue. This meant that the expression of longevity is not autonomous to each cell but rather is determined within the

256 Chapter 7 Genetic Determinants of Longevity in Animal Models

Yeast Specific Gene & Metabolite Names

Glucose

Gpr1

Ras2

Cyr1 (cAMP)

Sch9

PKA

Msn2, Msn4

MnSOD, Catalase, heat shock proteins, others.

LONGEVITY EXTENSION

Functional Category of the Yeast Component

Nematode Pathway

Ligand

Insulin-like molecule

Receptor

daf2 (insulin receptor)

G-proteins

Second messengers

Serine-threonine kinases

Stress resistance transcription factors

Stress resistance proteins

Physiological Outcome of Altered Cell State

age1/daf23

akt/PKB

daf16

MnSOD, CuZnSOD, heat shock proteins, others

LONGEVITY EXTENSION

Figure 7.10 The yeast caloric restriction/stress-response pathway compared with the insulinlike signaling pathway of the nematode. The functional similarities arise out of their common evolutionary origin. (After Longo and Finch 2003.)

neurons and then signaled to all other cells of the body in a systemic fashion. This findng does not conflict with the view that aging is fundamentally a cell-level phenomena. It simply means that the neuroendocrine system samples the environment and transmits that information thoughout the body, allowing each cell to respond in its own fashion. But sampling and transmission is not made within every neuron. Studies of sensory perception in the nematode revealed that two loss-offunction sensory mutants blocked the extension of longevity normally seen with ISP mutants such as

daf-2 (Apfeld and Kenyon 1998). This finding implied that the daf-2-containing interneuron depends on receiving a sensory signal originating in the worm’s external environment and transmits to it via a cholenergic synapse with a sensory neuron (Tissenbaum et al. 2000). Loss-of-function mutants have indicated that the interneuron likely depends on syntaxin (a protein involved in synaptic transmission) and CAPS (a protein involved in Ca2+-stimulated peptide secretion) for its ability to secrete a (still unknown) insulinlike ligand. This ligand represents the transduction of some environmental signal into an intercellular signal.

7.4 Caenorhabditis elegans

Why is this signal important to the animal? It is believed that the environmental signal being picked up by the sensory neuron has to do with the abundance of food (bacteria, in the case of the nematode). For a normal nematode living not in a petri dish in the laboratory but in the dirt of the real world, it is important for the developing worm to be able to sense the availability of food and the density of its fellow worms. If conditions are favorable (i.e., lots of food and few competitors), then the worm will continue developing into a mature reproductive adult. If adverse conditions are sensed by the developing worm, then it will form a dormant dauer larva and lie quiescent until the conditions are favorable. The animal needs a mechanism by which the necessary environmental signals can be detected and this information accurately relayed to all the tissues of the animal. The ISP, composed of the genes and cells described above, appears to be that mechanism. Its operation is diagrammatically presented in figure 7.11. This model for the regulation of life span in the nematode by sensory perception and the ISP works in the following manner. Environmental signals are perceived by sensory cilia and transduced to secretory neurons, possibly through cholinergic synapses. Upon stimulation by abundant food, appropriate temperature, and low pheromone, these cells release an insulinlike protein ligand. This ligand binds to the DAF-2 receptor kinase in the membrane of the somatic cell and activates a signaling cascade resulting in phosphorylation of DAF-16. Nonphosphorylated DAF-16 represses genes needed for reproductive growth and metabolism. This repression is relieved by phosphorylation, and the phosphorylated DAF-16 leaves the nucleus and enters the cytoplasm. It can no longer act as a repressor. The genes for reproductive growth and metabolism are accordingly expressed. This results in the development of a reproductive adult with a normal (i.e., short) life span. Note that phosphorylation of the transcription factor acts so as to shift the protein from one functional state to another, a process reminscent of the acetylation of histones and other proteins covered in the discussion of yeast.

257

Under the adverse environmental conditions of overcrowding (high pheromone) and low food, sensory input decreases, the interneuron is not stimulated to release the insulinlike ligand, and the DAF-2 insulinlike receptor is not stimulated. The weakening of the signal results in an increase in the proportion of nonphosphorylated DAF-16 present, which subsequently enters the nucleus and progressively represses the genes for reproductive growth and metabolism. Although it is not explicitly shown in the model, the nematode homologue of the yeast SIR2 gene, sir-2.1, extends longevity in C. elegans, possibly by silencing genes upstream of daf-16 that would otherwise suppress it, and thus suppressing the expression of extended longevity (Tissenbaum and Guarente 2001). This observation not only tells us that we do not yet have a complete inventory of every gene involved in regulating this pathway, but it also confirms that the SIR2 gene of yeast acts in an important public mechanism of aging. To make the story even more complicated, the DAF-16 protein has multiple transcriptional targets, and thus many different traits could be affected by its activation or repression. Not all of these traits need be directly related to aging. For example, the long-lived daf-2 and age-1 mutants are significantly more resistant to lethal bacterial pathogens (such as E. faecalis, S. aureus and P. aeruginosa) than are normal-lived controls (Garsin et al. 2003). The apparent involvement of the animals’ innate immunity suggests that the ISP regulates important processes not thought to be a direct part of the aging mechanism but that do play a role in determining longevity in particular environments. A more general examination of the C. elegans genome revealed at least 947 genes that contained at least one perfect DAF-16 binding site (TTGTTTAC) within their promoter region (Lee et al. 2003). Only conserved genes affecting public mechanisms were examined further. This particular subset of genes was identified by comparing the worm and fly genomes, selecting only those genes found in both organisms. This process revealed 17 orthologous DAF16 target genes, and three of these previously unknown genes were found to affect aging. The

258 Chapter 7 Genetic Determinants of Longevity in Animal Models

Low Pheromone, Abundant Food

High Pheromone, Low Food

Sensory Neuron

Sensory Neuron CNS

Interneuron

Interneuron

ILP released

no ILP released

DAF-2

DAF-2

DAF-16+P=> DAF-16P

Somatic Cell

DAF-16+P => DAF-16P

cytoplasm Absence of factor in nucleus:

Presence of factor in nucleus:

Growth Genes Active Somatic Maintenance Genes Inactive

Growth Genes Inactive Somatic Maintenance Genes Active

Figure 7.11 A model for the regulation of life span by sensory perception and insulinlike signaling. Environmental signals are perceived by sensory cilia and transduced to secretory neurons, possibly through cholinergic synapses. Upon stimulation, these cells release insulin-like protein (ILP) ligand. This ligand binds to DAF-2 receptor kinase and activates a signaling cascade resulting in phosphorylation of DAF-16. The nonphosphorylated DAF-16 represses genes needed for reproductive growth and metabolism. This repression is relieved by phosphorylation, and phosphorylated DAF-16 leaves the nucleus and enters the cytoplasm. Under adverse conditions, insulinlike signaling weakens, and the portion of nonphosphorylated DAF-16 increases, resulting in progressive repression of the genes for reproductive growth and activation of transcription of various stress-response genes, including specific catalase and superoxide dismutase variants. Under conditions of high dauer pheromone and low food, the signaling cascade is off. L2 juveniles arrest growth, molt to produce dauer larvae, and fully express the survival program. The genetic pathway for dauer larva formation is not shown here. (Redrawn from Braeckman et al. 2001.)

7.4 Caenorhabditis elegans

it now seems as if it must be viewed as another instance of the nuclear–mitochondrial communication, which we first discussed in the section on yeast aging. The clock mutants are a heterogeneous collection of at least four different genes that slow down by at least twofold the animal’s growth, behavior, development, or physiology and consequently allow the animal to live about 30% longer (Hekimi et al. 2001; Lakowski and Hekimi 1996). Molecular analysis of the clock genes shows that it is unlikely that they share a common mechanism of action. I include them in this one section, despite their diverse mechanisms, because they independently lead to a common longevity phenotype: The affected animals lead a slow but long life at all temperatures. Temperature independence is what distinguishes these mutant effects from the effect of temperature on cold-blooded organisms such as nematodes (table 7.1). More recent data have shown that the extended longevity arises out of antagonistic directions in age-specific mortality rates: The mortality is higher in younger animals relative to controls but is significantly lower in older animals. The clock-1 (clk-1) gene codes for a conserved mitochondrial enzyme that is required for the synthesis of ubiquinone (Q), a necessary cofactor in the electron transport chain and in several other processes (Hekimi 2000; Stenmark et al. 2001). Cofactor exists in several isoforms that differ in the

main point is that the de repression of the DAF16 protein allows the activation of a large number of genes encoding gene products involved in resistance to oxidative and other forms of stress as well as the regulation of other processes apparently important for the expression of extended longevity. These processes take place in different cells of the body. How then is the DAF-16 signal transmitted from the neural cells to all cells of the body? How do such cells respond to that signal, or to its absence? These questions are discussed below, but first I review another class of metabolic control genes in C. elegans.

7.4.3 Metabolic Control of Longevity: Nuclear–Mitochondrial Interaction A classic view of aging, first proposed by Buffon in 1749, is that longevity is inversely proportional to metabolic rate. This theory has been criticized (see chapter 10 for details) and was more accurately restated as follows: “The rate of aging is directly related to the rate of unrepaired molecular damage inflicted by the byproducts of oxygen metabolism, and is inversely related to the efficiency of antioxidant and reparative processes” (Sohal 1986, p. 41). This interpretation was once thought to explain the extended longevity of the clock family of mutants, and it is not wrong, but

Table 7.1 Effect of Temperature on the Mean Life Span of Normal and Mutant Nematodes Carrying Different or Multiple Clock Mutations Mean Life Span (days) at the Indicated Temperature Genotype Wild type clk-1 clk-2 clk-3 gro-1 clk-1/clk-2 clk-3/clk-1 clk-3/clk-2 clk-3/gro-1

15°C 22.0 29.3 24.6 25.7 26.0 34.5 36.7 33.8 24.4

± ± ± ± ± ± ± ± ±

0.3 0.5 0.8 0.7 0.7 0.9 1.4 0.9 0.6

259

18°C 14.9 18.4 18.7 20.4 19.2 28.2 22.7 22.3 15.9

± ± ± ± ± ± ± ± ±

0.3 0.4 0.5 0.4 0.6 0.9 0.6 0.8 0.8

20°C 16.1 17.3 18.0 19.9 19.7 23.1 17.8 20.6 14.6

± ± ± ± ± ± ± ± ±

0.2 0.4 0.5 0.6 0.6 0.7 0.5 0.6 0.4

25°C 9.2 ± 0.3 11.6 ± 0.5 11.7 ± 0.7 13.0 ± 0.8 15.6 ± 0.5 No data 17.4 ± 1.4 12.7 ± 0.5 No data

Source: data taken from table 1 of Lakowski and Hekimi (1996). Note: reading horizontally illustrates the effects of temperature on a given genotype; reading vertically illustrates the effects of different genes at the same temperature.

260 Chapter 7 Genetic Determinants of Longevity in Animal Models length of a lipid component of the molecule and are denoted as Qx. The Q9 isoform is the normal form endogenous to the worm, and the clk-1 mutation interferes with the final step in its synthesis (Miyadera et al. 2001). The animal survives because it can utilize the Q8 isoform synthesized by the bacteria it eats. However, if the clk-1 mutant is fed on a bacterial strain that cannot make Q8, then the nematode dies during development (Jonassen et al. 2001). Ubiquinone is an essential component. What is interesting is that if either wild-type C. elegans or daf-2 type mutants (both of which can make Q9) are fed on this same bacterial strain that cannot make Q8, then each of these worms shows a life span extension of 50– 60% compared to their observed life span on a normal-bacteria diet (Larsen and Clarke 2002). Daf-12 mutants do not respond to a Q-less diet. How does Q affect aging? An important clue is the finding that mitochondria of animals whose genes responsible for Q9 synthesis were silenced (and thus produced a clk-1 type phenotype with a decreased amount of total Q) produced significantly (30–50%) less reactive oxygen species and thus less oxidative stress (Ascencio et al. 2003). This suggests that an abundance of total Q8 plus Q9 may accelerate the rate of electron transport and thus increase the release of reactive oxygen species from the mitochondria. Q is also a cofactor in the activity of uncoupling proteins that are involved in proton leakage during aging (see chapter 11). Reducing the Q content may reduce the levels of oxidative stress and thus the aging rate. Either method of reducing oxidative stress provides an explanation for the extraordinarily long life spans of the daf-2:clk-1 double mutants, for they each reduce oxidative stress via a different and independent pathway, and so the double mutants can express a synergistic interaction of the two pathways. Alternatively (or in addition), perhaps the balance of Q8 plus Q9 isoforms can affect the signaling between the mitochondria and the nucleus that normally results in adjusting the level of gene transcription so that it is in balance with the metabolic state of the mitochondria. Either or both of these hypotheses offers a plausible explanation of the clk-1 effect on longevity. Yet they raise an interesting question: How can

the animal survive if it limits its mitochondrial energy metabolism? An interesting hypothesis put forth by Rea and Johnson (2003) is that energy generation in C. elegans occurs by a differential flux through two coexisting mitochondrial metabolic pathways: the conserved aerobic respiration pathway and the fermentative malate dismutation pathway, which is likely not conserved in higher organisms. In the latter process, fumarate is terminally reduced at complex II to succinate, and this process generates less reactive oxygen species than does aerobic respiration. This is a unifying explanation for the genesis of this unique longevity phenotype. Despite the fact that fermentive malate dismutation is not found in mammals, the gene is conserved, and orthologues of clk-1 are found in mice and humans (Asaumi et al. 1999). The human gene is predominantly transcribed in heart and skeletal muscle, with much lower levels in other tissues. Its mode of action is not yet clear, but it does not seem to involve a limitation of mitochondrial energy metabolism (Braeckman et al. 2002; Hekimi et al. 2001). The clk-2 gene encodes a cytoplasmic protein involved in regulating the length of telomeres (Benard et al. 2002). Its mode of action may involve altering telomere length directly or indirectly altering patterns of gene expression, but definitive data to support this hypothesis is not yet available (Hekimi et al. 2001). A third member of the clock family, gro-1, encodes a highly conserved enzyme (isopentylpyrophosphate:tRNA transferase) that modifies certain tRNAs. In both yeasts and nematodes, the gene is localized both in the cytoplasm and the mitochondria. Expression of the wild-type mitochondrial gene is sufficient to rescue the mutant phenotype. Presumably, other members of this gene family code for equally diverse gene products and require equally different modes of action. This assumption is supported by the fact that worms carrying different clk mutant genes show an additive effect for life span (table 7.1), indicating multiple mechanisms that can independently and additively slow down the rate of living and extend the longevity relative to the wild type at any temperature. Why, then, should diverse genes all yield a slow rate of living and an extended longevity? It

7.4 Caenorhabditis elegans

has been suggested by Branicky et al. (2000) that the clk mutants affect, in different ways, the interactive coordination between the nucleus and mitochondria. Switching to an alternative energygenerating pathway, as was discussed above in the case of the clk-1 gene, would be one way of altering this coordination. Loss of this coordination, which can apparently happen in a number of different ways, provides incorrect information to the nucleus about the state of the mitochondria, and vice-versa. This results in the reorganization of the nuclear genetic interaction network to a state consistent with a slow rate of living. The extended longevity characteristic of the clk mutants probably results from the fact that a slow rate of living results in a lower level of energy production in the mitochondria and an accompanying lower production of reactive oxygen species (see chapter 10), the molecules responsible for oxidative damage (see chapter 9). The daf-clk double mutants live extraordinarily long (table 7.1) because they both produce little oxidative damage and repair any damage that does occur efficiently.

7.4.4 Stress Resistance and Extended Longevity The nonphosphorylated DAF-16 generated by inactivation of the ISP does more than just repress the genes associated with reproductive growth and metabolism (see figure 7.11). It was independently shown by two groups that the long-lived age-1 mutant is highly resistant to oxidative stress and expresses elevated levels of two enzymes known to protect cells against oxidative damage (copper-zinc superoxide dismutase [CuZnSOD] and catalase; Larsen 1993; Vanfleteren 1993). The age-1/daf23 mutant also overexpressed the mitochondrial form of superoxide dismutase (MnSOD; Honda and Honda 2001). The age-1/ daf23 mutant lived significantly longer than controls when raised under either normal conditions or under conditions of high oxidative stress; however, the age-1–daf16 double mutant had normal levels of SODs and lived a normal life span (Honda and Honda 2001). This experiment demonstrates that the mutational downregulation of

261

the ISP has no effect if the daf16 transcription factor is mutationally altered so that it cannot activate or repress its target genes. Another important antioxidant enzyme is glutathione-S-transferase (GST), a multifunctional enzyme family used in detoxification mechanisms. Organisms usually express multiple GSTs, each with a specialized function. C. elegans, for example, contains more than 50 putative GSTs, only one of which (GST-p24) is thought to be involved in modulating oxidative stress (Leiers et al. 2003). Overexpression of GST-p24 in transgenic animals raised under normal aerobic conditions did not impart a longer life span on the mutant relative to the wild-type controls. However, when the animals were exposed to oxidative stress, the mutant animal had a higher survival compared to the wildtype controls. Conversely, when using RNA interference to silence the GST-p24 transgene, mutant animals then exposed to oxidative stress had a significantly lower survival rate than did the untreated mutants (Leiers et al., 2003). Thus, there is a strong correlation between the animals’ level of GST-p24 gene expression and their ability to survive in a stressful environment. Taken together, these data demonstrate that the high levels of several different antioxidative stress enzymes have a functional effect on the organism. Man-made synthetic molecules can mimic the effects of these antioxidative stress enzymes. Melov et al. (2000) showed that two synthetic molecules (EUK8, EUK134), which exhibit both CuZnSOD-like and catalaselike catalytic activities, could, when fed daily to normal C. elegans adults, increase longevity by an average of 44%, thus giving them a life span intermediate between untreated wild-type animals and the long-lived age-1–daf23 mutant. Moreover, feeding one of these mimetics to the oxygen-sensitive, shortlived mev-1 mutant increased its life span by 67%, effectively restoring it to a normal longevity by ameliorating an endogenous oxidative stress. However, the effect of the EUK8 mimetic was not replicated by a different laboratory, suggesting that minor environmental differences might alter its effectiveness (Keaney and Gems 2003). Nonetheless, this category of synthetic antioxidant mimetics promises to be a most useful possible

262 Chapter 7 Genetic Determinants of Longevity in Animal Models intervention in various disease states as well as in aging (Golden et al. 2002). A transient exposure of young age-1–daf23 mutants to high oxygen significantly protected the animals against a subsequent exposure to severe oxidative stress conditions (Honda and Honda 2001; Yanase et al. 2002). The importance of this experiment is that it demonstrates the adaptive nature of these stress-resistance enzymes; the animal’s exposure history determines its subsequent response to a severe stress. Transient, low-level stress protects the animal against the lethal effects of more severe secondary stress. This hormesis effect, which I discussed in chapter 6, is an important feature of the animal’s stress response. Oxidative damage is the major mechanism underlying age-related loss of function (see chapter 9), and the reactive molecules that cause such damage mostly originate in the mitochondria and then spread to the cytoplasm, where they cause damage to various cellular components. Thus, overexpression of these antioxidant genes and enzymes would be expected to reduce the level of oxidative damage in these animals and increase their longevity, if this oxidative damage really does play an essential role in the aging process. Taken together, these findings strongly support the oxidative damage theory of aging, a topic to which I will often return. More important, these findings demonstrate that extended longevity in the nematode is intimately associated with an increased ability to withstand a number of different environmental stresses, of which oxidative stress is one of the more important. Another common cytoprotective response to a wide variety of stresses is the induction and increased production of a family of proteins known as the heat shock proteins (HSPs). The term “heat shock” reflects their initial discovery in Drosophila as a protein family induced in response to hyperthermia. HSPs are now known to be ubiquitous in living organisms, being found in all forms from bacteria to vertebrates to plants. Increased awareness of their induction by a multitude of stressors other than heat shock led to their being more generally referred to as cellular stress proteins. They are also called, molecular chaperones, in recogni-

tion of their functional roles. Their normal roles in the cell as well as their role in the genesis of nonheritable stress resistance is discussed in chapter 10 (see figure 10.8), but in this chapter I discuss the role of HSPs in extending the longevity of each of our model organisms. A great deal of evidence has been gathered regarding the effect of HSPs on longevity in C. elegans. This work was encouraged by the observation that the dauer constitutive mutants previously identified in this species (Riddle et al. 1981) were not only long-lived (Kenyon et al. 1993), but they were also significantly resistant to heat stress (Lithgow et al. 1995). These observations led to the hypothesis that longevity is the result of the overexpression of stress-resistance genes and the consequently increased ability to cope with macromolecular damage (T. E. Johnson et al. 1996). This hypothesis has been supported by empirical data. HSP16 is a classical stress-response protein in the nematode. It is not expressed in unstressed younger worms but is found in older (>16 days) adults. Overexpression of HSP16, induced by either mild heat stress or by inserting extra copies of the gene into experimental animals, leads to an increase in both thermotolerance and longevity (Walker and Lithgow 2003). What is even more interesting is that the maximal expression of the hsp16 gene is critically dependent on the functioning of the DAF16 transcription factor regulated by the ISP. It has been shown that the expression of the hsp70 and hsp90 genes are also dependent on ISP-DAF16 function (Munoz 2003), as are other types of stress-resistance genes. For example, the old-1 gene is also dependent on ISP function and appears to activate both the DNA repair system as well as promoting thermotolerance. Taken together, these findings suggest the existence of multiple regulatory pathways modulating the expression of the various heat shock genes as well as that of genes conveying resistance to other forms of stress (see figure 9.6). The activity of a stress-resistance gene network can be regulated by environmentally modulated genesignaling pathways (nutrition, temperature, etc.). Evidence to support and refine this suggestion is the fact that exposure of C. elegans to low doses of oxygen stress, heat stress, or oxidative stress

7.4 Caenorhabditis elegans

led not only to the development of resistance to the effects of subsequent high doses of the initial stressor, but it also led to the development of cross-resistance to subsequent high doses of a different stressor (Cypser and Johnson 2002). The treated animals were not only stress resistant but were also long lived when allowed to age normally, thus supporting the idea that stress resistance and longevity are closely linked phenomena. It turns out that the ISP and the heat shock transcription factor HSF1 interact with one another so that mutations in one pathway have measurable effects on the amount of extended longevity brought about by the other pathway (Morley and Morimoto 2004). These data also support a direct role for HSF1 as a positive regulator of longevity. These data are discussed in chapter 9 as evidence for the proposition that the transition from the health span to the senescent span is dependent on the cell’s level of stressresistance molecules and their capacity to deal with accumulated, unrepaired cellular damage.

7.4.5 Genomic Stability There has been little work specifically designed to elucidate the role of genomic stability in modulating the longevity of C. elegans. However, experiments done for other purposes show that both the nuclear and mitochondrial genomes are subjected to the usual sources of DNA damage, that there is a variety of DNA repair genes that normally repair such damage, and that mutational inactivation of some of these DNA repair genes leads to abnormal effects and shortened life spans in adults (for reviews, see Boulton et al. 2002; Mandavilli et al. 2002). It seems reasonable to conclude that in the nematode, as in other organisms, both normal and extended life spans depend on the existence of a conserved system to support genomic stability and that such stability is an essential prerequisite for the ability of other gene pathways to bring about the expression of extended longevity. Loss of genetic stability can come about via either genetic mutation or genome rearrangement. The discussion of genomic stability in yeast (see figure 7.8) presented evi-

263

dence that the same DNA repair and recombination mechanisms deal with mutation and gene management. There is reason to believe that the same is true in the nematode and other organisms, and this belief has been supported by recent data. For example, the p53 tumor-suppressor gene plays a critical role in maintaining the stability of the mammalian genome by regulating cell-cycle progression and apoptosis in response to DNA damage. The p53 gene also activated by various stress signals. C. elegans has a p53 homolog, termed cep-1; the normal functioning of which is required for the normal inhibition of DNA damage-induced cell death or for the normal meiotic chromosome segration in the germline (Derry et al. 2001). Interesting in the current context is that mutant cep-1 animals are hypersensitive to the lethal effects of hypoxia and have a shortened life span when subjected to starvation. The nematode also contains homologues of the human XPA gene (Park et al. 2002) and the human Cockayne syndrome B protein (Lee et al. 2002), two other DNA repair genes that reduce longevity in humans. The degree of evolutionary relationships of the various DNA repair genes was demonstrated by a recent genomics approach. The investigators screened the entire C. elegans genome for genes that, when inactivated, increased the level of spontaneous mutation (Pothof et al. 2003). They detected 12 genes homologous to known DNA repair and replication genes, 9 genes homologous to known chromatin organization and remodeling genes, 11 genes homologous to known cell cycle and checkpoint genes, and 29 novel genes. Other work shows that these genes interact with each other and constitute a dynamic gene interaction network (Boulton et al. 2002). Genes with known DNA repair functions closely interact with genes of unknown function (Boulton et al. 2002). This observation suggests that the genes of unknown function may also be involved in that same function of DNA repair. This hypothesis needs to be confirmed, but it nevertheless gives us an appreciation of the fact that our genes do not work by themselves but rather as members of a dynamic gene interaction network. I return to this theme in chapter 14. There is an elaborate mechanism in the nematode for the maintenance of genomic stability. It

264 Chapter 7 Genetic Determinants of Longevity in Animal Models is reasonable to assume that this mechanism is very similar to that found in the yeast cell (figure 7.8). All in all, the available data certainly support the conservation of mechanisms responsible for the maintenance of genetic and genomic stability—and thus of longevity—in C. elegans.

7.4.6 Reproductive Effects The evolutionary theory of aging (chapter 4) posits a tension between reproduction and longevity based on competition for a finite amount of bodily resources. Our knowledge is now at the point where we can verify this assumption and begin sketching out the molecular and physiological mechanisms involved in this antagonism. Kenyon and colleagues performed a series of experiments in which the different compo-

nents of the embryonic gonad were removed by either laser ablation (Hsin and Kenyon 1999) or genetic means (Arantes-Oliveira et al. 2002), and the life span of the resulting adult was measured. As shown in figure 7.12a, the newly hatched C. elegans larva is not sexually mature but rather contains four precursor cells, termed Z1 to Z4, which give rise to the entire adult gonad. The gonad consists of the somatic gonad and the germ cells. The two outer cells, Z1 and Z4, give rise to the somatic gonad; the two inner cells, Z2 and Z3, give rise to the germline. The hundreds of reproductive cells that eventually compose the germline are derived from these two cells via a small number of germline stem cells located in the distal tips of the gonad. As development proceeds, germ cells located farthest from the distal tips of the gonad enter meiosis and differentiate initially into sperm and

1.0 Z1 Z2 Z3 Z4

Fraction alive

0.8

0.6

0.4

0.2

0.0 0

10

20

30

40

50

60

Days Figure 7.12 The nematode gonad at hatching consists of only four cells. Z1 and Z4 give rise to the somatic gonad, and Z2 and Z3 give rise to the adult germ line. These four cells are represented by the four circles in the legend, as listed. An X indicates which of these precursor cells were ablated by laser in the first instar larva. The somatic gonad is needed for germline development. The intact control lived 19.4 ± 0.29 days. Ablation of Z2 and Z3 (the germline precursors) led to extended longevity of 31.8 ± 0.97 days. The other treatments yielded essentially normal longevities (17.8 to 22.5 days). (After Hsin and Kenyon 1999.)

7.4 Caenorhabditis elegans

later into eggs. As a result, a pool of proliferating stem cells is maintained well into adulthood. The question is which, if any, of these reproductive cells have an effect on longevity of the adult. Using a laser to specifically remove either all four Z cells or just the Z1 and Z4 cells gave rise to adults that had life spans essentially identical to that of the intact wild-type control (figure 7.12). The fact that these sterile animals with no gonad lived no longer than the intact controls indicates that reproduction as such does not alter life span. However, when just the Z2 and Z3 cells were ablated, then the resulting sterile animals possessing only the adult somatic gonad tissue lived significantly longer (~60%) than did the others (figure 7.12). Such germline-ablated animals are significantly more resistant to heat and oxidative stress, once again indicating the close connection between stress resistance and extended longevity (Arantes-Oliviera et al. 2002). Note that the somatic gonad tissue must be present if germline ablation is to produce a long-lived animal (since ablation of all four cells yields a normal-lived animal). Essentially identical results are obtained if one uses various mutants to cause the germ cells to either die or to fail to proliferate. One implication of these results is that the germline cells of normal animals appear to generate a signal that accelerates aging, whereas the somatic gonads might generate a different signal that slows down the aging process. The fact that ablating the germline precursor cells yields long-lived animals in daf-2 mutants but not in daf-16 mutants strongly suggests that the DAF16 transcription factor is required if the longevity-enhancing signal from the somatic gonads is to be effective. This means that the reproductive effect on longevity is closely intertwined with the operation of the ISP itself. One possible process that would tie reproduction and longevity together is a postulated hormonal pathway by which the ISP signal for reproduction can be sent to all the cells of the body, including the reproductive germline and somatic gonad cells; this process triggers a feedback response that affects longevity (Gerisch et al. 2001). One working interpretation of the

265

data is that stimulation of the ISP in the neuroendocrine cells triggers the secretion of some unknown pro-reproductive hormone that alters the animal’s metabolism and physiology to favor reproduction-related activities and suppresses somatic maintenance activities (Tatar 2002). The germline stem cell activity then provides a positive feedback signal to the neuroendocrine system, which brings about the suppression of DAF-16 (i.e., maintaining DAF-16 in its phosphorylated state and out of the nucleus), thus maintaining the ISP in a pro-reproductive mode. In addition to the DAF-16 effects listed in figure 7.13, the active form of DAF-16 extends longevity in part because it inhibits daf-9 activity, which in turn activates daf-12 activity (Gerisch and Antebi 2004; Jia et al. 2002; Mak and Ruvkun 2004). DAF-9 codes for a cytochrome P450 involved in steroid hormone biosynthesis, and DAF-12 codes for a nuclear hormone receptor. Both these gene products might affect the activity of the postulated hormone depicted in figure 7.13. A model of how this feedback system might possibly be integrated with the ISP as described above is shown in figure 7.13. Environmental cues induce production of the insulinlike ligand by the sensory neurons (see figure 7.11). Neuroendocrine cells in the central nervous system respond by activating daf-2 and repressing daf-16. This promotes the reproductive mode, including the failure of daf16 to inhibit daf-9 in the two hypodermal sensory neurons in which this cytochrome P450 is active and thus its failure to repress the production of some pro-reproductive secondary hormone (which is still unknown in worms but whose homologue in flies is the well-known juvenile hormone). This secondary hormone stimulates the germline cells to produce a signal that represses both the DAF16 protein and the synthesis of the unknown DAF-9–dependent life-extending steroid in the somatic gonad cells. The repression of this putative hormone allows the somatic cells to remain in the pro-reproductive phase associated with phosphorylated DAF-16. The outcome of this system is that it connects the onset of reproduction with the repression of

266 Chapter 7 Genetic Determinants of Longevity in Animal Models

External Cues

CNS

Insulin/IFG-1 ligand

Longevity

DAF-2

DAF-16

Somatic Cells

DAF-9

DAF-12

Unknown Hormone

Germline Cells

Gonad

Figure 7.13 An integrated model for the endocrine regulation of reproduction and aging in the worm and fly. External cues (diet, crowding, pheromones, etc) stimulate production of insulinlike peptides (ILP) that induce the insulin/insulinlike growth factor 1 (IGF-1) signal transduction pathway of the CNS. This induction indirectly suppresses a transcription factor (DAF-16) and indirectly permits synthesis or release of a downstream proreproductive hormone (unknown in C. elegans but assumed to exist based on homologies to the fly). This hormone suppresses life maintenance systems and helps up regulate reproductive activity. The somatic cells of the gonad inhibit the IGF-1 receptor (DAF-2). The germline cells are stimulated by the presumed hormone, while they simultaneously inhibit the DAF-16 transcription factor. The repression of this factor results in peripheral tissues exhibiting a gene expression pattern that stimulates growth and reproduction at the expense of somatic maintenance activities and in this way results in active aging. (Redrawn from Tatar et al. 2003.)

somatic maintenance. This then provides a mechanistically plausible process by which senescence should begin just when the animal becomes sexually mature (see figure 2.15).

7.4.7 Patterns of Senescence When organisms undergo a senescent-driven loss of function, they usually do not degrade in a random manner but rather in some sort of pattern. Only recently has a detailed analysis of senescent patterns become available for the nematode (Herndon et al. 2002) The animal does not senesce in a random nor a uniform manner, but in a characteristic pattern. Using various imaging tools, Herndon et al. determined that the nervous system is well maintained in both structure and function even in advanced old age. In contrast, there is a gradual and progressive deterioration

of the muscles beginning in mid-life. There is also a late-life dysregulation of certain types of macromolecular biosynthesis and turnover, including yolk protein. These tissue- and stage-specific decrements may serve as the basis for biomarkers of aging (see chapter 2). Longevity-enhancing mutations, such as the age-1 mutant, delay the onset of the muscle-specific senescence pattern but not that of yolk protein dysregulation. Thus, such mutants appear to extend longevity via their effects on specific tissues (e.g., muscles) important to bodily function but not via a universal delaying effect on all systems. Animals die when one vital system falls below some functional threshold. The observations of Herndon et al. (2002) suggest the possibility of someday putting together combinations of different longevityextending mutants that each affect different parts of the body and thus exhibit a synergistic effect on longevity.

7.5 Drosophila melanogaster

7.5 Drosophila melanogaster 7.5.1 Background Information Our model organisms display a basic conservation of aging mechanisms that should now be apparent, and many of the same mechanisms discussed in yeast and nematode will reappear in the discussion of the fruit fly, but in more complexity. The fly has perhaps a million cells, which is about 1000-fold more cells than the nematode, and probably about 1000-fold fewer cells than a mouse or human. Systemic controls not needed for a simple system such as the worm make their appearance in the fly. Its intermediate level of complexity is one of its unappreciated virtues and allows Drosophila to serve as an important stepping stone between a nematode with only 959 cells and a mouse with some indeterminate but very large number of cells. Most of the existing Drosophila mutants had their life spans shortened as a result of deleterious processes unrelated to aging (for a review of the early literature, see Arking and Dudas 1989). Existing mutants provided no useful clues to the genetic control of aging and longevity. Instead, long-lived strains and mutants had to be created before genetic analysis of the aging process could begin. Beginning in 1979, several different groups of investigators decided to use artificial selection to create extended-longevity strains and to develop these as models to examine the mechanisms underlying aging (Arking 1987a; Luckinbill et al. 1984; Rose 1984). This approach marked the beginning of modern biogerontology in the sense that these experiments marked the first time that longevity was successfully and significantly extended in any model organism.The end result of the selection process is summarized in the survival curves of figure 7.14, which show that selection effectively increased the mean and maximum life span of the R strain by about 50% to yield a new long-lived L strain. The fact that the use of reverse selection allowed the conversion of the long-lived strain into a normal-lived strain (ReL) shows that the aging mechanisms are plastic enough so that they can be manipulated in both directions.

267

This increase in the life span did not occur as a result of stretching out the entire life cycle like a rubber band. The life cycle of the fruit fly can be viewed as consisting of a developmental period (10 days), an immature adult phase (<1 day), a mature adult reproductive phase (30 days), and a postreproductive senescent phase (10–30 days) that culminates in death (figure 7.15). The use of biomarkers (see chapter 3) allowed us to determine that the flies were living longer because they stayed healthy and delayed the onset of senescence. Whatever the long-lived strains were doing, they had to start doing it sometime early in adult life. This turned out to be a general rule, obvious now but not then (see Srgo and Partridge 1999). These data also led to the identification of the health span and senescent span as two independently manipulable portions of the adult life span, a topic I return to in chapters 9 and 14. Subsequent studies done on these and other strains and mutants have allowed us to deepen our collective understanding of the conserved mechanisms underlying longevity and aging in the fly (see Helfand and Rogina 2003a,b). I discuss this new information using the same format as for the review of the genetic mechanisms underlying aging in yeasts and nematodes.

7.5.2 Metabolic Control of Longevity via Caloric Restriction Although caloric restriction (CR) was first demonstrated in rodents in 1934, and although there have been a number of investigations of nutritional effects on longevity of Drosphila, it was not until 1996 that the first replicable experiment was published showing that CR was effective in altering the fly’s life span (Chapman and Partridge 1996). (This long time lag probably occurred for technical reasons having to do with how one regulates the food intake of a fly, the details of which are reviewed by Gerhard [2001]). Figure 7.16A shows that raising the adults on food containing only 33% of the nutrients (i.e., calories) of standard food allowed about a 34-day increase in the median life span and a 31-day increase in the maximum life span (Pletcher et al. 2002). These

268 Chapter 7 Genetic Determinants of Longevity in Animal Models

1

Fraction Surviving

0.8

L

0.6

ReL

0.4 R 0.2

0 0

15

30

43

61

77

96

Age in days

Figure 7.14 Longevity in Drosophila is plastic and can be reversibly and stably modified. The graph presents the survival curves for the nonselected R strain animals which served as the progenitors of the selected long-lived L strain animals. Forward selection (R to L) was done as described in the text. Several years after their creation, some of the L strains were reverse selected for a shortened longevity as described. The main genetic alteration involved the antioxidant genes, but their alteration depended on coordinate changes in mitochondrial and metabolic functions. Selection in either direction is repeatable as indicated by the similar response of the two independent replicate lines within each set. (After Arking et al. 2000.)

survival curves imply that there is a delay in the onset of senescence, an assumption supported by the patterns of the age-specific mortality rates depicted in figure 7.16B, for their age-specific mortality does not begin to increase from the minimal values observed in the young until about 15 days in the control animals and about 45 days in the CR animals. This increase in the age-specific mortality can be taken as indicating the age of onset of senescence; the 30-day delay in the age of onset of senescence in the CR animal essentially accounts for all the extra longevity noted in the survival curves. From a demographic point of view, the population effectively does not age during this interval, and so the extra longevity is incorporated into the young and healthy portion of the life span. I return to this theme of delayed senescence as a general process in chapter 14. At each of the time points indicated in figure 7.16A, animals were taken from both the control

and CR groups, and their gene expression patterns were assayed using DNA microarray analysis (Pletcher et al. 2002). This procedure, when applied to the indicated fly populations, gives us a detailed view of changes in gene expression as a function of both age and diet. An overall summary of the results is presented in figure 7.17. Of the 14,028 gene probes assayed (essentially the entire fly genome), only 1264 probes (representing 1203 genes; <1%) showed highly significant age-related changes in either or both of the control and CR cohorts. Most genes do not change their expression very much (see the section on genetic stability below), and so our interest is focused on those few genes that do change. The Venn diagram shows the numbers of gene probes that change with age in the control and/or CR cohort, and the associated boxes show the types or classes of genes that disproportionately express the indicated change (figure 7.17). Thus, the 754 gene

7.5 Drosophila melanogaster

269

Pho Geo 100 Pho Geo

Feo

Feo

80

Percent surviving

Pro

Pro

60

L R

40

20

Met Met

0

0

20

40

60

80

100

Age (days) Figure 7.15 The age-dependent loss of behavioral and physiological traits in adult females of the long-lived (L) and normal-lived (R) lines of Drosophila. The traits observed are positive phototaxis (Pho), negative geotaxis (Geo), fecundity (Fec), amino acid incorporation in vivo (Pro), and mean daily metabolic rate (Met). The arrows indicate for each trait the point in time at which half of the animals can no longer meet defined quantitative criteria for each test. Both strains show the functional loss of each trait in the same sequence and at approximately the same portion of the life cycle. The aging process in the L strain is extended due to the specific lengthening of the presenescent portion of the life span (i.e., the “health span”), but the senescent span is entirely normal by comparison with the normal-lived strain. (After Arking and Wells 1990.)

probes that increase with age in both sets comprise mostly genes involved with innate immunity and detoxification, enzyme inhibitors, and all sorts of genes involved in the response to various stresses. As a general rule, these gene probes increase in both cohorts but do so more slowly in the CR set than in the control set, an observation consistent with the data of figure 7.16B. The data above suggest that aging animals— regardless of their diet or chronological age—are under increasing stress from pathogens. It appears as if the proximate cause of death in old flies is

infection, a supposition confirmed by Tower et al. (2004). Various other endogenous and exogenous stressors are also involved. One possible source of such stress is indicated in the 458 gene probes down-regulated with age, which suggests loss of mitochondrial function. Data presented in chapter 10 show that oxidative stress increases as mitochondrial efficiency decreases; the aging animals are probably under increasing oxidative stress. The loss of energy production associated with decreased mitochondrial function is a major factor in the loss of function characteristic of aging. Also

1.0 0.8

B In Mortality Rate

A Survivorship

270 Chapter 7 Genetic Determinants of Longevity in Animal Models

0.6 0.4 Calorically Restricted Control

0.2 0.0 0

20

0 −2 −4 −6 −8

40

20

0

60

40

60

Age (days)

Age (days)

Figure 7.16 Survival curves (A) and age-specific mortality curves (B) for flies maintained on either control (enriched) medium or caloric restriction (CR) medium, which contains 66% less yeast and glucose. The mean (SE) life span of the control and CR flies was 25.4 (0.46) and 46.2 (0.24) days, respectively. The maximum life span in CR conditions was 78 days, and the last 50 flies alive on enriched media were assayed on day 47. (After Pletcher et al. 2002.)

A. Immunity & Detoification Enzyme inhibitors Stress response

Control (1041)

Up-regulated with age (754)

D. DNA repair & Cell cycle Protein metabolism & degradation Oogenesis & development Aging gene (methuselah)

130

534

3

87

81 0

284

B. No consistent pattern

90

Caloric Restriction

Control

Caloric Restriction (989)

C. Female reproduction Mitochondrial ETP Nucleolar products Down-regulated with age (458)

Figure 7.17 Animals raised under normal or caloric restriction (CR) conditions have different patterns of gene expression. The expression changes in the 1041 genes analyzed in the control group and the 989 genes in the CR group are displayed here in the form of a Venn diagram. The 667 genes up regulated with age in the control flies include 534 genes which are also upregulated with age in the CR flies, but there are 133 upregulated control genes not observed in the CR flies. The major pathways and processes represented by these changes are summarized in the text boxes and are discussed in the text. (After Pletcher et al. 2002.)

7.5 Drosophila melanogaster

note that down-regulation of reproductive genes occurs in aging females, as one would expect. However, the gene probes down-regulated by CR include additional genes involved in oogenesis and development, again indicating the inverse relationship between reproduction and extended longevity. Most interesting is the CR-induced down-regulation of at least one gene (methuselah, mth; see below), which significantly extends longevity when mutated. Its down-regulation under CR conditions suggests that the wild-type mth might act as a negative regulator of extended longevity. The fact that the expression profiles in the control and CR cohorts were qualitatively similar to one another suggested that CR extends life span in the fly by ameliorating many of the normal transcriptional changes that occur with age. This conclusion is supported by the fact that the extended longevity in these CR animals is not due to a reduction in reproductive activities, and so it cannot be attributed only to a shift of energy from reproduction to somatic maintenance (Mair et al. 2004). These gene changes do not seem to arise as a result of large-scale changes in chromosomal architecture, suggesting to the investigators that aging and death result from a large-scale break down in homeostasis and concomittant dysregulation of gene expression in older animals. A model for the mechanisms involved in the onset of senescence is presented in chapter 9. Mair et al. (2004) found many different patterns of changes in gene expression in Drosophila, but perhaps the most useful are those genes whose expression in both the CR and control animals is tightly correlated with mortality. The genes are not correlated with time, for mortality takes place over a longer period of time in the CR flies than in the controls, but they are correlated with the proportion of the cohort that is still alive. If the expression of these genes proves to be robust and repeatable, then their expression level might be used as a gene-level biomarker indicating the amount of life span left for the surviving members of the cohort. Should such genes be conserved and found in mice or humans, then their use might allow us to quickly determine if some putative antiaging intervention is working.

271

As was noted with CR in the mouse, different strains or mutants may have different sensitivity spectrums to CR. The chico mutant (affecting the ISP; see below) expresses its maximum life span at a higher food concentration than does the wildtype control (Clancy et al. 2002). Other mutants show similar changes in their optimal food concentration. Thus, the amount of food (i.e., calories) needed to trigger the CR response is not fixed in a species but is the outcome of the organism’s genotype and nutritional environment. The organism has several pathways that sense its nutritional state. The target of Rampamycin (TOR) signaling pathway (see figure 7.28) senses the amino acid levels so as to modulate growth or longevity. There may well be other specific nutrient-sensing pathways. Their variable interactions with the nutritional environment and with each other may account for the existence of different sensitivity spectrums to CR. The preceding discussion describes what CR does to a fly. But how does it do it? What gene pathways are involved? The evidence for mechanisms is less detailed but still informative. Two different histone deacetylases are involved in mediating the CR response. Flies carrying a mutant rpd3 deacetylase gene and raised on normal media have a mean and median life span about 40% longer than that of wild-type controls (Rogina et al. 2002). But rpd3 mutants raised on low-caloric food have a life span identical to that obtained if they are raised on normal food. The fact that the CR treatment has no effect on these mutants implies that the wild-type allele of the rpd3 gene plays a role in the CR response. Even more interesting is the response of the Drosophila Sir2 gene to CR and to rpd3. Wild-type animals subjected to CR show an approximate doubling in the level of Sir2 gene expression, while rpd3 mutants raised on normal food also show an approximate doubling in the level of Sir 2 gene expression. Thus, the normal allele of rpd3 seems to inhibit the Sir 2 gene expression, which is necessary for the CR effect. In another series of experiments, it was shown that if the Sir-2 histone deacetylase gene of Drosophila is up-regulated by feeding a known sirtuin activator such as resveratrol (the longevityextending component of red wine), then life span

272 Chapter 7 Genetic Determinants of Longevity in Animal Models is significantly extended, even though the animals are fed ad libitum (Wood et al. 2004). Conversely, flies lacking a functional Sir-2 gene failed to extend their longevity when fed resveratrol. Thus, activation of Sir-2 seems to be essential for the expression of CR-dependent extended longevity. It is reasonable to conclude that both these genes are components of the genetic pathway mediating CR expression, related perhaps in the manner suggested in figure 7.18. Another gene possibly involved in CR is the indy gene, which encodes a metabolite transporter protein responsible for the uptake and

transport of Krebs and citric acid cycle intermediates through the gut epithelium and into the appropriate organs (Knauf et al. 2002). Homologous indy genes exist in nematodes and mammals. Flies carrying a mutation in this gene display a 90% increase in their mean life span and a 50% increase in their maximum life span (Rogina et al. 2000). In flies, the indy gene is normally expressed primarily in the midgut, fat body (liver equivalent), and oenocytes, which are the main sites of intermediary metabolism. It is not unreasonable to assume that the mutant-induced down-regulation in the activity of this metabolite cotransporter pro-

Normal Food

CR

?

?

indy

indy

?

?

rpd3

rpd3

Sir2

Sir2

no longevity effects

gene effects (ISP?) leading to extended longevity

Figure 7.18 The indy gene presumably acts early in the caloric restriction (CR) process by transporting metabolites. Down-regulating its activity via mutations may lead to a state similar to that of a normal animal under CR conditions. Under normal nutritional conditions, the two histone deacetylases may be inactive and thus not lead to any prolongevity effects. Under CR nutritional conditions, the genes may somehow be activated and thus lead to prolongevity effects. Other interpretations are possible. Question marks indicate possible unknown reactions. (After Rogina et al. 2000, 2002).

7.5 Drosophila melanogaster

tein brings about a lower effective concentration of essential metabolites in the cell, thus creating a metabolic state similar to that induced by CR. One possible interpretation of its role in the CR pathway is shown in figure 7.18. An important insight into the nature of the CR response was provided by an experiment in which flies were initially raised on either a CR or ad-libitum (AL) dietary regime and were then switched at various ages to the alternative diet (Mair et al. 2003). AL-raised flies normally have a higher age-specific mortality rate (qx; see chapter 2) than do CR-raised animals. But after an AL to CR shift, the animals rapidly adopt the agespecific mortality rate characteristic of animals that have been raised on a CR regime for their entire life. A corresponding rapid upward shift in mortality rate is observed after a CR to AL shift. What this means is that aging has no cellular memory. An AL to CR shift, for example, induces the animal to alter its regulatory pathways and shift its gene expression pattern from a progrowth pattern to a prostress resistance pattern. The consequent change in damage patterns affects the mortality rate. The mortality rate is due only to the current levels of damage in the organism. The animal’s prior history does not directly play a determining role in its current mortality rate, although certainly the existence of prior unrepaired damage must have some effect. This finding implies that CR/ISP-dependent aging is an environmentally dependent cell-level function, and the presence of the systemic regulatory mechanisms present in multicellular eucaryotes does not invalidate this statement. A cautionary note is in order here: Not all flies react to CR in the same manner as described above. Mediterranean fruit flies (Ceratitis capitata) are distantly related to Drosophila and might reasonably be expected to react to CR in the same manner if the CR mechanism is highly conserved and public. However, a recent study reported that the Mediterranean fruit fly shows a more or less constant longevity at various levels of diet restriction, with a sharp decrease in mortality once the diet falls below 50% of the ad libitum level (Carey et al. 2002). There is no evidence for an increase in longevity at some level of dietary restriction.

273

Reproduction occurred across the range of diets tested. The mortality increase occurred in both sexes, even though males have no obvious counterpart to the energetic demands of egg production in females. These same flies lived longer when subjected to a “feast or famine” dietary regime (Carey et al. 2004). This regime likely resembles the normal situation in the wild, and so perhaps that environment selects for animals that can go into a survival mode when food becomes transiently scarce. Carey et al.’s study also suggests that not all species react in exactly the same manner to the same type of CR regime. This means that a conserved mechanism may be found in many species but may still not be universal, perhaps due to the interactions between the species’s different nutrition-sensing signaling pathways.

7.5.3 Metabolic Control of Longevity via the Insulinlike Signaling Pathway The generic structure of the ISP as determined from studies in yeasts and nematodes is shown in figure 7.10. This conserved, public longevityregulation mechanism is also found in the fly (figure 7.19). The ISP was initially investigated for its effects on growth and size of the fly, and mutants affecting the activity of ISP did affect these parameters (Weinkove and Leevers 2000). The ISP also affects blood sugar levels. The fly has five insulinlike proteins with significant homology to mouse and human insulin proteins. These proteins are expressed in several tissues but most particularly in small clusters of insulinproducing cells (IPCs) in the brain (Rulifson et al. 2002). Rulifson et al. showed that ablation of IPCs caused retarded growth and elevated carbohydrate levels and that a normal phenotype could be restored by expressing one of the Drosophila insulinlike proteins. Thus, there is a remarkable conservation of the insulin-based glucoregulatory mechanisms in flies and mammals with respect to their effects on growth and blood sugar. Once the experimental data allowed the concept of a conserved longevity-regulation mechanism to become clear from the nematode work, the fly’s ISP was investigated to see if it also regulated

274 Chapter 7 Genetic Determinants of Longevity in Animal Models

Yeast Pathway

Glucose

Gpr1

Sch9

Functional Category Nematode Pathway

Fly Pathway

Ligand

Insulin/IGF1-like molecule

Insulin/IGF1-like

Receptor

daf2 (insulin receptor)

InR

Ras2

G-proteins

Cyr1 (cAMP)

Second messengers

age1/daf23

PI3K

PKA

Serine-threonine kinases

akt/PKB

Akt/PKB

Msn2, Msn4

Stress resistance transcription factors

daf16

Fkh; dFOXO

MnSOD, Catalase, heat shock proteins, others.

Stress resistance proteins

MnSOD, CuZnSOD, heat shock proteins, others

MnSOD, CuZnSOD, heat shock proteins, others

LONGEVITY EXTENSION

Physiological Outcome of Altered Cell State

LONGEVITY EXTENSION

LONGEVITY EXTENSION

IRS

Figure 7.19 The yeast caloric restriction/stress-response pathway compared with the insulinlike signaling pathway of the nematode and the fly. The functional similarities arise out of their evolutionary. See text for details. (Adapted from Longo and Finch 2003.)

longevity in this organism. There are some obvious phenotypic differences between homozygous and heterozygous mutants in genes composing the ISP, but the important point is that certain mutants can significantly increase longevity. As you might expect, the absence of ISP is lethal. In rare cases, one can construct homozygous flies that contain two different mutations, each with a defect in a different part of the gene. Such unusual animals have a low but detectable level of ISP activity (~25% or less) and can survive through adulthood. These heteroallelic homozygotes contain two different mutations in their insulin receptor gene (InR; homologous to

the C. elegans daf-2 gene) and give rise to dwarf adults with different longevity effects on the two sexes (Tatar et al. 2001). Females express a delayed onset of senescence, with an 87% increase in their mean life span and about 45% increase in their maximum life span. These homozygous mutant females are small because the ISP controls cell size, and the low activity levels result in small cells and slow growth. They are sterile because the low ISP activity results in a significant decrease in the juvenile hormone levels that are essential for reproduction (see below). Male homozygotes are also small and semisterile, for the same reasons. In contrast, the males show no

7.5 Drosophila melanogaster

In Mortality rate

statistical change in their mean life span, even though they have an increased mortality during early to mid-adult life, followed by a lower mortality rate thereafter (figure 7.20). Incidentally, the smaller effect of the ISP on males seems to be a general phenomenon, and I touch on it in the section on reproductive effects below. Both heterozygous and homozygous InR mutants have an impaired synthesis of the steroid hormone ecdysone (Tu et al. 2002). This suggests that the increased longevity associated with the heterozygous InR mutants are dependent on a decreased level of juvenile hormone synthesis and the consequent reduction of ecdysone synthesis to levels that do not repress the animal’s level of stress resistance (see section on reproductive effects below for a description of the mechanisms that tie these observations together). Flies heterozygous or homozygous for a certain mutation in their IRS proteins (associated with the cytoplasmic side of the InR; see Figure 7.19) also yield long-lived females but not long-lived males (Clancy et al. 2001; figure 7.21). These flies were also observed to be more resistant to oxidative stress than controls. Thus, disabling the InR and/or IRS genes is sufficient to down-regulate the entire ISP. Finally, the dFOXO gene of Drosophila was shown to have sequence homology with the daf16 gene of C. elegans and to function in the regulation of growth (Kramer et al. 2003). Flies heterozygous for the dFOXO gene are incapable of expressing the extended longevity otherwise

resulting from mutational down-regulation of the InR gene (Hwangbo et al. 2004). This functional test demonstrates that the dFOXO gene is downstream of InR. Thus the structure of the ISP is the same in yeast and nematode and fly, for effects on the up-stream genes such as InR must be mediated through the down-stream dFOXO transcription factor. If the latter is mutationally inactivated, then the up-stream signals are ignored. This factor is believed to activate various stress-resistance genes and repress various progrowth genes in a manner similar to that demonstrated in the nematode. Both of these actions will result in a slower accumulation of age-related oxidative damage and in a delayed onset of senescence, as described below in the section on stress resistance. The simplest conclusion of these several experiments is that the modulation of longevity via the ISP is highly conserved and seems to involve similar (but perhaps not completely identical) mechanisms in our several model organisms. The fact that longevity regulation in the fly is intertwined in the same system with the regulation of growth, cell size, and fertility suggests that flies ISP has multiple specific regulatory functions compared to nematodes. This is consistent with the fact that the fly has at least four independent isoloci of the PI3K gene, each of which apparently control different sets of processes; humans have at least 16 PI3K genes (Samuels et al. 2004). This increased signaling complexity, which might have arisen as a consequence of the increased size and

0

0

-2

-2

-4

-4

-6

275

-6 0

20

40 Age (days)

60

0

20

40

60

Age (days)

Figure 7.20 Age-specific mortality of InR mutant genotypes (open symbols) relative to the wild-type (shaded symbols). The mutant in this case is homozygous for two different mutant alleles of the InR gene. Female data are in left panel; male data are in the right panel. Note the sex-specific effect. (After Tatar et al. 2001.)

276 Chapter 7 Genetic Determinants of Longevity in Animal Models

Proportion surviving

Female

Male

1

1

.75

.75

.5

.5

.25

.25

0

0 0

20

40 60 Time (days)

80

100

0

20

40 60 Time (days)

80

100

Figure 7.21 The effect of an IRS mutation (chico1) on life span of females and males either homozygous for chico1/ chico1 (solid circles), heterozygous for chico1/+ (diamonds), or wild-type (open circles). Note the sex-specific effect. (After Clancy et al. 2001.)

morphological complexity of the fly, might account for its apparent greater genetic complexity. The ISP is not the only signaling pathway involved in assaying nutrient availability. The TOR signaling pathway is an important regulator of growth and size and is found in organisms from yeast to humans. Experiments with Drosophila show that the TOR pathway senses amino acid availability and uses this information to modulate activity of the S6 kinase regulatory gene to enhance growth and repress extended longevity (Kapahl et al. 2004). TOR also plays an important role in regulating autophagy, or the digestion of the cell’s own components for energy (Klionsky 2004). Either overexpression of the upstream Tsc1 and Tsc2 genes or dominant-negative mutations in in the TOR or S6K genes give rise to extended longevity (Kapahi et al. 2004). As summarized in figure 7.28, these data suggest that the two upstream genes act as negative regulators of the TOR or S6K genes, which themselves act as negative regulators of longevity and a positive regulators of growth. It seems likely that the TOR pathway acts in parallel with, and perhaps even overlaps, the ISP.

7.5.4 Metabolic Control of Longevity via Nuclear–Mitochondrial Interaction There is a large body of information regarding the role of mitochondria in aging, which I discuss in chapter 10. But there is almost no definitive data

regarding the role of nuclear–mitochondrial interaction in the Drosophila aging process. This, however, does not mean that nuclear–mitochondrial interaction does not occur in flies, for there are at least three lines of suggestive evidence to the contrary. Taken together, they suggest but do not prove that the observed changes stem from some alteration in nuclear–mitochondrial communication. First, James and Ballard (2003) found that mitochondria and nuclei taken from flies indigenous to different parts of the species’ range yielded lowfitness flies when combined. This might come about if the mitochondria had evolved to be most effective in certain environments. Combining a (genetically different?) mitochondria from one area with a nucleus that evolved in a different region might result in organelle incompatibilities and hence a decreased life span. Second, the same laboratory collected strains of Drosophila simulans (a close relative of D. melanogaster) from comparable environments and assayed their longevity as well as aspects of their mitochondrial function. A particular long-lived strain was characterized by a high mitochondrial efficiency. Complex IV of their mitochondria’s electron transport chain (see figure 11.2) is more efficient than that of normallived control strains, as evidenced by their lower than normal oxygen consumption but normal levels of ATP production (Melvin et al. 2005). The nuclear-encoded protein subunits of the

7.5 Drosophila melanogaster

complex responsible for the increased efficiency contained several point mutations that resulted in amino acid changes, the presence of which are highly correlated with the altered mitochondrial function. It is reasonable to suspect that the selection of chance mutations in nuclear-encoded mitochondrial proteins supports the existence of metabolic signals that coordinate nuclear and mitochondrial activities and provide the feedback signals necessary to the selection process. Third, the selection experiment depicted in figure 7.14 yielded a long-lived La strain and a normal-lived Ra control strain. Their extended longevity arises in part from the fact that the La mitochondria produce 20–40% less hydrogen peroxide than do normal Ra mitochondria (Ross 2000). Using the Ra and La strains, Driver and Tawadros (2000) set up various crosses that allowed them to combine mitochondria from different selected strains with the same normal-lived Ra nucleus. They then examined the resulting “cybrid” strains to see if any of the combined genomes had an effect on longevity. As shown in table 7.2, combining mitochondria from either normal-lived strain (Ra or Rb) with the Ra nucleus led to no change in longevity, but combining either of the long-lived mitochondria (La or Lb) with the Ra nucleus led to a significant change. Thus, the Lamt strain achieves about 40% of the extended longevity seen in the La strain solely because it has La-type mitochondria. It would seem that mitochondria can make a significant contibution to longevity. Possible mechanisms for this phenomenon have been put forth elsewhere (Arking et al. 2002a,b). The important point here

277

is that different mitochondrial genomes appear to set up different stable metabolic equilibrium settings with the Ra nucleus, and each of these permit different longevities to be expressed. Whether any of these three cases constitute nuclear–mitochondrial interaction has not been determined. The fact that two different labs using different strains and different approaches have observed similar results might indicate that nuclear–mitochondrial interactions also take place in Drosophila.

7.5.5 Stress Resistance and Extended Longevity It has long been observed that mild or nonlethal stress often has the paradoxical effect of benefitting the organism by increasing longevity (Minois 2000). Conversely, it has also been suggested that all long-lived strains and mutants exhibit some form of stress resistance (T. E. Johnson et al. 1996; Parsons 1995). This relationship is thought to reflect the fact that the natural environment usually exerts substantial albeit variable stresses on organisms. Evolutionary considerations of Darwinian fitness will thus impose a premium on genotypes conferring metabolic efficiency and stress resistance (Parsons 1997, 2003). The magnitude of the effects of stress resistance on longevity is summarized in table 7.3. I examine four complementary lines of evidence bearing on the relationship of stress resistance and extended longevity in Drosophila. The first involves the use of strains selected for

Table 7.2 La Mitochondria Confer Longevity Source of Strain

Nuclear genome

Mitochondria genome

Ra Rb-mt La-mt Lb-mt La

Ra Ra Ra Ra La

Ra Rb La Lb La

Mean Life Span (days)

Lifespan Relative to Ra value

69 70 86 80 110

1.00 1.01 1.25 1.16 1.60

Source: data from Driver and Tawadros (2000).

278 Chapter 7 Genetic Determinants of Longevity in Animal Models Table 7.3 Effect of Nonlethal Stressors on Longevity % Response of experimental animals over controls

Stressor Cold Heat Hypergravity Physical Activity Irradiation Caloric restriction Oxidative Stress Resistance

5 10 12 20 20 50 33–60b

Molecules Involveda HSP HSP, ADS HSP ADS ADS, HSP ADS, others ADS, others

Source: data adapted from figure 1 of Minois (2000). aHSP,

heat shock proteins; ADS, antioxidant defense system proteins.

bResponse

noted after selection or transgenic modification of genome.

extended longevity, followed by an analysis of the mechanisms responsible for the altered phenotype. The second involves standard mutational techniques in which candidate genes thought to play an important role in longevity are inactivated and the life span of the experimental animals assayed to see if the original hypothesis was correct. The third approach involves genetic engineering or transgenic work in which extra copies of candidate genes are inserted into experimental animals and assaying their longevity to determine if the added gene has a significant effect on the life span. The fourth approach involves the environmental induction of stress-response genes and the analysis of their effect on the longevity of the tested animals. As a result of studies on the biochemistry and stress-resistance properties of the long-lived La strains, we knew that the only predictive factor clearly and significantly associated with extended longevity in these strains was an enhanced resistance to oxidative stress (Arking et al. 1991; Force et al. 1995). Thus it seemed logical to conclude that La animals probably live long because of higher than normal activity of the antioxidant defense genes early in life. Quantitative trait loci (QTL) mapping is a genetic method of identifying small chromosome regions that have a significant statistical effect on longevity. QTL may be viewed as a genome-wide scan that allows one to identify genes for further investigation. Curtsinger and Khazaeli (2002) conducted QTL mapping on recombinant inbred strains

derived from the La and Ra strains discussed above. They found four QTLs located on chromosomes 2 and 3 of the (La × Ra) recombinant inbred strains that accounted for almost all of the selection response. The major QTLs for both paraquat resistance and longevity are coincident with each other and are centered over a small region of chromosome 3L, which contains the loci of the CuZnSOD gene, and of several heatshock protein (hsp) genes. Both of these gene sets are involved in stress resistance. The other QTLs seem to be involved in maintaining female fertility. (One of these other QTLs appear to involve the ecdysone receptor gene, which is involved in reproductive processes; see Reproductive Effects section below.) The several genes involved exert their effects over the first 6 weeks of life but not thereafter. There are other minor contributors to the extended longevity phenotype (Curtsinger et al. 1998), but the available data indicate that the La strains live long primarily because of specific up-regulation of antioxidant defense genes. This finding supported the candidate gene approach in which researchers assayed the quantitative changes in the mRNA levels and the antioxidant enzyme activity levels of several loci during the development and early adult life of the normal-lived Ra and long-lived La strains (Dudas and Arking 1995). In addition, Hari et al. (1998) used antibodies to measure the amount of CuZnSOD protein present in these strains. Figure 7.22 summarizes many of these changes in

7.5 Drosophila melanogaster

279

3 2.8

Control strain Long-lived strain

Relative value

2.6 2.4 2.2

p < 0.05

2 1.8 1.6 1.4 1.2 1 0.8

ADS mRNAs

Enzyme active

Y HA

H AD

GS T

T CA

SO D

H XD

GS T

T CA

SO D

0.6 0.4 0.2 0 Non-ADS mRNAs

Figure 7.22 Relative levels of expression of antioxidant genes in young (5-day-old) normal-lived and long-lived strains of Drosophila. For each item listed, the left-hand bar represents the normalized value of the normal-lived strain, and the right-hand bar represents the relative value of that item in the long lived strain. Note that three of the four antioxidant defense system (ADS) mRNAs are significantly up regulated in the long-lived strain even at this early age, as are all three of the ADS enzymes. It is significant that none of the non-ADS mRNAs is elevated, suggesting that the ADS is specifically up regulated in these long-lived strains. (Data from Dudas and Arking, 1995.)

gene expression. The mRNA data demonstrate that at day 5 in the L strain, there appears to be a coordinately regulated significant increase in the mRNA levels of CuZnSOD, catalase, and xanthine dehydrogenase. There is a nonsignificant increase in GST mRNA during the same time period. These increases in mRNA levels are accompanied by significant increases in the enzyme activity of CuZnSOD, catalase, and GST. Other experiments showed that the amount of SODspecific protein is proportionately increased in the La strain during the same period (Hari et al. 1998). It seems reasonable to conclude that these alterations in gene expression in the long-lived strain are the result of a transcription-level change that alters the enzymatic arsenal available to the organisms. But these changes in gene expression have biological meaning only if they reduce the amount of oxidative damage in the long-lived animals. Figure 7.23 shows that the higher level of antioxidant gene expression and enzyme activities do in fact bring about a life-long and signifi-

cant reduction in the levels of the most common oxidative damages in proteins or lipids in the long-lived L strain. There is an inverse correlation between the levels of antioxidant enzyme activity (figure 7.22) and the levels of oxidative damage (figure 7.23). It seems reasonable to conclude that these animals developed the ability, as a consequence of artificial selection, to turn on a regulatory process that coordinately activates the antioxidant defense genes early in life, thereby protecting the animals against the oxidative damage to vital molecules and thus delaying the onset of senescence until their antioxidant defenses fall to normal levels. Reverse-selecting these longlived strains for shortened longevity (see the ReL strains of figure 7.14) reverts their antioxidant gene expression patterns to control levels. Only the antioxidant genes (and certain other enzymes operationally connected to them) show these correlated and coordinate changes in gene expression. Other metabolically important enzymes that do not affect the antioxidant proteins do not show any significant changes as a result of selection

280 Chapter 7 Genetic Determinants of Longevity in Animal Models

20

normal ADS levels, normal lived 15

10

high ADS levels, long lived

5

0 1

5

9

12

20

22

30

40

50

54

67

70

Life span in days

Figure 7.23 The selected long-lived L strains have significantly lower levels of oxidative damage products, such as lipid hydroperoxide, relative to the normal-lived R strains thoughout their life span. Overall, the L animals have about 57% of the lipid hydroperoxide levels of the R animals and about 77% of the R protein carbonyl level (data not shown). (Redrawn from data presented in Arking et al. 2000.)

(Arking et al. 2000a, 2002b). The reversal in life span was accompanied by a specific reversal in the expression of only the antioxidant genes and genes necessary to their function. Incidentally, an independent replicate strain (Lb) has the same longevity patterns as does the La line (see figure 7.14) but uses different specific patterns of antioxidant gene expression. One interpretation of these data is that the overall oxidative stress level of the organism may be more important than which particular antioxidant gene is overexpressed. Taken together, this series of experiments reveals the existence of a causal relationship between antioxidant gene expression, oxidative stress resistance, levels of oxidative damage, and longevity in these selected strains of Drosophila. In a parallel experiment, Rose (1984) used a different wild-type progenitor stock but the same indirect selection protocols as used for the Wayne State University (WSU) lines to create a set of long-lived strains called the University of California-Irvine (UCI) long-lived lines. The physiological traits associated with the UCI selected lines overlap those associated with the WSU lines, for

both sets of strains are significantly more resistant to environmental stresses than their respective controls. The two stains are resistant to a different spectrum of stressors, which may well be the result of the different genetic backgrounds used in their different progenitor stocks. The UCI lines are resistant to starvation, dessication, and oxidative stress; and selection for increased starvation or dessication resistance was shown to lead to increased longevity (Harshman et al. 1999; Rose et al. 1992). The WSU lines are mildly resistant to dessication but mostly resistant to oxidative stress (Force et al. 1995). Although several kinds of stress resistance are associated with extended longevity, it may not be a coincidence that the only common stress resistance in all these strains is that to oxidative stress. Genetic data generally support these selection experiments. Phillips et al. (1989) created a CuZnSOD-null mutant of Drosophila and showed that the absence of this SOD activity significantly decreased viability and longevity. Subsequent analysis (Parkes et al., 1998) showed that the absence of the CuZnSOD gene has a num-

7.5 Drosophila melanogaster

ber of important pleiotropic effects such as (1) adult sensitivity to paraquat, (2) male sterility, (3) female semisterility, (4) adult hyperoxia sensitivity, (5) larval radiation sensitivity, (6) developmental sensitivity to glutathione depletion (an important antioxidant molecule), and (7) adult life span reduction. Before one could confidently interpret these results, it was necessary to determine whether these alterations stemmed from an increase in the rate of aging or from some abnormal pathology. An experiment was done showing that these CuZnSOD-null mutants showed an acceleration, relative to the wild-type control, of the normal age-related temporal changes in the expression of certain other genes (Rogina et al. 2000). Because the acceleration in the temporal expression of these other genes was proportional to the shortened life span, this was interpreted as showing that the shortened life span of the CuZnSOD-null mutants is due not to an abnormal pathological process, but to an increase in the rate of aging. This interpretation suggests that the aging rate is directly proportional to the animals’ level of antioxidant capacity and resistance to oxidative stress, a finding in keeping with the selection results discussed above. In contrast to the SOD data, acatalesemic mutants of Drosophila are essentially normal when reared under standard conditions as long as they have at least 3% of the normal catalase expression level (Mackay and Bewley 1989), a finding consistent with the transgenic work done on this gene (Orr and Sohal 1992).This suggests that catalase is not normally a rate-limiting factor in longevity. Transgenic technology, which allowed the insertion of an extra gene into an otherwise normal organism, was used to test the effects of increasing the level of CuZnSOD gene expression on the longevity and aging of the altered animals. Early experiments inserted single copies of either CuZnSOD or catalase into the test organisms, but results were inconclusive for a variety of reasons. However, the tandem overexpression of both CuZnSOD and catalase in the same animal did extend median and maximum longevity by up to 34% in some lines, while simultaneously retarding oxidative damage and increasing oxidative

281

resistance (Orr and Sohal 1994; Sohal et al. 1995a). Sun and Tower (1999) used a controllable transgenic system which allowed them to control when CuZnSOD overexpression would take place in adult flies. This system allowed increases in mean life span of up to 48%. These two transgenic alterations of gene expression ostensibly affected all tissues of the organism at all stages. But there is much information showing that most genes have characteristic tissue- and stage-specific expression patterns. Thus it was important when Parkes et al. (1998, 1999) showed that a GAL4UAS transgene expression system that selectively targeted CuZnSOD expression to the adult motor neuron was capable both of restoring the normal adult life span of CuZnSOD-null mutants and extending by 40% the adult life span of an otherwise wild type fly. Overexpression of CuZnSOD in the adult central nervous system, adult muscle, or larval body has no effect on adult longevity. Adult Drosophila have a surprising lack of CuZnSOD activity in their central nervous system relative to the rest of the body (Klichko et al. 1999). It may be that the fly’s motor neurons have the lowest age-related failure threshold of the whole body and would normally be the first critical tissue to fail. Thus, using transgenes to increase their resistance to oxidative stress may have the effect of postponing the age at failure of this critical tissue and thus lengthen the life span. Scientific progress does not occur in a straight line. The validity of the three transgene experiments described above have been challenged by two of the researchers involved on the basis that an increased longevity was only observed when the control flies had a comparatively short life span (Orr and Sohal 2003). According to Orr and Sohal, shorter lived control lines are helped by CuZnSOD overexpression, but genetically robust controls showed little or no effect. This reassessment casts doubt only on the efficacy of increasing life span by increasing CuZnSOD expression; it does not invalidate experiments showing that suites of antioxidant genes are overexpressed in long-lived strains (see figure 7.22), the finding that different antioxidant genes act together in a cooperative manner in the fly (Missirlis et al.

282 Chapter 7 Genetic Determinants of Longevity in Animal Models 2001), or experiments involving robust control flies. It is a likely possibility that the reason for the failure of some transgenes to significantly affect the life span of some flies is that an organism with an inefficient metabolism and low levels of available ATP is simply not in a position to effectively reallocate the energy saved due to lowered oxidative damage levels to increased somatic maintenance. The mechanism discussed in figure 7.27 may be applicable here as well. In the La strain, the effective use of the increased antioxidant defense enzymes requires simultaneous alterations of metabolism (e.g., shifting from glycolysis to the pentose shunt) so as to support the enzyme functions, and it also requires mitochondrial changes that yield an increased efficiency and thus increase the levels of available ATP (Arking et al. 2002a,b). Both genetic and metabolic pathways need to be changed if an animal is to live long. Altering the expression of one gene without altering the necessary pathways may bring about only a weak effect. MnSOD is the mitochondrial version of superoxide dismutase. Given the crucial role of mitochondria in energy metabolism and generation of reactive oxygen species, it seems logical that MnSOD plays an important role in modulating the life span. This assumption is borne out by selection data (Arking et al. 2000) and by the transgenic data of Sun et al. (2000). Sun et al. used their controllable transgene system to induce the overexpression of MnSOD in only adult flies but not in the developmental stages, thus avoiding complications in the analysis. They reported that MnSOD showed increases in expression of up to 75%. This yielded a 33% increase in mean life span and a 37% increase in maximum life span. The simultaneous overexpression of both CuZnSOD and MnSOD led to a complicated situation wherein each transgene partially inhibited the overexpression of the other, but nonetheless the two genes still had partially additive effects on life span (Sun et al. 2004). Phillips and colleagues (2000) also used transgenes to overexpress the MnSOD gene in wild-type animals and found that they obtain life span extensions of about 30%. They also noted that the overexpressed MnSOD gives an incomplete

rescue of the CuZnSOD-null mutant, establishing that the two enzymes operate in functionally different compartments. Destroying the MnSOD mRNA and thus silencing this one gene in a normal animal disrupts mitochondrial function, increases sensitivity to oxidative stress, and yields a striking 80% reduction in mean and maximum life span (Kirby et al. 2002). In contrast, Mockett et al. (1999) reported that their transgenic MnSOD lines did overexpress MnSOD mRNA, protein, and enzyme activity but did not show an increased life span relative to the controls. This finding may have to do with the genetic background of their control strain; if so, it represents a limitation, but not a refutation, of the ability of any one specific antioxidant gene to increase life span for the reasons presented above. Given the complexity of the stress-resistance process, it is inevitable that other genes are involved. Glutathione is perhaps the most abundant low-molecular-weight antioxidant present in the animal and represents a potential clue to other candidate genes. Mockett et al. overexpressed the glutathione reductase gene in transgenic Drosophila and obtained up to 100% overexpression of the enzyme. Longevity was significantly enhanced under hyperoxic conditions but not under normoxic condtions, suggesting that glutathione reductase may not be a rate-limiting factor in antiaging defenses under normal conditions but may be a factor when the level of oxidative stress is elevated. This is similar to the effects of GST on nematode longevity (Leiers et al. 2003). Another set of candidate genes are those that regulate the expression of the antioxidant structural genes. Four different mutant searches have identified such genes. In the first, Lin et al. (1998) did P-element mutagenesis of the third chromosome and screened for long life (relative to the white control strain) at 29°C. One homozygous mutant, methuselah (mth), lived up to 35% longer and was more resistant to paraquat, starvation, and high temperature. The mth gene appears to code for a transmembrane G protein-coupled receptor presumably involved in the regulation of stressresponse genes. Other data suggests that mth may be a negative regulator of stress-response genes. Recent data suggest that the ligand for this mth

7.5 Drosophila melanogaster

receptor is the stunted (sun) protein, but the functional pathways involved are not yet known. The second approach also used P-element mutagenesis but focused on the second chromosome and identified two groups of trans-acting mutants, one of which acted as if it were a normally positive regulator of CuZnSOD and catalase in wild-type animals and the other of which acted as if it were normally a negative regulator. The third approach showed that up-regulation of the Jun-nuclearkinase (JNK) signaling pathway (see table 7.11) made the animals much more resistant to oxidative stress while increasing both their mean and maximun life spans (Wang et al. 2003). Finally, the fourth approach used microarrays to conduct a genome-wide search for genes that responded to chronic exposure to oxidative and other stresses (Giradot et al. 2004). The data show the existence of both general and specific responses to these different stressors and indicate the existence of a complex interlocking network of stress-response genes. Four independent experiments have thus demonstrated that single genes extending longevity seem to do so by up-regulating the animal’s abil-

283

ity to withstand various types of stress, and these individual genes may be components of a larger network of stress-response genes. A summary of the various experiments discussed above is given in table 7.4. The heat-shock protein genes were initially found in Drosophila. These proteins have significant effects on stress resistance and longevity in all organisms (see discussion of nematodes above and figure 10.9). There are two complementary findings: first, the expression of the hsp genes is affected by aging, and second, the up-regulation of some (but not all) of these genes can significantly affect longevity. Aging animals usually express abnormal patterns of hsp expression relative to young animals (Niedzwiecki and Fleming 1990). Two studies showed that hsp22 mRNA is up-regulated during aging, particularly in the head (King and Tower 2000; Wheeler et al. 1995). In addition, an earlier onset of hsp22 and hsp23 mRNA accumulation in Drosophila selected for increased longevity was reported (Kurapati et al. 2000), suggesting a possible correlation between hsp22 levels and longevity.

Table 7.4 Summary of Genetic Interventions Testing the Relationship between Resistance to Oxidative Stress and Altered Longevity Manipulation

Genes involved

Effect on life span

Effect on stress resistance

Selection for long life

CuZnSOD, MnSOD Catalase No data

Increased Increased Increased

Catalase reduced Catalase increased SOD1 reduced SOD1 increased

Catalase Catalase CuZnSOD CuZnSOD

Decreased No effect Decreased Increased

Increased Increased Increased oxidative stress, increased starvation and dessication resistance No data Increased Decreased Increased

SOD1, Catalase increased

CuZnSOD, catalase

Increased Increased (some strains) (some strains) Increased Increased

SOD2 reduced SOD2 increased

MnSOD MnSOD

Note: SOD, superoxide dismutase.

No increase over SOD alone Decreased Decreased Increased No data No effect No effect

Reference Arking et al. (2000a) Arking et al. (2000b) Rose (1984)

Mackay and Bewley (1989) Orr and Sohal (1992) Plillips et al. (1989) Parkes et al. (1998) Sun and Tower (1999) Orr and Sohal (1993) Orr and Sohal (1994) Sun and Tower (1999) Kirby et al. (2002) Sun et al. (2002) Mockett et al. (1999)

284 Chapter 7 Genetic Determinants of Longevity in Animal Models Using genetic techniques to overexpress hsp22 in motor neurons led to a 30% increase in longevity and stress resistance (Morrow et al. 2004b). Conversely, knocking out the hsp22 gene so that flies could not express the HSP22 protein led to a 40% decrease in their longevity, coupled with an increased sensitivity to stress (Morrow et al. 2004a). Finally, ubiquitous overexpression of hsp22 throughout the body led to a reduction in life span as well as to increased sensitivity to heat and oxidative stress (Bhole et al. 2004). We can see that increased longevity and stress resistance depends not just on the overexpression of the hsp22 gene, but on its being expressed in the appropriate tissues at the appropriate time and in a balanced manner relative to the expression of stress-resistance genes in other tissues. Failure to achieve this balanced expression harms, rather than helps, the organism. This may come about because an unbalanced expression of one stress-response gene may inappropriately down-regulate other stress response genes. This hypothesis assumes that there is a stress-response gene network of some sort, as was noted in the nematode. Empirical evidence to support this assumption is provided by a DNA hybridization screen that identified at least 13 genes activated by exposure to heat, oxidants, and starvation (Wang et al. 2004). A beneficial effect of HSPs on aging and stress resistance is observed when organisms are preconditioned by exposure to a mild stress before being exposed to a subsequent damaging stress (hormesis; see chapter 10; Le Bourg et al. 2001). The animal’s ability to resist stresses likely plays an important role in determining its longevity, and this observation forms the basis of the model presented in chapter 9 to explain the mechanisms underlying the organism’s transition from a healthy state to a senescent state. Taken together, the different types of experiments discussed above show that (1) there is a variety of different but interacting stress-resistance pathways and (2) almost all tested longlived animals are also stress resistant. Parsons (2003) has argued the existence of a tight relationship between stress resistance and longevity.

Reciprocal experiments in which one directly upregulates a known stress response gene, such as hsp26 or hsp27, have been done and result in flies having significantly increased stress resistance and longevity (Wang et al. 2004). This experiment shows that at least some stress-resistant animals are also long lived. The experiments thus verify the fundamentals of Parson’s (2003) argument.

7.5.6 Genetic Stability There are two aspects to genetic stability in Drosophila. One aspect has to do with the structural integrity of the DNA and the genome. The other has to do with the nonstructural epigenetic modification of the genome. I discuss them here in sequence. It was once hypothesized that aging might be the result of genomic instability and/or gene dysregulation, but recent data indicate instead that the genome is very stable during the adult period. For example, even though 86% of the genes assayed significantly changed their expression during the entire life cycle (i.e., embryo to young adult; Arbeitman et al. 2002), only 9% of the genes in the Drosophila genome showed significant agedependent changes in their levels of expression during adult aging (see figure 7.17). Furthermore, plotting the genomic distribution of this small set of changeable genes reveals no obvious clustering of the genes at particular chromosomal regions, such as telomeres or centromeres, once thought to be particularly unstable (Pletcher et al. 2002). In addition, there is no evidence for a genome-wide gene dysregulation. There is, however, extensive evidence that gene expression during adult aging is a precisely regulated process. Independent studies using enhancer-trap techniques show that the expression of many genes changes in their own stereotypical patterns with age (Helfand and Rogina 2000). If these aging changes were the result of some stochastic loss of function, then one would not expect such a random process to yield different but predictable signature patterns. A different enhancer-trap study found that the rate of change of expression in some cases was correlated with changes in longevity under various conditions (i.e.,

7.5 Drosophila melanogaster

temperature), suggesting that the changes are indicators of physiological function and not of some stochastic loss of genome stability (Seroude et al. 2002). Genes encoding functionally related proteins, such as cell-cycle proteins or metabolic enzymes, tend to be expressed at similar times during the life cycle, again suggesting a functional basis for the observed changes (Arbeitman et al. 2002). Finally, a spatial analysis of individual gene patterns of expression throughout the entire adult body revealed that each gene had its own characteristic signature tissue localization—a finding incompatible with stochastic genomic instability (Seroude et al. 2002). Each of these phenomena can be best understood as reflecting the operation of signal transduction mechanisms over the life course (see also chapter 12). Of course, genetic stability is affected by the relative levels of mutation and repair processes. As was documented for the other model organisms, decreased DNA repair activity leads to widespread somatic mutations and a consequent loss of function. For example, Leffelaar and Grigliatti (1984) isolated several temperature-sensitive DNA repair mutants that had significantly decreased adult longevity relative to controls when raised under restrictive conditions. Another DNA repair protein (Rrp1), when overexpressed, was capable of reducing the level of somatic mutations in animals exposed to oxidative stress (Szakmary et al. 1996). The normal functioning of these and other repair genes likely plays an important role in maintaining genetic stability. Taken together, the data suggest that the genome is remarkably stable during the aging process, that most of the genes do not show significant alterations in gene expression during this process, and that the observed changes in gene expression represent the dynamic regulation of specific genes in response to the changing physiological status of the organism. Note that none of the foregoing implies that these regulated gene expression patterns constitute an aging program. If the genome of an organism is so stable, then how can environmental influences affect its operation? Epigenetic regulation of gene expression occurs when somatic cells use either DNA methylation to modify the genetic material itself or when

285

they use histone acetylation/methylation to modify proteins that intimately interact with the genetic material (Czermin and Imhof 2003; Jaenisch and Bird 2003). These epigenetic modifications arise in an orderly fashion during development, stochastically during aging, or as the result of long-term external influences such as diet. This phenomenon was first detected in Drosophila in 1930 as the occurrence of variable phenotypes in neighboring somatic cells of an individual fly. Neighboring cells in the same tissue had different patterns of gene expression, suggesting a particular instability of their genomes. A gene later shown to be important in that process, Su(var)3–9, is now known to be a histone methyltransferase and has conserved orthologs in yeast and humans (Schotta et al. 2003). The importance of these phenomenon for the topic of aging is that various stressresistant genes in normal-lived Drosophila are repressed in part by the epigenetic effects of histone deacetylation in those nucleosomes about which the chromatin is coiled. The absence of acetyl groups on the histone causes a tight winding of the chromatin about the nucleosome, making it difficult for various transcription factors to gain access to the gene promoters. Acetylation of the histone causes a loosening or partial unwinding of the chromation about the nucleosome, allowing the genes to be activated by the various factors. A drug approved by the Food and Drug Administration for use with various diseases, 4-phenylbutyrate (4PB), inhibits the action of histone deacetylases and thereby induces hyperacetylation of the histones. When Kang et al. (2002) fed 4PB to normal-lived Drosophila, the optimal concentrations increased the mean and maximum survival in both sexes by 30– 52%, depending on the sex and genetic background (figure 7.24). Figure 7.24 also shows that the largest effect on life span was observed when flies were fed the drug during their mid- and latelife phases. The drug had no effect on the flys’ fecundity or body weight. Furthermore, the treated flies were resistant to starvation, oxidative stress, and loss of locomotor ability. All of these observations are consistent with a heightened stress resistance. Assays of gene expression showed that a number of stress-resistant genes

286 Chapter 7 Genetic Determinants of Longevity in Animal Models

Day 12

A

fed PBA from days 0-12 100 80 Virgin females fed PBA from day 12 onward

60 control 40 20 0 0

10

20

30

40

Days at 29 C° Figure 7.24 The drug 4-phenylbutyrate (PBA) extends lifespan whether fed early or late in life. Newly emerged w1118 flies were fed on normal food containing 10 mM PBA for 12 days and were then transferred to normal medium without the drug for the rest of their life span. Another group was fed on normal medium for the first 12 days, then with PBA for the remainder of life. (After Kang et al. 2002.)

(including CuZnSOD) were induced by the drug, and their induction was accompanied by altered levels of histone acetylation. These findings are intriguing for several reasons. First, they show that manipulation of the epigenetic code to specifically affect genetic stability is sufficient to significantly alter gene expression. Second, they demonstrate that pharmecutical intervention is capable of significantly altering longevity. Third, they show that treatment of middle-aged adults yields significant delays in the onset of senescence. I return to these three points in chapter 15.

7.5.7 Reproductive Effects The antagonistic interaction between the soma and germ line cells observed in the nematode (figure 7.13) is also found in the fly. In fact, we have much more information on the hormones involved and their various inhibitory and stimulatory effects in flies. A summary view of these relationships and their evolutionary conservation is shown in figure 7.25, and the following descrip-

tion is taken from Tatar et al. (2002, 2003) and Hwangbo et al. (2004). The par intercerebralis, cells of the central nervous systems, secrete insulinlike peptides (ILPs) that are directed to certain target cells as well as being released systemically. These ILPs (particularly dILP2) directly or indirectly stimulate certain neuroendocrine cells such as the corpora allata to produce one of the two key hormones in the insect, juvenile hormone. dILP2 activates ISP and thus inactivates dFOXO and the stress resistance/somatic maintenance pathways. The ISP (figure 7.19) likely operates in the cells of the central nervous system as well as in the peripheral tissues. In the fly, the fat body cells in the head play a particularly important role because these cells appear to regulate dFOXO, and thus control the aging of the organism, when activated (Hwangbo et al. 2004). The ISP operating in the gonad–somatic cell axis can activate or inhibit the synthesis of juvenile hormone. Thus, the longevity-extending effects of mutants affecting the ISP (see figures 7.20 and 7.21) presumably arise out of interference with the operation

7.5 Drosophila melanogaster

287

tive form of a second key insect hormone, 20hydroxy-ecdysone (20HE), which is well known for its effects on development. The molecular details of the action within the cell of 20HE have been worked out, particularly the fact that the 20HE must bind with a bipartite receptor protein in the nucleus if it is to specifically activate its target genes (Riddiford et al. 2000; Tatar 2003). What is particularly interesting in the current context is that the known inhibitory effect of juvenile hormone on the stress resistance of the adult may well be mediated through 20HE.

of the ISP portion of the process. As illustrated in figure 7.25, longevity arises from the interaction of control centers in the gonads, central nervous system, and head fat-body cells with the various peripheral somatic tissues. Juvenile hormone is essential for reproduction. It promotes vitellogenesis and reproduction in insects and inhibits adult diapause (a nonreproductive long-lived somatic state characteristic of, for example, overwintering adults). In the gonad, juvenile hormone stimulates egg development and also stimulates the synthesis of the ac-

External Cues

NE System pars intercerebralis Head Fat Body InR

Peripheral Tissues d ILP2

dFOXO

InR

Other TF dFOXO

InR

dFOXO JH

Somatic Cells

Germline Cells

GONAD

Figure 7.25 A model for the endocrine circuits of aging regulation in Drosophila. Environmental cues act on the neuroendocrine (NE) system and stimulate the cells of the pars intercebralis to produce insulinlike peptides (ILPs), one form of which (dLP2) is most effective at stimulating the insulin receptor (InR) in all tissues. Activation of the InR results in the inhibition of dFOXO (the fly homologue of DAF16) via the mechanism shown in figure 7.19 and the activation of other transcription factors (TF) to activate growth and reproduction genes. The fat-body cells of the head secrete dFOXO, which inhibits dILP2. The corpora allata (endocrine gland) produces juvenile hormone (JH). JH promotes vitellogenesis and suppresses response to external stress in peripheral tissues by indirectly allowing the phosphorylation of dFOXO, which restricts it to the cytoplasm (see figure 7.11), inhibits the InR-activating effect of the somatic gonad cells, and stimulates the germline cells to inhibit dFOXO. This sets up a series of opposing circuits involving the somatic germ cells and dFOXO on the one hand and the germline gonad cells, InR, JH, and dILP2 on the other. (Redrawn from Tatar et al. 2002; Hwangbo et al. 2004.)

288 Chapter 7 Genetic Determinants of Longevity in Animal Models The relationships of these two hormones are quite complex; refer to Pu et al. (2005) for details. Simon et al. (2003) have shown that animals bearing heterozygous mutations in the ecdysone receptor (EcR) protein, one component of the nuclear receptor protein complex, exhibit increased longevity relative to control animals (figure 7.26). This locus is the probable site of a QTL involved in longevity (Curtsinger et al. 1998). The heterozygous animals have a higher metabolic rate and a lower rate of spontaneous activity. They are also significantly more resistant to oxidative stress, heat, and starvation than are normal animals. This implies that the increased life span and the increased stress resistance both are the result of a decrease in the effective concentration of the 20HE in the cells. Female age-specific fecundity was increased in these heterozygous mutants relative to normal animals, suggesting that moderate levels of 20HE are compatible with both extended longevity and enhanced fecundity. A comparable increase (~42%) in life span was obtained by use of a temperature-sensitive mutant (DTS-3)

which affects ecdysone synthesis in females. Because this gene product is inactivated at high temperatures, one can effectively turn it on or off simply by transferring the fly from a permissive temperature (20°C) to a restrictive temperature (29°C). Doing such shifts at different times indicates that the greatest effect of the decreased hormone levels on longevity takes place within the first 2 or 3 weeks of adult life. Finally, feeding 20HE to these mutant female flies abolished the extended longevity in a dosesensitive manner and also reversed their resistance to the stress of starvation. As little as 10–3 M 20HE added to the food allowed the females to have a normal life span. Absence of ecdysone is lethal. But these several experiments taken together clearly show that high levels of 20HE have a negative effect on longevity and on stress resistance. Thus the stimulation of the gonads by juvenile hormone to synthesize and secrete high levels of 20HE leads to a shorter (i.e., normal) life span. This decreased life span may arise from the concomittant repression of stress resistance observed in these experi-

Surviving flies (%)

100 cn EcR V559fs bw cn EcR + bw

cn EcR + bw

50

0 0

40

80

Time at 25 C ° (days) Figure 7.26 Extension of life span in an animal heterozygous for a mutation (EcRV559f) affecting the ecdysone receptor (EcR). Survival curves for cn EcRV559fs bw/cn EcR+ bw males and females and for cn EcR+ bw controls raised at 25°C are shown. Male and female heterozygotes showed a significant 45% increase in mean life. (After Simon et al. 2003.)

7.5 Drosophila melanogaster

ments. Using mutants to reduce the level of 20HE to a moderate (~50%) level brings about extended longevity and enhanced fecundity, as do experiments in which the effective level of the receptor protein is manipulated. Note that the approximate 45% increase in mean life span obtained with either manipulation affects both sexes and is about equal to the increase in life span obtained with selection (figure 7.14), caloric restriction (figure 7.16), or ISP mutants (figures 7.20 and 7.21). There is another aspect of the reproductive effects on longevity. Reproduction requires the female fly to expend a considerable amount of energy. Individuals in which the energy demands of the reproductive system are not synchronized with the energy production ability of the somatic cells are likely to die when the demands exceed the supply. A formal analysis of the interplay between the age-related energy demand and supply led to a hypothesis about a mechanism that predicts two critical periods in the life history of an individual fly (Novoseltsev et al. 2003). As shown in figure 7.27, the first crisis occurs at early ages when the increased energy demand becomes greater than the available energy supply. This would often result in a premature or nonsenescent death and would presumably involve females in which their intrinsic rate of egg production is greater than can be supported by their intrinsic mitochondrial energy production. In other words, the weaker flies would die at this first stressful period in their lives. The stronger and surviving flies lay eggs at some more or less constant rate, while their available energy supply is gradually decreased by various senescent processes. Eventually, they do not produce enough energy to simultaneously maintain their reproductive activities and to resist the various environmental stresses associated with living. The initially strong flies perferentially die late in life from a senescent-caused death precipitated by their inability to meet the cumulative energy demands placed upon them. Note that these two mechanisms of reproductive effects on longevity—the hormonal mechanisms and the energy demands—are not contradictory but are really the same process as

289

described on the one hand from the cell and tissue level of the geneticist and on the other hand from the organismic level of life-history theory.

7.5.8 Patterns of Senescence Do all tissues of the fly age together and at the same rate? The data of figure 7.15 imply that both normal- and long-lived animals undergo the same sort of senescent process, losing different traits in the same sequence and at the same stage. Because the animals do not lose all functions at the same time, different tissues must age at different rates. Are there any data to support these implications? The fact that there is a defined spatio-temporal pattern of gene expression during the aging process (Seyroude et al. 2002), coupled with the fact that at least 9% of the genes undergo transcriptional changes during aging (Pletcher et al. 2002), suggests that it is unlikely that all tissues of the fly age together and at the same rate. Some data to support this supposition were presented by Cook-Wiens and Grotewiel (2002). They showed that different behaviors of the fly, utilizing different sensory systems, age at different rates. For example, locomotor and olfactory performance decreases with age even while the aging animal maintains its ability to respond to electric shock and light. Different sensory-motor behaviors decline at different rates within the same animal. What is most interesting is the fact that the olfactory and locomotor declines occurred at the same rate and at the same times in normal-lived control animals as they did in longlived, stress-resistant (mth) mutants. Thus, genetic manipulations that enhance resistance to oxidative stress and extend life span do not necessarily protect against functional senescence in all pathways. We not only age in an individual manner, but each of us ages in a heterogeneous manner. Figure 7.28 summarizes the known activation and/or inhibition signals generated by the longevityextending mechanisms known to be operative in flies, which I have discussed in this section. The schematic is somewhat speculative in detail but its tying together of the several methods of inducing stress resistance with the expression of extended longevity is most likely correct in concept.

290 Chapter 7 Genetic Determinants of Longevity in Animal Models

Demands and Supplies

Strong flies Highest investments into reproduction Weak flies

Lowest investment into reproduction

Age Distribution of deaths M2

M1 Strong flies Weak flies

X1

X2

Age

Figure 7.27 Predicted general mechanism of premature deaths in female flies. Two groups of flies, strong and weak, are shown with age-related trajectories of the power supply represented by thin dashed lines. The solid-line patterns represent age-related individual power demands endowed for the highest to lowest levels of egg production. The demand/supply intersection points define the ages of death. Weak flies mostly die early, thus forming the M1 mode of the death distribution curve around age X1, whereas the strong flies die around age X2, at the M2 mode. Normally, the flies suffer senescence-caused death, but in the case of overly endowed rates of egg production, death may occur prematurely. A bimodal distribution of deaths arises, with M1 mode corresponding to premature deaths, and M2 corresponding to senescence-caused ones. (From Novoseltsev et al. 2003.)

7.6 Mice and Other Mammals 7.6.1 Background Information The laboratory rat and mouse have many characteristics, such as small size, short life span, and a physiological similarity to humans, that have endeared them to laboratory scientists over the past century or so. One indication of the increasing popularity of these animals is to compare an older bibliography that lists about 750 references to studies of aging done with genetically defined strains of rodents in the 11 years between 1977 and 1988

(Hazzard and Soban 1988) with the finding that about 750 published studies on the topic of mouse aging were listed in 2002 alone on the PubMed web site. Expensive and time consuming as mice may be, our increased knowledge of aging mechanisms makes it more likely that studying these furry relatives of ours will yield information with practical significance for humans. A much larger variety of inbred strains and mutants is available in mice than in rats, so we concentrate our genetic inquiries on the former. In fact, during the past half century and as a result of the meticulous accumulation and study of genetic variants, the mouse has become one of the world’s best organisms for genetic research. The

7.6 Mice and Other Mammals

291

Longevity Pathways in Drosophila Stress Resistance Pathways

ISP Pathway Environmental Factors

Growth Regulation

Caloric Restriction

Drugs

Selection

Genetic Engineering

Indy?

ISP Gene Cascade

4PB

Rpd3 HDA

CNS

?

Tcs Sir2 TOR

dFOXO High Conc.

Low Conc.

Juvenile Hormone

Other HDA SGK

20HE

?

Breed for long lived flies JNK

Insert Extra Genes

dFOXO Other Anti-stress HDA Increased Genes Repair Genes Mitochondrial Efficiency

Gene Activation

Stress Resistance Gene Activity

Reproduction Stress Resistance

ROS Levels

Tissue Damage

LONGEVITY Figure 7.28 A schematic diagram integrating all the pathways empirically known to extend longevity in Drosophila.

mouse has proven useful as a genetic tool for exploring mammalian developmental genetics; its value in studies on aging is now widely realized as well. However, I incorporate relevant data from other rodents and from primates (i.e., monkeys and humans) when available to test the public nature of the mechanism under discussion. Historically, investigations into the genetics of aging in the mouse began with a comparison of longevity among different strains. There is a wealth of survival data for these inbred strains, and their genealogies are well known (Atchley and Fitch 1991). Their use spread since their characteristics were assumed to be stable and repeatable both within and between generations. Phelan (1992) compiled data illustrating that each of these inbred

strains has a unique qualitative and quantitative pattern of nonfatal lesions, which make them useful for disease-specific studies. Some strains are very susceptible to certain lesions; other strains are not. For example, 30% of DBA/2NNia mice show amyloidosis of the kidney glomeruli, compared to 0% of A/HeNNia mice. The main scientific rationale for the use of inbred strains was that such strains would exhibit the smallest amount of variation between individuals, and this was thought to be a useful experimental characteristic. Imagine then the consternation when a thorough review and comparison of the uniformity of inbred strains compared to F1 hybrids, involving plant, invertebrate, and vertebrate species, revealed that inbred strains often displayed

292 Chapter 7 Genetic Determinants of Longevity in Animal Models significantly more phenotypic variability than did the hybrid lines (Phelan and Austad 1994). The reasons for this involved the comparative developmental and physiological stability of heterozygous compared to homozygous genomes; the latter are paradoxically less stable than are hybrids and so exhibit greater variance for most phenotypic traits. This finding undercut the main reason for the use of inbred strains. This situation led to the deliberate development of a genetically heterogeneous mouse strain (HET-NIA) derived from four different inbred parents (Jackson et al. 1999). The procedure for generating the HET-NIA strain is shown in figure 7.29. Mating these animals in a controlled manner yielded an F2 generation of sibling mice, each of whom receives 25% of its genes from each of its four grandparents. The virtue of this genetic history is that while no two mice are genetically identical, each mouse shares a random 50% of its genetic endowment with any other mouse in the population. They are, in effect, full genetic sibs. Inasmuch as any investigator can at any time take the same four inbred lines and generate a statistically identical population, this genetically heterogeneous population is replicable. It is also more similar genetically to the obviously outbred human population, and thus more likely to allow the uncovering of useful biomarkers of aging. And if one is searching for mechanisms that give rise to differences in longevity, then perhaps one needs a genetically heterogeneous strain so that

one can correlate the meiotic segregation of chromosome segments with the phenotypic segregation of longevity markers. Genetics can only be done if genetic differences exist, and this HET-NIA strain yields multiple replicable differences. The heterogeneous HET-NIA strain has already yielded some interesting results. One is the identification of immunological biomarkers of aging, as described in chapter 12 (R. A. Miller 2001). Another is the finding that loci associated with life span exhibit both sex-specific and epistatic (i.e., interaction) effects that lead to complex patterns of genetic effects. The confirmation that low body weight in early life predicts future longer life span suggests that our senescence patterns may have their roots in early life processes (R. A. Miller et al. 2002c). Finally, the fact that low levels of certain hormones in early- or mid-life are sex-specific predictors of life span recalls some of the Drosophila data and lends support to the neuroendocrine data discussed below (Harper et al. 2003). The choice of a strain must depend on the purpose of the experiment, however, for neither inbred nor heterogenous strains are suitable for all uses (Festing 1999; McClearn 1999). Investigators must still make an intelligent decision regarding experimental goals. Investigators can now use replicable inbred or outbred lines as they think most appropriate for the goals of the experiment. In addition to the inbred strains mentioned above, a series of recombinant inbred strains that

Production of Genetically Outbred Heterogeneous Mouse Lines

P

1

BALB/cJ x C57BL/6J

F (BALB/cJ x C57BL/6J) 1

F 2

x

C3H/J x DBA/2J

inbred lines

(C3H/J x DBA/2J)

hybrid lines

genetically heterogeneous sibling offspring

test lines

Figure 7.29 The mating scheme used to generate statistically identical populations of genetically heterogenous mice.

7.6 Mice and Other Mammals

undergo genetically mediated accelerated aging also exist (Takeda et al. 1981). These senescenceaccelerated mice (SAMs) were created from an accidental outcrossing of AKR/J mice with AJ (and perhaps other) mice to yield a series of genetically and phenotypically distinct lines, each of which display a variety of senescence-associated pathological phenotypes (Kitado et al. 1994a). Some of the lines no longer display the lymphomas characteristic of their parent AKR/J strain, suggesting the loss or lack of expression of certain parental genes. They exhibit no obvious alteration in their developmental processes; they simply senesce significantly earlier than does the AKR/J strain from which they were derived. There has been a long standing consensus among biogerontologists that the study of animal models with reduced life span, such as the SAMs, will eventually prove to be unrelated to the mechanisms of normal aging (Harrison 1994). The researchers working with the SAMs disagree (Kitado et al. 1994b). We might consider viewing the SAMs within the same conceptual framework that we view disease: as indicating the weak points and failure modes of the evolved animal model, the study of which may indicate by default the processes that need to be modulated if longevity is to be extended (Warner and Sierra 2003). I discuss the genetics of aging in mice using the same framework as for the other three model systems, with one alteration. There is now sufficient data regarding the changes in tissue-specific gene expression as the mouse ages under normal conditions that we may incorporate this information into our discussion, particularly with respect to CR. A listing of genetically altered mice, which includes informative summaries of each, may be found online at http://sageke.sciencemag.org/resources/experimental/transgenic/. I refer to many of these strains in the following discussion.

7.6.2 Caloric Restriction Modulation of Normal Patterns of Gene Expression in Aging I discussed the effects of CR on mammalian longevity in chapter 6 and concluded that it is a robust intervention that significantly extends

293

longevity by delaying the onset of senescence (see figure 6.1). The evidence presented there suggested that CR exerts its effects on the individual cells via a neuroendocrine mechanism, and this might explain how this nutritional alteration brings about such a systemic alteration in the organism. The problem we are interested in here has to do with describing and understanding the effects of CR on the genetic functioning of the different tissues of the mouse. In order to answer this question, we would first need to know how the genetic functioning of different tissues changes under conditions of normal aging. The use of DNA microarrays to simultaneously scan thousands of different genes and assay their individual gene expression patterns has increased exponentially since 1997. The experimental paradigm of choice for aging research would use this technique to simultaneously measure the patterns of gene expression in same-aged control and experimental groups of ad libitum-fed (AL) and CR mice, followed by the appropriate analysis and comparison of the CR patterns with the AL patterns to highlight any changes. The technique is technically difficult, financially expensive, and statistically daunting; reviews of the technologies may be found in Weindruch et al. (2002) or in a special issue of Nature Genetics (see Axion 2002). It is now clear that DNA microarrays is not reliable in detail, since false positives are generated by differential splicing and/or inadvertent cross-hydridization and are aggravated by the different standards used by different microarray manufacturers (Marshall 2004). In many cases, however, the data have been confirmed by independent polymorase chain reaction analysis of individual genes. A database summarizing the known CR effects on biomolecules and oxidative stress is freely available at http:// agingdb.bio.pusan.ac.kr/ (Park et al. 2003). It is currently agreed that this technique can yield useful information if the comparative studies focus on biologically robust broad themes rather than only on the expression of specific genes (Marshall 2004). By focusing on the broad biological themes (e.g., the general categories of genes up- or downregulated by various treatments) uncovered by microarray analysis, we will get a reasonably informative view of what is going on. Table 7.5 sums up the overall results of some recent experiments

294 Table 7.5 Comparison of Gene Expression Changes Brought about by Aging and by Caloric Restriction (CR) in Different Tissues of the Mouse Brain

Ad libitum fed

Skeletal muscle

Heart muscle

Neocortex + cerebellum

↑ Stress response HSP response DNA repair genes Oxidative Stress genes

↑ Structural Proteins ECM components Cell adhesion Cell growth Collagen deposition

↑ Inflammation Complement MHC molecules Microglial factors Inflammation peptides

↑ Neurodegeneration Nerve injury markers

↑ Stress response HSP response Oxidative stress genes Proteases

↑ Neuronal injury Reinnervation Neurite growth ↓ Energy metabolism Reduced glycolysis Mitochondrial dysfunction

↑ Carbohydrate Metabolism ↓ Fatty Acid Metabolism FA translocases FA oxidation enz.

↓ Protein turnover Ubiquitin suppression ↓ Growth factors Developmental genes Trophic factors Neural migration

Cortex

Hypothalmus

↑ Inflammation (Oldest mice only)

↑ Proteases Signal processing Cell death b-amyloid regulation

↑ Proteases Signal processing Cell death b-amyloid regulation ↑ Neurodegeneration Synaptic integrity * protein & AD Sleep disorders ↓ ATPases ↓ Synaptic Proteins Neurotransmission Memory Cortex specific prot. ↓ Neuronal Struc. Neural growth Neural migration Synapse formation ↓ Stress Response Hsp response

↑ Neurodegeneration Synaptic integrity tau protein and AD Sleep disorders ↓ ATPases ↓ Synaptic Proteins Neurotransmission Memory Hypo.specific prot ↓ Neuronal Struc. Neural growth Neural migration Synapse formation ↓ Stress Response DNA repair genes ↑ Metabolic Enz. Mito. ETC

CR fed

↑ Protein metabolism Synthesis Turnover ↑ Energy metabolism Pentose phosphate shunt Gluconeogenesis ↑ Biosynthesis Fatty acid synthesis Nucleotide precursors ↓ Macromolecular damage Hsp response Detoxification response DNA repair

Reference

Lee et al. (1999)

↓ Structural Proteins ECM components Actin and tubulin Collagen deposition Basement membrane ↓Immune Genes MHC genes Complement genes ↓Genetic Stability DNA damage DNA repair genes Apoptosis inhibited

Lee et al. (2002)

↑ Growth Factors Developmental genes Neurotrophic support ↑ DNA synthesis Nucleotide precursors DNA replication factors

No data (Note: specific gene changes are seen if animals are raised in an enriched environment)

Altered Neuroendocrine Levels and circadian rhythms ↓GHRH, POMC, and CP ↑ NPY (Note: specific gene changes are seen if animals are raised in an enriched environment)

↓ Protein synthesis tRNA synthesis Elongation factors ↓ Stress Response Ox stress response Hsp response NfB signaling Immune Modulation Interferon induction Inflam. suppression Lee et al. (2000)

Jiang et al. (2001)

Shimokawa et al. (2003); Jiang et al. (2001)

Note: HSP, heat shade protein; ECM, extracellular matrix; MHC, major histocompatibility complex; AD, Alzheimer’s disease; GHRH, growth hormone releasing hormone; POMC, proopiomelanocortin; NPY, neuropeptide Y

295

296 Chapter 7 Genetic Determinants of Longevity in Animal Models involving several different tissues. By scanning horizontally across the top half of the table, one can see that each tissue has its own characteristic pattern of age-related changes during normal aging. Skeletal muscle and heart muscle, for example, are both postmitotic tissues with similar functional and similar structural cellular organization. Yet they age in very different manners. Skeletal muscle aging is characterized by an increase in stress-induced damage, loss of motor neurons followed by reinnervation of muscle fibers by the remaining neurons, and a large decrease in energy metabolism and biosynthesis. Cardiac muscle, on the other, shows a different spectrum of processes affected by normal aging. There is a striking change in its energy metabolism from a reliance on fatty acid metabolism in the young cell to one on glucose metabolism in the old cell, and this is accompanied by changes in neural integrity and various structural proteins. The striking contrast that takes place in the metabolic shifts associated with normal aging in these two tissues, and their differences when compared with the various neural tissues also profiled in table 7.5, is evidence enough that each tissue has its own unique transcriptional and functional alterations. These changes in gene expression are not unique to the mouse in particular but may be characteristic of mammalian aging in general. An analysis of gene expression changes in the skeletal muscle of rhesus monkeys revealed that normal aging resulted in selective up-regulation of genes involved in oxidative stress and inflammation and down-regulation of genes involved in mitochondrial respiration and energy production, a pattern similar to the changes seen in mice (Kayo et al. 2001). If one examines the proteins produced by older animals, one finds similar sorts of changes. For example, the pituitary gland of old mice shows a pattern of protein increases and decreases suggesting that this gland is undergoing increased oxidative stress and has a greater burden of damaged proteins (Marzban et al. 2002). This protein pattern is compatible with the gene expression changes seen in the neural tissues (table 7.5). Despite the tissue specificity of these patterns, there are certain broad functional similari-

ties in the aging processes of the several tissues described in table 7.5. All tissues show alterations in their stress responses, and this response is likely coupled to increased levels of reactive oxygen species (ROS) in both muscle and neural tissue and/or to their increased level of damage or injury to various cellular components (i.e., increased hsp expression). They also show changes in their ability to process signal proteins or to synthesize and degrade other proteins. Both increased ROS production and decreased signal sensitivity/protein turnover might set off positive feedback cycles that progressively degrade the aging cell’s performance in these and other areas. What is particularly interesting is that the listed changes in gene expression between young and old animals can be plausibly connected in a mechanistic manner to the morphological and functional changes that take place in each of these tissues. If one could delay or alter these gene expression changes, then one might be able to delay or alter the pattern of aging. Caloric restriction brings about a variety of changes that seem to maintain the youthful function of the tissue by reducing the stress levels within the cell while retaining the optimal metabolic, biosynthetic, and turnover capabilities of the cells (see bottom half of table 7.5). It would be incorrect to say that CR simply and completely reverses the gene changes characteristic of normal aging (Dhabi and Spindler 2003; Han et al. 2000). The present data suggest that CR does repress some normal aging changes while inducing other changes that alter the metabolism of the cell. The resulting new profile of gene expression is more likely to resist the age-related loss of function characteristic of AL-fed normal animals. One can construct a plausible connection between these CR-induced gene expression patterns and the maintenance of function. What is needed is data from the same tissue on both the mRNA expression changes and the changes in protein levels and activity. Such data exist for the mouse liver. Caloric restriction administered to young mouse over periods of time ranging from 2 to 8 weeks (short term CR) produced about 70% of the gene expression changes induced in mouse liver over a much longer term

Table 7.6 Effects of caloric restriction (CR) on Gene Expression in the Mouse Liver Genes increased with age

CR effects

Genes decreased with age

CR effects

Genes unaffected with age

CR effects

20 ↑

14↓

26↓

13↑, 13 NE

33

11↑, 22↓

Inflammation Stress/HSPs

8 5

6↓ 3↓

Cell cycle/DNA repair Xenobiotic metabolism

6 5

2↑ 3↑

Energy metabolism Apoptosis/cell growth

11 7

2↑, 9↓ 4↑, 3↓

Apoptosis Other

3 4

2↓ 3↓

Urinary proteins Other

4 11

4↑ 4↑

Xenobiotic metabolism Intracellular signaling Stress/HSPs Other

5 4 2 4

3↑, 2↓ 4↓ 2↓ 2↑, 2↓

Normal fed

Summary Total genes altered by CR = 79 (0.007%) Apoptosis/cell survival 16 Energy metabolism 11 Xenobiotic metabolism 10 Inflammation 8 Stress/HSPs 7 Urinary proteins 4 Intracellular signaling 4 Other 19 Source: after Cao et al. (2001). Note: NE = no CR effect relative to control; HSP, heat shock protein.

297

298 Chapter 7 Genetic Determinants of Longevity in Animal Models CR (27 months; Cao et al. 2001). A summary of these gene expression changes is shown in table 7.6. Cao et al. used microarrays to assay about 11,000 genes. Although not the complete genome, this sample is large enough to yield representative results. Note that most of the genes in the mouse liver show no change with either normal aging (e.g., an AL diet) or with CR aging. Only 79 genes (~0.0071%) showed significant changes, reinforcing the idea that the genome is remarkably stable during adult life. Note that these data are fully compatible with the conclusions drawn from the AL/CR microarray analysis of Drosophila aging (figure 7.17). A review of the general categories of genes altered by CR suggests that the normally aging AL cell is under increasing stress as evidenced by inflammatory and stress responses, decreased energy and xenobiotic metabolism, and changes in cell survival and apoptosis. This finding is in general agreement with the results of other studies examining CR effects on other tissues of the mouse (see table 7.5). The CR effects in mice manifest rather quickly, even in old mice, and these results are summarized in table 7.7 (Dhahbi et al. 2004). Seven-month-old mice were raised on either a control diet (AL) or a CR diet until they were approximately 32 months old. At that time, the AL mice were shifted to CR for 2, 4, or 8 weeks, and their gene expression was assayed at the end of that period. The long-term CR mice were simultaneously shifted to a control diet for 8 weeks or allowed to remain on the CR diet, and their gene expression was assayed as well. Caloric restriction begun relatively late in the life span (~32 months) was as effective as long-term CR (begun at 7 months) in reducing the mortality rate and extending the mean and maximum longevity. CR appears to act rather rapidly (e.g., within an 8week period) in bringing about these mortality and survival changes. Eight weeks of CR enabled the expression of about 72% of the total CR genes induced, and this partial expression of the CR genomic profile was sufficient to significantly alter mortality to the same extent as did lifetime CR. In an independent study, it was shown that short-term CR decreased mitochondrial oxidant

production but paradoxically increased the level of oxidative damage (Judge et al. 2004). This probably came about because the animal’s ability to scavenge oxidants was transiently suppressed by the initial reduction in mitochondrial antioxidant enzyme activities. Finally, the early and intermediate genes are most responsive to nutritional changes, and thus these two categories likely contain many of the regulatory genes that initiate the subsequent gene cascades. (Note that the experiments described in tables 7.6 and 7.7 used different strains and genders of mice. The numbers are not directly comparable, although the general patterns certainly are.) The CR and AL mouse livers have also been analyzed at the enzyme level. The liver plays a critical role in maintaining glucose homeostasis, mediated in part by various hormones and in part by the metabolic activities of the liver. Glucose can be generated (gluconeogenesis) from the breakdown products of peripheral proteins, and it can be destroyed (glycolysis) by the normal catabolic activities involved in the Emdem-Meyerhoff pathway and the Krebs cycle. Normal aging is accompanied by a decline in the enzymatic capacity to undergo gluconeogenesis, suggesting that normal aging is accompanied by a decrease in the turnover of peripheral protein. One effect of this change is a decrease in the liver’s ability to precisely regulate glucose levels. Another effect is diminished ability of the liver to support high levels of protein turnover. Caloric restriction increases the liver’s capacity for gluconeogenesis while somewhat reducing its capacity for glycolysis. This change enhances its capacity to support high levels of protein turnover. As shown in figure 7.30, glycolytic metabolism involves three irreversible, regulated steps. Normal (AL) aging alters the enzyme activities in the liver so as to encourage glycolysis and inhibit gluconeogenesis. The CR-induced enzyme alterations strongly send the metabolic flux in the other direction. Figure 7.31 shows that extrahepatic tissues such as muscle derive part of their energy requirements by using amino acids derived from protein turnover to generate energy via the Krebs cycle. The ammonia generated in these reactions is captured as glutamine and transported

Table 7.7 Temporal Effects of CR on Gene Expression in Mouse Liver Age at CR start 33 months 32 months 7 months 33 months

CR period

Temporal response

No. of genes affected

% of Total

Effect on mortality

Genes reversed by normal feeding

% of CR reversed

2–4 weeks 8 weeks 27 months 2 weeks

Early genes Intermediate genes Late genes Oscillating genes

15↑, 22↓ 7↑, 10↓ 14↑, 21↓ 16↑, 18↓

57.7 71.5 100.0 —

— 42%↓ ~40%↓ —

11, 5 5, 10 14, 21 11, 16

43 88 100 —

Source: after Dhahbi et al. (2004).

GLUCOSE CR: 1.6&2.3 Age: 29%

G6 Pase

GK

Diet: NC Age: NC

G6P

Fru 6-P

Diet: NC Age: NC

PFK-2/FBPase-2 Fru 1,6-P2ase

PFK-1

Fru 2,6-P2

Diet: NC Age: 1.3

CR: 7&21% Age: NC

Fru 1,6-P2

RNA CR: 1.9&2.2 Age: 17% ACTIVITY CR: 1.7

PEP PEPCK PK

RNA CR: 58&66% Age: 1.6 ACTIVITY CR: 60%

OA PYR AA AA PDH

CR: 7&21% Age: NC

Acetyl-CoA Figure 7.30 Summary of the effects of age and caloric restriction (CR) on glycolytic and gluconeogenic pathways of the mouse liver. Only the irreversible steps are shown; see text for details. Substrates are not boxed, enzyme names are in shaded boxes, and amino acids are indicated by “AA.” The value after “Age” is the main effect of age. The values after “CR” are the fold change in the young and old mice, respectively. A down arrow indicates the percent decrease, an up arrow indicates the fold increase. NC is no change. The value given for “Age” is a combination of both dietary groups. (From Dhahbi and Spindler 2003.)

299

300 Chapter 7 Genetic Determinants of Longevity in Animal Models

MUSCLE

LIVER

Glutamate Synthetase

Glutamate

Muscle Protein

Glutaminase Amino Acids Transdeamination TCA Cycle

NH4 + CR: 1.3&2.1 Glutamate Glutamate AGE: 29% Synthetase

Glutamate CR: 2.5&2.2 + AGE: NC NH4 CR: 1.9&1.8 AGE: 20% CPSI Carbamyl Phosphate

RNA: CR: 1.6&2.0 AGE: 37% ACTIVITY CR: 37&31% Age: 27%

Tyrosine Aminotransferase CR: 1.6&2.0 AGE: 37%

Glutamate Urea

Urea Cycle

Tyrosine

Figure 7.31 A summary of the effects of age and diet on muscle and liver nitrogen metabolism in mice. Substrates are not boxed; enzyme names are in shaded boxes. The value after “Age” is the main effect of age. The values after “CR” are the fold change in the young and old mice, respectively. A down arrow indicates the percent decrease, an up arrow indicates the fold increase. NC is no change. The value given for “Age” is a combination of both dietary groups. If muscle is being rapidly turned over, as in caloric restriction (CR), then glutamine is released from muscle and transported to liver where it enters the urea cycle. Non-CR animals do not show this effect. (From Dhahbi and Spindler 2003.)

to the liver, where it is fed into the urea cycle and excreted. Note that the enzyme changes brought about by AL aging inhibit this process, while those activity changes induced by CR enhance it. A large part of the normal decrease in muscle protein synthesis (and the cellular remodeling implied by its existence) must flow from these changes in hepatic gene expression and enzyme activities. These data should make clear that CR does not just make the experimental animal thinner than the AL control; it makes it different. Although the studies are not complete, it certainly seems as if primates, including humans, also respond to CR treatment with the same panoply of metabolic and physiological changes seen in rodents (Mattison et al. 2003; Walford et al. 2002). It will be most interesting to see if primates and rodents react to CR with the same broadly similar types of gene expression and enzymatic changes. Certainly, older humans respond to CR conditions within a matter of months (Walford et al. 2002), in a manner reminiscent of the mouse (table 7.7). There are other ways by which nutrition can affect longevity. Restricting the intake of certain amino acids (e.g., methionine) of rats on a AL dietary regime results in a significant increase in

longevity similar to that seen with the traditional CR regime (Zimmerman et al. 2003). This likely arises via an effect on the TOR pathway which senses amino acid levels and interacts with the ISP (see figure 7.28). The pattern of CR-induced gene changes in the several tissues discussed above is not the only gene pattern consistent with extended longevity. Long-lived mice with mutations in the neuroendocrine system seem to do it differently. For example, only about 9% of the mouse liver genes affected by CR are also altered in the livers of long-lived dwarf mice (R. A. Miller et al. 2002a). This observation gives rise to two possibilities. Continuation of the present work will allow us to eventually describe all the gene patterns consistent with extended longevity, however induced. Given that information, we should be able to identify those genes that are altered in the same manner in all sorts of different types of long-lived mice and that are thus common to all models of delayed aging. Such a finding would indicate a common core mechanism of longevity extension. Failure to find a common core would indicate that there are multiple independent pathways to extended longevity. In this context, we should also note that other non-nutritional interventions

7.6 Mice and Other Mammals

(e.g., exercise, an enriched environment) can also induce specific changes in gene expression patterns in these same tissues (see chapters 6 and 12), some of which may have a synergistic effect on longevity. Our genes are quite literally in touch with our environments and change accordingly (see genetic stability section below). The concept of normal gene activity probably needs to be calibrated against precisely defined environmental components. At this time, I believe it likely that both outcomes are correct: Extended longevity depends on a common core of essential gene expression patterns plus environment-specific patterns of expression in other supporting genes. Evidence to support this view was presented by Tsuchiya et al. (2004). Long-lived dwarf mutant mice or CR normal mice show changes in 389 genes, of which 100 are altered in both types of long-lived mice. These data may be interpreted as showing that the additive effects of dwarfism and CR on longevity develop both from their independent effect on unique genes and their additive effects on other common genes. The common set likely provides us with a sampling of genes closely associated with the regulation of longevity in mammals. Similar results were found by Mastermak et al. (2004). It is reasonable to assume that CR-induced gene expression changes prevent deleterious agerelated changes in function. Preventing these changes without having to undergo a 40% reduction in calories would be desirable. The extent to which such age-related gene expression changes have occurred in a tissue may allow us to measure the effectiveness, or lack thereof, of potential interventions. There are both technical and conceptual problems associated with this proposal, but using the major gene expression changes as tissuespecific biomarkers of aging may offer a way to develop a new functional measure of timeindependent aging. Recent work in Drosophila has shown that there are certain genes whose change in expression signals the amount of life yet to be lived (Pletcher et al. 2002) or the onset of senescence (Tower et al. 2004). If similar sorts of gene changes are identified in mammals, then the possibility of gene-based functional biomarkers may become real.

301

The most wide-ranging change induced by CR might be its effect on the hypothalmus (table 7.5). The hypothalamus is an integral part of the neuroendocrine system and plays an essential role in coordinately regulating bodily functions (see chapter 12). There is much evidence, discussed below and elsewhere in the text, that CR may bring about its systemic effects on all the different tissues of the body via some sort of hormonal effect. The work of Shimokawa et al. (2003a) shows that CR treatment alters the circadian rhythms and concentrations of certain neuroendocrine trophic factors. Neuropeptide Y (NPY; table 7.5) and proopiomelanocortin have opposite effects on food intake and energy expenditure. Fasting activates NPY neurons. Increased NPY concentration reduces the synthesis of growth hormone releasing hormone (GHRH), which in turn affects growth hormone (GH) levels. GH indirectly affects the levels of insulinlike growth factor 1 (IGF-1), which is a component of the ISP and is implicated in regulation of longevity (see below). Not surprisingly, the sera of CR animals differs from that of AL animals in that the former have significantly lower levels of triglycerides, insulin, and IGF-1 (de Cabo et al. 2003). Most important, it has been shown that if cells taken from an AL animal are exposed to CR sera in vitro, they will respond as if they were CR cells, and vice-versa (de Cabo et al. 2003). It has also been shown that the mammalian Sir2 gene (SIRT1) is induced by CR in rats, as well as in human cells in vitro treated with sera from these CR rats (Cohen et al. 2004). This also works in vivo, for the effects of CR on ROS production and mitochondrial bioenergetics are reversed if CRmaintained rats are continuously infused with physiological levels of insulin (Lambert and Merry 2003). There seems to be a complex neuroendocrine cascade connecting a CR diet with genetic mechanisms. CR certainly appears to be a cellular-level process, but in complex eukaryotes CR primarily affects the neuroendocrine cells that send out the appropriate circulating signals (IGF-1, etc.) to allow individual cells to respond in the corect manner to the nutritional signals from the

302 Chapter 7 Genetic Determinants of Longevity in Animal Models environment. The CR-dependent activation of SIRT1 results in the indirect inhibition of stressinduced apoptotic cell death, which likely promotes the long-term survival of mammals by promoting the long-term survival of irreplaceable cells (Cohen et al. 2004). Thus, it seems most likely that the decreased calories taken in by a CR mouse are transduced into a sera-based (low glucose/low [insulin/low IGF-1) signal, which then affects the gene expression pattern of the hypothalmus and, by affecting its neuroendocrine output, can alter the ISP in key cells, and indirectly the proximal circulating hormone levels, thus systemically affecting all cells and tissues exposed to the new hormonal regime. This explanation also provides a plausible mechanism by which the condition of maternal nutrition during pregnancy and lactation can affect the life span of the offspring (chapter 3).

7.6.3 Metabolic Control of Longevity via the Insulinlike Signaling Pathway Studies on invertebrate models have suggested that down-regulating the ISP leads to a delayed onset of senescence and to extended longevity. A key question is whether these effects are conserved in vertebrates. Recent evidence suggests that they are and that manipulating the ISP results in significant alteration of longevity in mice and rats. The ISP is more complex in mammals than it is in flies. There are, for example, more different types of somatic tissues, each of which requires its own retinue of signals. What was one ISP consisting of multiple ligands and one receptor in the fly has become diversified into complementary ISP-type systems with specific ligands, multiple receptors, and different effects on longevity. Nonetheless, the basic structure of the mammalian ISP is highly conserved and similar to that found in invertebrate models (figure 7.32). A brief summary of this mammalian ISP, based on that of Carter et al. (2002), follows. GH is produced in the anterior pituitary and is normally present in very low levels. GH is modulated by two hypothalmic hormones: GHRH, which

stimulates both the synthesis and secretion of GH; and somatostatin (SS), which inhibits GH release in response to GHRH. In addition, GH feeds back to inhibit GHRH secretion, possibly via a direct effect on the GH-producing cells. GH receptors are found throughout the body. Activation of GH receptors stimulates the synthesis and secretion of the IGF-1, mostly in the liver but also in many other cell types. A portion of this IGF-1 protein appears to be homologous to proinsulin—hence its name. Binding of the IGF-1 protein to its receptor (IGF-1R) alters macromolecular synthesis, causes mitogenic effects, and so forth. Both GH and IGF-1 binding proteins play a role in regulating the level of IGF-1. Other IGF isoforms exist and have different effects. Mice whose GHR gene has been disrupted can survive as homozygotes or heterozygotes (Coschigano et al. 2000). However, a significant extension of mean longevity is only seen in the homozygotes (table 7.8). The homozygotes weigh less and have very low concentrations of IGF-1 in their plasma. Homozygous null IGF-1R mutations are lethal. However, the heterozygous IGF-1R –/+ animals are viable and live long when maintained on an AL regime (figure 7.33; Holzenberg et al. 2003). The 33% increase in female mean life span relative to controls is statistically significant, but the 16% increase in the male longevity (not shown) is not. Note that in this strain of mice (129/J), males usually outlive females, and so the treatment reverses the normal situation. Neither body weight nor metabolic rates were affected, but the males were hyperglycemic and were not resistant to oxidative stress, while the females were normo-glycemic and were significantly more resistant to oxidative stress. These sex-related dimorphisms in glucose homeostasis and oxidative stress resistance may have masked any pro-longevity effects of the IGF-1R reduction in the males, which lived a normal life span. It is possible that the higher blood glucose levels in the males somehow activated their cell-level ISPs despite the loss of IGF-1R (de Cabo et al. 2003). In the females, the 50% reduction in IGF-1R levels altered all major IGF-1–based intracellular signaling processes, including a reduction in the

7.6 Mice and Other Mammals

Functional Category

Yeast Pathway

Nematode Pathway

Fly Pathway

303

Mouse Pathway GH

Glucose

Ligand

Insulin/IGF like molecule

daf2 (insulin receptor)

Receptor

Insulin/IGF1-like

IGF-I

InR

IGF-1R

Gpr1 G-proteins

Ras2

Second messengers

Cyr1 (cAMP)

age1/daf23

PI3K

PI3K

PKA

akt/PKB

Akt/PKB

Akt/PKB

Msn2, Msn4

daf16

Fkh; dFOXO

FOXO

Stress resistance proteins

MnSOD, Catalase, heat shock proteins

MnSOD, CuZnSOD, heat shock proteins

MnSOD, CuZnSOD, heat shock proteins

SOD, catalase, heat shock

Physiological Effect

Glycogen

Fat & Glycogen

Fat

Fat

LONGEVITY EFFECT

LONGEVITY EXTENSION

LONGEVITY EXTENSION

LONGEVITY EXTENSION

LONGEVITY EXTENSION

Serine-threonine kinases Stress resistance transcription factors

Sch9

IRS

Ras

Figure 7.32 The yeast caloric restriction/stress-response pathway compared with the insulinlike signaling pathway of the nematode, fly, and mouse. The functional similarities arise out of their conserved evolutionary relationship. (Redrawn from Longo and Finch 2003.)

phosphorylation of the Shc66 protein, which modulates oxidative stress resistance and may be the operative mechanism in these mice. Just as important as the increased life span is the fact that this extended longevity seems to be cost-free, in that the females were normal-sized and had normal fertility. This observation indicates that interventions at the IGF-1R level may not have detrimental effects. Rats whose GH/IGF-1 secretory axis has been affected such that it yields a 40% reduction in the animals’ IGF-1 levels also yield a modest but statistically significant 9–13% increase in mean and maximum longevity (Shimokawa et al. 2003b).

Table 7.8 Effect of Growth Hormone Receptor Knockout on Life Span Gender Males

Females

Genotype

Lifespan (days)

+/+ +/– –/– +/+ +/– –/–

629 ± 72 668 ± 51 975 ± 106 (155%) 749 ± 41 701 ± 36 1031 ± 4 (38%)

Source: from Coschigano et al. (2000).

304 Chapter 7 Genetic Determinants of Longevity in Animal Models

Females

Fraction alive (%)

100 80 60

lgf1r lgf1r

40

+/-

+/+

20 0 0

6

12

18 Age (months)

24

30

36

Figure 7.33 Life span extension in Igf1r+/– mice, which have a presumed 50% reduction in the number of their IGF-1 receptors, with respect to Igf1r+/+ (wild-type) mice. Female Igf1r+/– animals (dark line) live a mean of 33% longer than their wild-type littermates (756 ± 46 compared with 568 ± 49 days; P < 0.01, t-test). Kaplan-Meier analysis of survival revealed a delayed onset of senescence in Igf1r+/– mice compared with wild type (P < .001, Cox’s test). Male data (not shown) shows that Igf1r+/– males live 15.9% longer than wild-type littermates (679 ± 80 compared with 585 ± 69 days); however, this is statistically nonsignificant. (After Holzenberger et al. 2003.)

A 12–14% increase in mean and maximum longevity is also observed in mutant mice whose insulin receptor (InR) levels are reduced by 50% but only and specifically in their adipose tissue (FIRKO or fat-specific insulin-receptor knockouts; Bluher et al. 2002, 2003). Such heterozygous InR -/+ animals are not dwarves but beginning at 3 months of age show a 15–25% decrease in body weight coupled with a 50–70% decrease in total fat mass throughout the rest of their life, even when raised under AL conditions. The FIRKO animals show no reduction in food intake relative to controls, implying that their energy expenditure is greater. In contrast to the IGF-1R mutants discussed above, both heterozygous FIRKO males and females appear healthy, show no signs of impaired glucose homeostasis, and both sexes live long due to a delayed onset of senescence. This study seems to suggest that adipose tissue is an important component of the neuroendocrine system regulating longevity and that an adipose tissue-specific restriction of the ISP may be sufficient to delay senescence and extend longevity without necessarily generating undesirable genderspecific effects (Bluher et al. 2002, 2003).

Mutants that affect the development of the mouse anterior pituitary during fetal life have also yielded informative results (Carter et al. 2002). The Snell dwarf mouse is homozygous for a mutation at the pituitary-1 (pit-1) locus, and the Ames dwarf mouse is homozygous at the “prophet of pit-1” (prop-1) locus. Both mutations inhibit the development of pituitary cells responsible for producing GH, prolactin, and thyroid-stimulating hormone, and so both types of mutants are deficient in all three of these hormones. Both mutants demonstrate increased mean and maximum longevity (50% in males, 64% in females) characterized by both a delayed occurrence and a diminished incidence of fatal neoplastic diseases (Ikeno et al. 2003). This extended longevity, however, is accompanied by other changes that make it difficult to precisely determine if the increased life span is a direct result of the lower GH/IGF-1 levels or of some other related neuroendocrine effect. In addition, there are significant differences in the manner in which different tissues respond to insulin, for example, and those subtleties are not considered here (Dominici et al. 2003). Table 7.9 summarizes the characteristics

7.6 Mice and Other Mammals

305

Table 7.9 Comparison of Characteristics of Long-lived Dwarf and Growth Hormone Knockout (GHR-KO) Mice with the Characteristics of Wild-Type Animals Subjected to Caloric Restriction (CR) WT-CR

Dwarf (Prop1df & Pitdw)

GHR-KO

Reduced Reduced Increased

Reduced Reduced Increased

Greatly reduced Modestly reduced Increased

Reduceda Reduced Reduced

Absent Greatly reduced Reduced

Elevatedb Greatly reduced Reduced

Plasma thyroid hormones Body core temperature

Reduced Reduced

Greatly reduced Reduced

Reduced Slightly reduced

Reproduction Sexual maturation Fertility

Delayed Reduced

Delayed Severely suppressedc

Delayed Reduced

Elevated Reduced

Normald Normale

Normald no data

Characteristic Glucose regulation Plasma insulin Plasma glucose Insulin sensitivity Somatropic axis Plasma GH Plasma IGF-1 Body size Thyroid function and metabolism

Glucocorticoids and adiposity Plasma corticosterone Percent body fat Source: data from Bartke and Turyn (2001). Notes: GH, growth hormone, IGF-1, insulinlike growth factor-1. aPulsatile

GH secretion is preserved in long-term CR rats.

bElevated

GH levels in these animals do not imply increased GH signaling because GH receptors are absent.

cFemale

dwarf mice are infertile due to primary prolactin deficiency and the resulting luteal failure.

dPlasma

corticosterone is normal in most comparisons, but some age-, sex- or time of day-related exceptions occur.

eModest

but statistically significant reductions in adiposity detected in older dwarf, along with reduction in plasma leptin levels.

of the two mutants relative to those of wild-type and GHR knockout animals raised under CR conditions. All three of these long-lived rodents share the same sorts of alterations to their glucose regulation, somatropic axis, and thyroid function. The comparison suggests that neither elevated glucocorticoid levels nor decreased body fat are required for the expression of extended longevity. The deleterious reproductive effects shared among these three models are not found in either the IGF-1R null mice or the FIRKO mice described above; thus these pathologies may not be requisite side effects of ISP effects on longevity. It cannot be assumed that the pituitary mutations in dwarf mice are exactly mimicking what happens in normal mice on CR. When Ames dwarf mice are subjected to 30% CR, they live significantly longer than do AL-fed dwarf mice

or CR-treated wild-type mice (Bartke et al. 2001). This effect is seen in both sexes. Although both the dwarf and CR effects probably depend on a down-regulation of the IGF-1 levels, it appears that the Ames mouse uses some nonshared pathway or mechanism that allows it to live longer. The survival plots of these mice show an interesting disparity. A comparison of the AL and CR wild type survival shows that senescence starts at the same time in both cohorts but proceeds slower in the CR set. A similar observation can be made when comparing the AL and CR dwarf cohorts. However, comparing either set of wild-type mice with either set of dwarf mice shows that senescence is delayed in the mutants. This led Bartke et al. (2001) to suggest that CR decelerates aging, while the prop-1 mutant locus delays onset of aging. Note, however, that a similar

306 Chapter 7 Genetic Determinants of Longevity in Animal Models delayed onset is observed in other experiments using CR-treated wild-type mice (e.g., figure 6.2). Thus, the different survival effect of CR on dwarf mice might be a private effect only, possibly stemming from developmental effects of hormone deficiencies on the gastrointestinal tract (Sonntag et al. 2001) or from other possible mechanisms (e.g., hypothyroidism, low metabolic rate, high reactive oxygen species scavenging levels, altered methionine metabolism) discussed by Bartke (2000) or Uthus et al. (2003). The effect of CR in dwarf mice might be better investigated using age-specific mortality data (e.g., see figure 2.14). Conversely, transgenic mice that overexpress GH have been shown to have a high frequency of various pathologies and a reduced life span (Brown-Borg et al., 1996; Steger et al., 1993). The data discussed above shows that genetic and nutritional techniques that underexpress the GH/ IGF-1 axis are associated with extended longevity. Longevity in yeasts, nematodes, fruit flies, and rodents is extended by a down-regulation of one or more components of the ISP. This pathway is a highly conserved and public mechanism of longevity regulation and seems to operate in the same fundamental manner in all organisms. Human centenarians are characterized by having a low degree of insulin resistance and reduced levels of oxidative stress (Barbieri et al. 2003), which implies that their GH/IGF-1 axis might be downregulated and their antioxidant defenses might be elevated (also see figure 3.21). One of the main benefits of using a comparative approach to understanding the biology of aging is that it allows us to develop plausible hypotheses regarding human aging based on robust comparative data. There are at least 250,000 sites on the Web that advertise GH for sale, ostensibly as an easy way to improve health and slow aging. Many of the sites claim to be supported by science, and they cite the 1990 study by Rudman et al. What they fail to cite is the fact that the preliminary data in that paper supporting a beneficial effect of GH in older men were reversed by follow-up studies which indicating that GH has significant deleterious side effects. And so supplement purveyors continue to recommend that users may

improve their health by up-regulating the GH/ IGF axis. This is based on a highly selective and erroneous interpretation of the data. Neuroendocrine aging is a real phenomenon (see chapter 12), and perhaps some sort of supplementation to restore age-depleted levels of certain factors might be beneficial. But supplementation might also be severely detrimental as well. Many of the advertisements tout GH as a means of altering body composition in elderly people. The existing data are consistent with this sales line, although the ads studiously ignore the documented deleterious side effects. Other ads tout GH as an anti-aging hormone, and these vendors have, I think, misinterpreted the data (Bartke 2001). Given its careless marketing, inappropriate users of GH may well be generating future pathologies for themselves as well as a surplus of prematurely ill individuals for society in the near future. The axiom that the buyer should beware certainly applies in this instance. Anyone who has read this chapter and still buys GH as an anti-aging drug does not understand the material.

7.6.4 Metabolic Control of Longevity via Nuclear–Mitochondrial Interaction Evidence for an important nuclear–mitochondrial interaction has been obtained by observing the effects of mitochondria-generated reactive oxygen species on the stability of the nuclear genome (Samper et al. 2003). Fibroblasts were obtained from wild-type mice and from mice heterozygous or homozygous for a null version of the mitochondrial MnSOD gene. Wild-type and heterozygous cells grew nicely when raised under standard cellculture conditons, but the homozygous null cells grew slowly if at all. These null cells had higher rates of cell death than did wild-type cells and had significantly higher levels of chromosomal aberrations, including double-strand breaks and translocations. The heterozygous MnSOD knockout mouse has a normal life span but increased morbidity, including increased rates of cancer with age (A. Richardson, cited in Samper et al. 2003), and the homozygous null mutant is extraordinarily short lived and dies as a result of cell death in

7.6 Mice and Other Mammals

the brain and heart (Melov et al. 1999). Thus, the stability of the nuclear genome is dependent on the MnSOD-dependent redox state of the mitochondria. Becuase superoxide (the toxic ROS generated in the absence of MnSOD) is tightly localized in the mitochondria, this implies that there is a cascade of ROS effects connecting the two organelles, allowing the mitochondria to indirectly affect genomic stability and longevity. The data discussed above show that increasing the ROS levels within the mitochondria by decreasing the MnSOD activity leads to higher levels of morbidity and a shorter lifespan. But does the reverse situation hold? Will decreasing the mitochondrial ROS levels result in increasing the life span of a mammal, as suggested by the mitochondrial free radical theory of aging (see chapter 11)? These questions have been answered in the affirmative, as evidenced by an experiment that created transgenic mice overexpressing the human catalase gene in their mitochondria (Schriner et al. 2005). Catalase is normally expressed in the cytoplasm only, and works in tandem with CuZnSOD to reduce harmful hydroxyl (OH–) radicals to harmless water and oxygen. Its targeting to the mitochondria would allow it to carry out the same reactions in cooperation with the MnSOD enzymes. In accordance with predictions, transgenic mice expressing catalase in their heart, muscle, and/or brain showed significantly lower levels of both oxidative damage to mitochondrial components and tissue-specific pathologies. They also had an increased (~20%) life span, which may have arisen from a delay in the onset of aging. Transgenic mice in which the human catalase gene was targeted to the nucleus or to the peroxisome did not exhibit these effects; thus, these data demonstrate that mitochondrial ROS levels may be an important limiting factor in determining mammalian longevity.

7.6.5 Stress Resistance and Extended Longevity The discussion of invertebrate models clearly showed that increased or decreased resistance to various stressors plays a major role in extending

307

or shortening the life span. These factors are also important in modulating mammalian longevity (Finkel and Holbrook 2000). For example, when one tests primary fibroblast cells taken from eight different mammalian species for their in vitro resistance to five different stressors, there appears to be a robust correlation between the cell type’s measured resistance to stress and the maximum life span characteristic of each species (Kapahi et al. 1999). Because the maximum life span of a species is inversely proportional to its susceptibility to oxidative stress (Agarwal and Sohal 1996), then it is logical to conclude that stress resistance is an important modulator of longevity and aging. But the actual data suggest that some stressors are more important than others, and some antistress genes are more important than others. These differences are magnified by the greater complexity, redundancy, and/or specificity of regulatory mechanisms in mammals relative to invertebrates. Simply increasing the level of antioxidant activity in a mouse by inserting an extra copy of the CuZnSOD (SOD1) gene, for example, did not lead to an increased life span, even though the enzyme was expressed (Huang et al. 2000) and even though it did decrease the normal agedependent levels of oxidative damage observed in the brain (Cardozo-Pelaez et al. 1998). The simplest way to explain this unexpected deviation from the fly data (see table 7.4) is to assume that the levels of cytoplasmic ROS in the mouse cell do cause damage but are not limiting factors on longevity, and so increasing the defenses against them would have little effect on life span. Mice homozygous for a disrupted the CuZnSOD gene do not have a decreased life span (Muller 2001). Although they appear normal, these mice covertly suffer from injuries arising from increased sensitivity to a variety of oxidants. Females are semisterile and have ovarian pathologies. Both sexes suffer from sensory (hearing) and motor neuron dysfunctions. These data suggest that the reduction in CuZnSOD activity exposes the animal to increased but not lethal levels of oxidative stress. Only those tissues in which oxidative stress exceeds some threshold display stress-induced pathologies. Of course, if we

308 Chapter 7 Genetic Determinants of Longevity in Animal Models expose the homozygous mutant to some highstress environment (e.g., 100% oxygen for their entire life, etc), then we should expect the animal to display more, and possibly life-shortening, stress-induced pathologies. A very different effect is seen when the MnSOD (or SOD2) gene is deleted (Muller 2001). Mice heterozygous for the deletion have a more or less normal life span under standard conditions but show signs of mitochondrial abnormalities and increased oxidative stress when compared to wild-type animals (Williams et al. 1998). In contrast, the homozygous MnSOD knockout mice die as neonates and exhibit cardiomyopathy, fatty liver, and neurodegeneration. The exact life span can vary, depending on the mouse strain involved, but it ranges only from 3–15 days after birth. Overexpressing the CuZnSOD gene fails to rescue the MnSOD homozygotes, in agreement with the idea that the two genes operate in different cellular compartments (i.e., cytoplasm and mitochondria, respectively). An interesting phenomenon occurred when an attempt was made to rescue these animals by treating them with chemicals that mimic the action of SOD and catalase and protect cells from oxidative stress (Melov et al. 1998). Mimetics that cannot cross the blood– brain barrier protected the hearts and livers from stress-related damage, but the animals eventually succumbed to paralysis stemming from extensive degeneration of the cortex and brain stem, which resembled that found in various mitochondrial diseases. Thus, the defect resides within the mitochondria of each cell, and there is a hierarchy of tissues sensitive to the reactive oxygen species damage produced by the defective organelles (heart = liver > cortex = brain stem). Mimetics that are able to cross the blood–brain barrier were effective in signficantly increasing the animals’ life span (Melov et al. 2001) and in significantly reducing the loss of neurons due to oxidative stress (Hinerfeld et al. 2004). The fact that deletion of the mitochondrial SOD gene causes so much more damage than the cytoplasmic SOD gene supports the mitochondrial free-radical theory of aging, which postulates that the mitochondrion is the source of most of the reactive oxygen spe-

cies responsible for the age-related damage to the cell (de Grey 1999; see chapter 11). Of course, the same conclusion was drawn from the experiment discussed above in which transgenic mice overexpressing human catalase in their mitochondria had significantly decreased oxidative damage and increased longevity (Schriner et al. 2005). These same mimetics can also inhibit the normal age-related learning and memory deficits observed between 8 and 11 months in C57BL/ 6N mice (Liu et al. 2003). Mice treated with the mimetic during this period showed neither the typical learning and memory deficits nor the higher levels of oxidative damage to the brain observed in control animals. This experiment supports a role for oxidative stress in age-related cognitive impairment, and it basically confirms results drawn from earlier studies. One such study showed a correlation between cognitive decline in humans and plasma levels of chemicals indicative of systemic oxidative stress (Berr et al. 2000), and another showed that oxidative stess reduced the performance of rats on cognitive tests relative to controls (Shukitt-Hale 1999). Mimetics may have a useful therapeutic role in both rodents and humans. The fact that mitochondria from female rodents exhibit higher antioxidant gene expression and lower oxidative damage than do those from males suggests that sex-specific differences in mitochondrial ROS production might underlie the gender difference in mortality (see table 8.5). Oxidative stress affects both quality and quantity of life in mice and humans. In addition to the two SOD enzymes discussed above, there are other enzymes and molecules that suppress the net formation of ROS by either quenching active oxygen or by decomposing peroxides. These are discussed in more detail in chapter 10. Mice have a more complex regulatory system than do invertebrates, and this applies to their oxidative stress resistance as well. There are mechanisms that regulate stress response, and a mutation in one such system is one of the few single genes known to increase life span in mammals. The p66shc protein is a cytoplasmic signaltransducer molecule involved in transmitting

7.6 Mice and Other Mammals

signals from various receptors to the Ras protein, which plays a regulatory role in many cellular systems. Mice homozygous for knockout mutation of this gene (p66shc–/–) are 40% more resistant to oxidative stress and live about 30% longer than do the controls (figure 7.34; Migliaccio et al. 1999). This occurs because p66shc interacts in an inhibitory manner with the mammalian forkhead (FOXO3a) transcription factor, which is an important component of the ISP and homologous to daf-16 (see figure 7.32). The complete absence of the p66shc protein in the homozygous mutant allows the forkhead factor to become active (Nemoto and Finkel 2002; Seydel 2002). It consequently activates the animal’s intrinsic antioxidant defense systems in the manner described earlier in this chapter and allows its cells to express significant resistance to oxidative stress by inducing both antioxidant genes and DNA repair genes. The p66shc protein has another function as well. In wild-type animals, cells severely damaged by oxidative stress or radiation damage are signaled to enter apoptosis (see chapter 11) and destroy themselves, presumably as a defense against these cells becoming cancerous and even-

309

tually killing the organism. In homozygous mutant animals, the level of apoptosis among oxidatively stressed cells (but not irradiated cells) decreases by about 80%, indicating that the p66shc protein is part of a signaling mechanism specific for certain types of damaged cells. The absence of p66shc not only makes the cells resistant to oxidative stress, but it also interrupts the normal signal which would instruct them to kill themselves. As a result, the mutant cells survive and are significantly more resistant to developing chronic diseases that stem from systemic oxidative stress, such as atherosclerosis (Napoli et al. 2003). The homozygous mutant animal lives longer because its cells are resistant to oxidative stress and perhaps because its tissues do not lose excessive numbers of cells. But longer life comes with the risk of developing cancers and other chronic diseases. The normal mouse avoids the problem of late-life chronic disease by killing its damaged cells which might give rise to such pathologies. Unlike invertebrate tissues, most mammalian tissues are mitotic, and presumably the destroyed cells might be replaced via mitogenic signals to appropriate progenitor cells in the

Cumulative survival

1.0 0.9 0.8 0.7 0.6 0.5 0.4 WT

0.3

p66+/-/p66

0.2 0.1

500

600

700

800

900

1,000

1,100

Time (days) Figure 7.34 Cumulative survival of wild-type, p66shc+/–, and p66shc–/– mice. Notice the apparent dose response. (After Migliaccio et al. 1999.)

310 Chapter 7 Genetic Determinants of Longevity in Animal Models tissue. The p66shc protein acts as a negative regulator of oxidative stress (and longevity) but a positive regulator of apoptosis for potential cancer cells. It is likely that the inability of late-life chronic diseases to evoke a strong early-life selection against them explains the continued existence of this negative regulator (see chapter 4). Mammals thus mark the appearance of a regulatory system that coordinates oxidative stress and cell death directly and longevity indirectly. It is common knowledge that older animals are less resistant to various stressors than are younger animals of the same genotype. This observation implies that the older animal’s response to stress must be impaired in some way. In the case of the mouse, it has been shown that the hearts of young (5 months), middle aged (15 months) and old (25 months) animals differ in their transcriptional response when the animal is given a systemic overload of paraquat-induced oxidative stress (Edwards et al. 2003). As shown in table 7.10, about 3% of the assayed genes respond to paraquat at any age, but there are significant age-related differences in the transcriptional responses. The most obvious difference is that decreasing numbers of the immediate early genes are activated as age increases. These genes are probably important, for their early activation suggests they play a regulatory role in setting up cascades of regulated gene expression. Their inability to be induced in older animals suggests that important parts of the gene response to oxidative stress might be ineffective in older animals. Even the common genes show altered transcriptional patterns as age increases. Thus, there is an age-related impairment of specific inducible

pathways in the aging animal’s response to oxidative stress. A given level of oxidative stress will thus have more deleterious effects in an older animal than in a younger animal. As oxidative stress levels rise, the effect on the animal comes not only from ROS damage to cellular components, but it also comes from oxidant effects on signal transduction systems (Martindale and Holbrook 2002). For example, mammalian fibroblast cells in culture are stimulated to divide by very low doses of hydrogen peroxide (~10 mM), stimulated to enter a temporary growth arrest at higher concentrations (~125 mM), stimulated to enter a permanent growth-arrest phase at still higher levels (~300 mM), and stimulated to die via apoptosis at titers above about 500 mM (Yoon et al. 2002). The low-level stimulatory effect likely arises because oxidants at this dose activate growth-factor receptors and thus mimic the action of natural ligands. How could an ROS do this? Many growth factor receptors are activated by being phosphorylated by specific kinases and inactivated by being dephosphorylated by various phosphatase enzymes. In other words, the presence or absence of a phosphate group serves as the cell’s version of an on or off switch for many of its enzymatic components. Low ROS levels inhibit phosphatases, thus leaving the already phosphorylated receptors activated and thereby stimulating some normal mitogenic response. Intermediate levels could affect the JNK pathway, for example, by inhibiting its inhibitor and thus activating it. High levels of ROS could initiate apoptosis via the p66shcbased mechanism.

Table 7.10 Aging Degrades the Animal’s Ability to Mount an Effective Response to Oxidative Stress Genes PQ-responsive genes Common Common genes with altered expression Immediate early genes expressed at each age

5 months

15 months

25 months

249 55 ↑4 12

298 55 ↓7 7

256 55 ↑ 2, ↓8 5

Source: data from Edwards et al. (2003). Note: PQ, paraquat; common = no. of genes expressed in response to PQ at all ages; immediate early genes = expressed in response to PQ within 1 hour of treatment.

7.6 Mice and Other Mammals

There are a very large number of interacting signal transduction pathways, the net effect of which is to turn on various transcription factors (see, e.g., figure 12.8). After all, much of the genetic control of longevity and aging is actually exerted by changes in the cell’s protein composition which then alters the quality or quantity of intra- and intercellular signaling. Signal transduction systems known to respond to ROS signals include the ISP, the nuclear factor-kappa B (NFkB) signaling system, the extracellular signalregulated kinase (ERK) system, the c-JUN amino terminal kinase (JNK system), the p38 mitogenactivated protein kinase (p38 MAPK) system, the heat-shock response factors (HSF), and the p53 cell-cycle checkpoint system (Finkel and Holbrook 2000). These systems all interact with one another. One possible advantage of a signaling network with multiple intersecting pathways is to allow a coherent response to numerous potentially conflicting signals. Such an outcome could come about by the cell summing up the different inputs and adopting the alternative with the most “votes”—an example of quantitative factors interacting to bring about a qualitative outcome. Given the large num-

ber of different genes positively or negatively regulated by these and other signal transduction systems and their transcription factors, it is not surprising that changing the ROS levels within a cell gives rise to changing patterns of gene expression. These different patterns of gene expression give rise to different outcomes, and it this ability to either protect or destroy the cells that connects signal transduction with stress resistance and aging (table 7.11) In this context, it is interesting that naturally occurring chemicals have a protective effect against oxidative and chemical stresses. For example, the polyphenols and isothiocyanates present in green tea or blueberries activate the JNK1 and ERK2 pathways, inducing an array of detoxification and antioxidant genes that presumably have some relationship to the beneficial effects reported by users of such agents (Ouwor and Kong 2002). The important lesson to be derived from such reports is that the variety of proteins acting as scavengers of superoxide and peroxides do not act independently but rather are bound by these multiple and overlapping signal transduction pathways into an antioxidant network

Table 7.11 Signaling Pathways activated by Oxidative Stress and Their Ultimate Cellular Outcomes Cellular outcome Signaling pathways p53 NFkB HSF1 PI3K/Akt ERK JNK p38 PLC JAK/STAT c-Abl

311

Enhanced survival

Cell death

+ + +++ +++ +++ ++ + +++ +++ +

+++ +++ – – ++ +++ + – – +++

Source: after Martindale and Holbrook (2002). Note: This table summarizes the prevailing evidence regarding the activation of these major pathways and how their downstream targets either enhance cell survival or promote cell death specifically in response to oxidant injury. A minus (–) means there is minimal or no evidence that this pathway influences this outcome; +, some evidence for this outcome; ++, much evidence that this pathway promotes this outcome; +++, predominant outcome for this pathway. See text for a description of the pathways.

312 Chapter 7 Genetic Determinants of Longevity in Animal Models that is activated or repressed in an integrated manner by changing ROS levels as well as by a variety of antioxidants. The antioxidant network is only one member of a larger set, for other stressors likely induce their own sort of network response. The response of these antistress networks must be considered as constituting the major portion of the genetic mechanisms underlying the animal’s resistance to stress. The age-related change in the animals’ transcriptional responses to stress (see table 7.10) not only stems from the age-related changes in the sensitivity of these signal transduction pathways but also assure that the aged animal will mount a progressively less effective defense against ROS and other stressors. There are stressors other than oxidative stress, and there are resistance genes other than antioxidant genes (see table 7.3). Many stressors induce the heat shock proteins (HSPs), a family of conserved proteins with diverse roles in the cell. Their most well-known role is to act as molecular chaperones and assist unstressed cells with the proper folding, localization, and degradation of their proteins. In stressed cells, the HSPs assist in refolding proteins and in otherwise protecting against protein damage (see figure 10.9). There is an age-related decrease in rats’ ability to induce hsp70, and this probably has deleterious effects on their ability to survive in a stressful environment. This decreased ability to induce hsp70 stems from conformational changes in the heat shock transcription factor (HSF1), and these altered proteins are increasingly incapable of activating the gene (Heydari et al. 2000). This suggests that the basal levels of HSPs are not sufficient to protect the transcription factor from damage. However, if one induces hsp70 in cultured human fibroblast cells in vitro by exposing them to repeated mild heat shocks throughout their cellular life span, a beneficial hormetic response results (see chapter 10 for details). Exposure to heat shock significantly increases the level of hsp70 expressed in both early and late passage cells; it concomitantly led to significant increases in resistance to a variety of different stressors in both populations of cells (Fonager et al. 2002). Thus, adaptively inducing high levels of HSPs early in life protects these cells by preventing or delaying functional damage.

There is a strong tissue/cell specificity in the response to HSPs that must also be considered. Skeletal muscle of aged rats is able to express a significant exercise-induced accumulation of HSP70 in the highly oxidative skeletal muscles (e.g., soleus), but only a blunted response in the fast-twitch muscles (e.g., white gastrocnemius; Naito et al. 2001). Overexpression of hsp70 in muscle of old mice protects against the exerciseinduced specific force deficits seen in the muscles of old control mice, so there are real consequences to this variable hsp70 expression (McArdle et al. 2004). Mammals have at least three different HSF isoforms. Perhaps there might be a correlation between the cell’s HSF isoform present and the cell-specific hsp70 response to exercise. It is likely that the depth and breadth of the cell’s stress resistance networks determines when it passes from the health span into the senescent span. A model for this important transition is presented in chapter 9, and it should be viewed as another example of the important role that stress resistance plays in modulating longevity.

7.6.6 Genetic Stability As was the case with Drosophila, genetic stability may be viewed as having two components; structural integrity of the DNA and epigenetic alterations of chromatin and nucleosome structure. Somatic mutations have long been suspected of playing a critical role in the aging process. I discuss this in more detail in chapter 10. The available data allow us to conclude that even though many DNA alterations increase with age, few if any such alterations routinely reach a critical threshold in normal animals sufficient to explain the observed aging changes. Normal animals do not have such an unstable genome that it threatens their ability to function. However, recent evidence suggests that as a result of unequal crossing-over, many people have variable numbers of different protein coding genes (Leslie 2004). The resulting duplications (and the ensuing overexpressions) might, in extreme cases, provoke the onset of disease. This large-scale genomic variation can be viewed as a particular instance of genetic instability. This in-

7.6 Mice and Other Mammals

stability may not affect longevity, but it is known to affect morbidity. Of course, animals with certain DNA repair mutations do have an unstable genome as measured by their higher mutation level and their usually shortened life span, whereas animals with multiple DNA repair defects are always shortlived (e.g., Vogel et al. 1999; also see chapter 10). Therefore, the maintenance of genomic integrity is a major factor in longevity (Hasty et al. 2003). Part of the reason that normal mice do not succumb to their somatic mutations may be that mutated cells are routinely destroyed by apoptosis. Newly mutated cells are usually damaged by some form of oxidative stress. As such, they may be detected by the p66shc protein and dispatched via apoptosis. Alternatively, they may be detected by the p53 cell-cycle checkpoint or some other signal transduction protein and similarily instructed to kill themselves (table 7.11). In any event, genomic instability does not appear to play a major role in regulating the longevity and aging of normal mammals, in large part because the normal animals have a complete complement of the genes necessary to maintain genomic integrity. Epigentic modifications of the genome are under intensive investigation in mice, but most of the available data concern development and do not seem pertinent to our discussion. However, the existence of variable somatic cell, epigenetically controlled phenotypes (such as that involving the agouti locus of mice) demonstrates that the DNA and histone acetylation/methylation processes that occur in flies also occur in mice. So far there is no information suggesting that drugs affecting histone deacetylation can affect mammalian longevity, as was shown in Drosophila (see figure 7.24), but the available data suggest that the mouse genome can be affected by an appropriate epigenetic stimuli.

7.6.7 Reproductive Effects Sexual maturity in mammals is brought about by alterations in sensitivity to and levels of various trophic hormones, which eventually result in increased levels of estrogen or testosterone and the

313

subsequent differentiation of gender-specific secondary sexual characteristics. During this process, thyroid hormones and growth hormone are activated, and their basal levels increase. The combined actions of high levels of insulin, IGF-1, sex steroids, and thyroid hormones promote growth and reproduction at the expense of anti-aging mechanisms. The Gompertz data show that humans are at their maximum health status just prior to puberty, when sex steroids are not at their highest level (figure 2.14). The onset of senescent changes in many normal adult functions are somehow correlated with the postreproductive changes in the hormonal milieu. Some interventions aim to maintain the hormonal milieu via hormone replacement therapy. Demographic and experimental data are consistent with the hypothesis that maintenance of the hormonal milieu is conducive to extended longevity. First, contemporary women who reported an early onset of natural menopause (<44 years) had significantly elevated mortality relative to those who experienced the onset of menopause at ages 50–54 years (Snowden et al. 1989; but recall that age at menopause is affected by maternal nutrition, as discussed in chapter 3). Second, a historical study of French-Canadian women of the 17th and 18th centuries, when large families were desired, showed that increased female fertility is statistically associated with a decreased postreproductive mortality, and thus longer life spans, relative to women in the cohort who were not as fertile or long lived (H.-G. Muller et al. 2002). Both analyses point to a protective effect of potential or actual late births (possibly reflecting the number of ovarian follicles) for the mother. These are both correlative studies and say nothing of causation, but a recent mouse study supports this correlation. If female mice were ovariectomized prior to puberty and then at much later ages had young ovaries transplanted into them, they showed active ovarian cycles at an age when the controls were postmenopausal. They also exhibited a significant (~38%) increase in life expectancy relative to the intact controls (Cargill et al. 2003). The young ovary did not change the maximum longevity but significantly delayed mortality to a late age, thus

314 Chapter 7 Genetic Determinants of Longevity in Animal Models accounting for the higher life expectancy. Even the control is informative, for the absence of a functional ovary accelerated mortality to earlier ages than in the intact control and decreased life expectancy by about 10%. The transplant of older ovaries into the experimental mice was found to yield smaller increases in life expectancy, showing that remaining life span is proportional to the relative youth of the ovary. Thus both the human and mouse studies are consistent with the idea that prolongation of ovarian function extends longevity in female mammals. Is there a mechanism that might play a role in regulating ovarian function, and thus life span? The ISP may play a role in regulating the age duration of oogenesis in mice. The transcription factor forkhead-like 1 (FKHRL1, also termed FOXO3a) of the mouse is regulated by the ISP (see figure 7.32) and is homologous to the Drosophila forkhead (dFOXO) and nematode daf-16 transcription factors. Homozygous inactivation of this FKHRL1 gene in female mice leads to a rapid activation and mobilization of oocytes, followed by oocyte depletion and early-life ovarian failure (Castrillon et al. 2003). Female sterility is also induced by homozygous inactivation of the IRS-2 component of the ISP (see figure 7.32; Burks et al. 2000). In contrast, homozygous inactivation of the pro-apoptosis Bax gene in female mice leads to a threefold increase in the number of ovarian follicles and a significant prolongation of the ovary’s functional life span (Perez et al. 1999). The studies described above suggest the existence of a multigene system, including but not limited to the apoptosis-inducing pathways and the ISP, regulating the rate of follicular activation and the rate of follicular apoptosis in mammalian ovaries. Mouse ovaries are reported to contain mitotically active germline stem cells that replenish and sustain follicle production in the adult ovary (Hubner et al. 2003; J. Johnson et al. 2004). If present in humans, this would constitute a third component of this multipathway system regulating ovarian function. The age at ovarian failure (menopause) in human females may possibly be the vectored outcome of such a regulatory system, such that some balanced ratio of stem cell activity, activation rates, and apoptotic levels might

synergistically spare the oocyte supply and extend functional life span. The possibility that the ISP plays some sort of role in modulating this system ties these observations to our current understanding of the major conserved mechanisms regulating longevity in animals. Mammals probably have a more complex version of the mechanisms schematically depicted in figure 7.25, in part because of this interacting apoptosis pathway. Two other studies tried to correlate the relationship between actual births and female longevity in the historical records of the British and European aristocracies, but reached opposing conclusions, possibly due to incomplete data (Gavrilov and Gavrilova 1999, 2002; Gavrilova et al. 2004; Westendorp and Kirkwood, 1998, 1999, 2001). It may not be the number of births that is important but rather the length of time the ovary is functional and active. The recent reports that estrogen appears to act as an inhibitor of mitochondrially derived reactive oxygen species production (Borras et al. 2003) and that the mitochondria have an estrogen receptor (ERb; Yang et al. 2004) suggests a mechanism by which estrogen might affect longevity. This hypothesis is strengthened by the finding that estrogen treatment protects neurons subsequently exposed to various toxins (Nilsen and Brinton 2004). Estrogen protects via two mechanisms. It increases the level of anti-apoptotic proteins (Bcl-2 and BclxL), and it increases mitochondrial respiratory efficiency and so decreases the oxidative load within the mitochondria. Taken together, the actions contribute to the estrogen-treated mitochondria shifting to a proactive defense against oxidative damage, thus preventing initiation of the deleterious mitochondrial spiral (see chapter 11 for more details). It will be interesting to determine if an estrogenlike molecule may be synthesized that would induce these mitochondrial protective effects without any feminizing effects.

7.6.8 Patterns of Senescence The data in table 7.5 and the information contained in chapter 5 make it clear that mammalian tissues do not age together or synchronously,

7.7 Summary

but rather follow their own characteristic senescent pathways with more or less characteristic types of functional failures in each. Nonetheless, mortality increases logarithmically with advancing adult age (see figure 2.14). Yet this increase in humans hides significant age-associated changes in the cause of death. The mortality rise in middle age has been attributed mostly to specific chronic diseases (e.g., cancer, heart attack) that develop prematurely in high-risk individuals, while the mortality increase in old age stems mostly from the senescent processes (e.g., senile dementia, heart failure) that stem from the longterm decrease in the body’s defense mechanisms and make most people more vulnerable to various extrinsic effects (Horiuchi et al. 2003). Thus, there are significant differences in the types of diseases/systemic failures that precipitate mortality at different ages. Furthermore, the morbidity found in older humans stems from the effects of potentially mortal diseases (e.g., cancer) as well as those that are rarely, if ever, fatal (e.g., arthritis). The percentage of the older population having one of these mortal conditions increases only slightly with age between 70 and 90+ years, while the likelihood of having one of the morbid conditions increases significantly during this same age range (Crimmins 2001). Thus, the particular senescent path that different people trace from vigorous adult to frail elder is the outcome of their individual intrinsic weaknesses as well as the extrinsic risks of their particular environment. The improvement in the general environment that has occurred during the last century has effectively lowered the levels of many extrinsic risks and thus promoted longevity. Additionally, many older individuals voluntary alter or restrict their own environment and thereby lower particular pertinent risks (Robine 2001). The wide individuality of senescent patterns in humans owes much to the fact that mortality and morbidity reflect different aspects of senescence, follow different age-related changes in incidence, and can be differentially modulated by environmental alterations and improvements. Finally, the role of development and chance cannot be underestimated. Figure 3.14 presents

315

some of the known developmental effects that affect the aging rate and the senescence pattern. More information on these effects is now coming to light. For example, early body weight in mice predicts future longevity such that big mice die young (R. A. Miller et al. 2002c). Parents who give their small, young children GH may be altering their future senescent path in an unfavorable direction. Chance plays a large and once unappreciated role in aging (Finch and Kirkwood 2000). One example of the role of chance should suffice. The signals that instruct some developing structure to stop cell division arise from the tissue itself and may reflect the effect of increasing tissue mass on the strength of the mitogenic signals. There is likely some variability as to when the stop signal is effectively received. Should the signal be received a bit late, then that tissue will undergo an additional round of cell division. This may not seem so important, but such an extra division could result in as much as a doubling of the cell number in that tissue. Should the increased cell numbers be maintained during development, then the resulting adult might have increased numbers of immune system cells, or neurons, or myocytes, or perhaps increased number of stem cells in different tissues. Increased cell numbers, obtained entirely by chance, may significantly increase the organism’s abilities to resist various stressors by known mechanisms. This hypothesis is compatible with reliability theory explanations of biological aging (Gavrilov and Gavrilova 2004).

7.7 Summary Having inspected all the individual “trees,” it is now time to step back and see the “forest.” Table 7.12 is designed to provide a summary of the main mechanisms, their evolutionary connections, and the different ways in which model organisms regulate and modulate longevity and senescence. As mentioned in the beginning of the chapter, it is helpful to scan this table prior to, during, and after reading this chapter. It is a tourist’s guide to the major aging pathways.

316 Chapter 7 Genetic Determinants of Longevity in Animal Models Table 7.12 Summary Comparison of Main Longevity Pathways in Model Organisms Organism and Pathway

Major genes or pathway

Effect

RAS2, PKA; requires HDAc

Low glucose represses PKA, which activatess Msn2 and 4, repressing pro-senescence pathways and activating stress resistance mechanisms

ISP

Not applicable; CR pathway is forerunner of ISP

—

RR

RAS2, Rtg1, Rtg2, Rtg3

Coordinates nuclear and mitochondrial activities, shifts cell metabolism accordingly

Stress resistance

CuZnSOD, MnSOD, DNA repair genes

RAS2 inhibits Msn2 and Msn4 tfs which repress multiple stress resistance genes. Adaptive effect noted.

Genetic stability

SIR2, DNA checkpoint genes, telomeres

Absence of SIR2 gene induces rDNA circles and genome instability. May be a private mechanism. Longevity requires all DNA repair genes.

Reproductive effects

—

Asymmetric mitotic divisions distribute damaged proteins to mother cell, promotes somatic senescence and shortens replicative life span.

Patterns of senescence

—

Mother cell undergoes senescence but daughter cells stay young.

Yeast Metabolic control CR

Nematode Metabolic control CR

(see ISP data) —

—

ISP

daf2, age-1, daf18, Akt1,2, daf16

Environmental signals activate sensory neuron and then interneuron to secrete insulinlike ligand. This activates ISP in neural cells, which systemically affect body via secondary hormones (see figure 7.11, 7.12).

RR

clk 1,2,3; gro-1

Genes alter mitochondrial–nuclear balance in different ways such that metabolism is shifted to slower rate.

Stress resistance

CuZnSOD, MnSOD, GST, catalase, mimetics heat shock genes, other stress resistance genes

daf16 activates stress resistance genes which are otherwise repressed if ISP is activated. Some genes have adaptive effect; others effective only in stressful environment. Inverse relationship between ADS enzyme levels and cellular oxidative damage levels; direct relationship with longevity in many cases.

Genetic stability

DNA repair genes, p53, others

Large interacting complex of DNA repair genes necessary for longevity. Not deeply investigated yet in this organism. (continued)

7.7 Summary

317

Table 7.12 (continued) Organism and Pathway

Major genes or pathway

Effect

Reproductive effects

ISP–gonad interactions

Germline stem cells send prosenescent signal which inhibits daf-16. This allows a postulated hormone to be released from somatic cells to down-regulate somatic maintenance activities and up-regulate germ cell proliferation, thus setting up a positive feedback. Somatic gonad cells send an antisenescent signal which inhibits daf-2 and thus activates daf-16.

Patterns of senescence

—

Nervous system is maintained throughout life span. Muscles deteriorate beginning in midlife, but this is delayed by age-1 gene activity. Macromolecular synthesis and turnover not tightly regulated in late life. Stochastic factors play an important role.

Fly Metabolic Control CR

indy, rpd3, Sir2

CR slows rate of gene expression change relative to normal feed. Down regulates synthesis, turnover, and reproduction. No consistent upregulation pattern. CR effect is relatively rapid.

ISP

InR, chico, forkhead

ISP affects hormones which indirectly affect longevity by altering stress resistance levels. Forkhead is key transcription factor which increases stress resistances.

RR

—

Cybrids show longevity intermediate between nuclear and mitochondrial donors.

Stress resistance

CuZnSOD, MnSOD, catalase, Gpx, hsp, others

Up-regulation of different ADS genes by various techniques usually leads to increased oxidative/other stress resistance. Likely basis of extended longevity; QTL confirmation of ADS basis in one strain. Tissue-specific protective effects.

Genetic stability

—

Genome appears to be highly stable in normal flies. No evidence for regional or overall genomic dysregulation.

Reproductive effects

ISP, JH, ecdysone, EcR, DTS-3

ISP induces JH synthesis which induces egg production and 20HE synthesis. At high levels, 20HE represses resistance to various stressors and reduces life span. At moderate levels, longevity and stress resistance significantly increase.

Patterns of senescence

—

Various sensory and locomotor abilities do not age in unison. Long-lived mutants do not delay the senescence of all their functions but retain some and lose others as do normal-lived animals. (continued)

318 Chapter 7 Genetic Determinants of Longevity in Animal Models Table 7.12 (continued) Organism and Pathway

Major genes or pathway

Effect

—

Both normal aging and CR induce tissuespecific alterations which can be quite different from each other. Generally, aging changes involve increased ROS production and decreased signal sensitivity/protein turnover, while CR represses some normal aging changes but induces other changes which alter cell metabolism and delay ROSdependent senescence. May involve hormetic effect.

ISP

pit-1, prop-1, GHR-KO, IGF-1R, others

Down-regulation of IGF-1 systemically increases longevity in females but not males, due to sex dimorphic response to glucose homeostasis and oxidative stress response. Adipose-specific down-regulation of InR increases longevity in both sexes. CR effect reversible by insulin.

RR

—

No evidence for or against its occurrence in mammals as yet.

Stress resistance

Various antioxidant defense genes, HSP genes, signal transduction (ST) pathways.

MnSOD and CuZnSOD have different effects on longevity and loss of function. Stress genes’ response diminishes with age, likely due to decaying ST pathways. Decreased oxidative stress apoptosis yield longer life span.

Genetic stability

DNA repair systems, epigenetic modifiers

Age-related hypermethylation many silence tumor-suppressor genes, allowing agerelated rise in cancer. Low methionine diet may enhance epigenetic up-regulation of stress genes.

Reproductive effects

ISP, GH, IGF-1, thyroid hormones, sex steroids

Possible that maintenance of hormonal milieu, perhaps linked to follicular atresia rate, is consistent with an extended longevity.

Patterns of senescence

—

Tissue-specific age-related alterations in function. Systemic treatments (CR, IGF-1, etc.) delay the aging of most tissues in an apparently coordinated manner.

Mouse Metabolic control CR

Note: CR, caloric restriction; PKA, protein kinase A; ISP, insulinlike signaling pathway; PR, rerrograde regulation; ADS, antioxidant defense system; QTL, quantitative trait luci; JH, juvenile hormone; 20HE, 20-hydroxy-ecdysome; HDAc, histone deactetylases; CuZnSOD, copper-zinc superoxide dismutase; MnSOD, manganese superoxide dismutase; GST, glutathime-S-transferase; EcR, ecdysme receptor; tfs, transcription factors; ROS, reactive oxygen species; IGF-1, insulin-like growth factor 1; ST, signal transduction; OX, oxidative.

8

Genetic and Social Aspects of Aging in Humans

8.1 Introduction Much of what we know about the genetic mechanisms that regulate longevity and aging in humans was described in chapter 7 in the context of mammalian aging. But humans live in complex environments, and an examination of human aging should include gene–environment interactions, for they are likely to have major effects on longevity. Many of the earlier human studies tried to determine whether longevity was heritable. Such demographic studies depended on accurate birth and death records, which could be used to generate life tables and test statistical associations between selected groups. There are statistical and methodological objections to both life tables and statistical associations, but until recently they were all we had. Several dozen such studies have been reported in the literature; I discuss only a representative few of them.

8.2 Genealogical Studies Alexander Graham Bell, of telephone fame, did a genealogical analysis of longevity among the descendants of William Hyde of Connecticut, who died in 1681. Bell’s analysis, published in 1918, of the 8798 individuals listed in these family records revealed an excellent correlation between parental and offspring life spans (table 8.1). The results not only showed a substantial (about 40%) difference in the life span of off-

spring whose parents had both died before the age of 60 years as compared to offspring whose parents had died after the age of 80 years, but also showed that the offspring of parents with intermediate life spans also had intermediate life spans. However, it was equally clear that the mean life span of the progeny was always much shorter than that of the parents. There appears to be a high correlation but a low heritability, which seems paradoxical. In part, this phenomenon probably reflects the role of accidents, disease, and other environmentally dependent premature deaths; it certainly emphasizes the fact that not all children of long-lived people will also be long lived. Bell (1918) concluded that the influence of the father on the offspring’s longevity appeared to be somewhat greater than that of the mother. In 1932, Yuan analyzed the longevity records of 7500 individuals who belonged to a single southern Chinese family and who were born and died in the 500-year span between 1365 and

Table 8.1 Parental Age at Death and Average Duration of Life of Offspring Father’s age at death <60 60–80 >80

Mother’s age at death <60 32.8 (n = 35.8 (n = 42.3 (n =

years 128) years 251) years 131)

60–80 33.4 (n = 38.0 (n = 45.5 (n =

years 120) years 328) years 206)

>80 36.3 years (n = 74) 45.0 years (n = 172) 52.7 years (n = 184)

Source: after Bell (1918).

319

320 Chapter 8 Genetic and Social Aspects of Aging in Humans tion of parental age at death revealed a striking linear relationship with the maternal age at death when the father’s age was held constant (figure 8.2). However, the relationship between the age of the offspring and increasing paternal longevity was much less obvious, becoming apparent only with very long-lived fathers. Jalavisto concluded, in contradiction to Bell, that the effect of maternal longevity exceeds that of paternal longevity. Mayer (1991) used a series of genealogies, including 14,549 individual data sets drawn from six white New England families covering 1578 to 1963, to answer two questions: What proportion of the recorded longevity of these individuals could be statistically ascribed to their genetic inheritance? And did the value of this genetic inheritance of longevity change during the 300-year period of the study, or was it a constant? Technically, the genetic inheritance of this or any other

1849. Yuan’s life table analysis also showed a correlation between the life expectancy of sons and the longevity of their parents: The offspring of long-lived parents had a substantially greater (about 25%) life expectancy than did offspring of shorter-lived parents (figure 8.1). The virtue of a life table approach is shown by Yuan’s calculation of life expectancy for sons at age 20. These individuals are already survivors of the high rates of childhood mortality (see figure 8.3). Consequently, there is much better agreement between parental life spans and the surviving son’s total expected life span at age 20. Comparison of the survivors at this age removes much of the early environmental effects, highlighting the genetic similarities and/or differences. Jalavisto (1951) analyzed 12,876 individuals from Scandinavian genealogies of the middle class and minor nobility from 1500 to 1829. Plotting the mean length of life of the offspring as a func-

50 Parents died at ages 20–49 Parents died at ages 50–69 Parents died at ages 70 and over

45

40

Figure 8.1 The further expectation of life in years for sons whose fathers and mothers both had short, intermediate, or long life spans. Note that the sons’ life expectancy at age 20 has the same rank order as the parental life spans. This difference disappears as the sons survive to older ages. (After Yuan 1932.)

Life expectancy (years)

35

30

25 20

15

10

5

0 20

25

30

35

40

45 50 55 Age (years)

60

65

70

75

8.2 Genealogical Studies

321

Life expectancy of offspring

45

Figure 8.2 The mean length of life as a function of parental age at death. In the maternal series, the mother’s age was held constant; in the paternal series, the father’s age was held constant. Note that the expectation of life for both sons and daughters rises steadily with increasing maternal length of life but that the increase in expectancy with increasing paternal longevity is much less, particularly daughters. (After Jalavisto 1951.)

40

35 Daughters Sons Daughters Sons

50

50–59

Maternal series Paternal series

60–69

70–79

80

Parental age at death (years)

trait is termed the heritability, and symbolized as h2. Heritability is broadly defined as the proportion of the total phenotypic variance that is genetic, and it is determined by dividing the genetic variance by the sum of the phenotypic and environmental variances. Mayer (1991) found that parent–child longevity had a heritability value of 10–30% (i.e., within the 95% confidence interval, or the range within which we are statistically certain the correct value falls). For sibs, he found a higher value of 33–41% (probably because this value also includes the effect of common developmental environments). A subset of mildly inbred families showed a slightly (but not significantly) higher h2 value. Finally, there was no evidence of significant alterations in the heritability values during the 300 years of the study, even though the U.S. population, including these families, changed dramatically from both a genetic and demographic viewpoint during the same period of time. The most reasonable interpretation of the data is that (1) a moderate genetic component determines about 10–33% of the vari-

ance in human longevity and (2) this heritable effect is independent of population-wide environmental and social changes. This is a general phenomenon, not limited to humans, for heritability of life span in baboons has been estimated at 23% of the variance (L. J. Martin et al. 2002). Finch and Tanzi (1997) surveyed heritability studies in various vertebrates and invertebrates. They concluded that, despite the evidence showing that some genetic variants or mutants can significantly affect senescence and longevity, in general, the heritability of life span is relatively low, accounting for less than 35% of the variability. (The only exceptions to this seems to be found among relatives of centenarians; see below.) For humans, this low heritability implies that choice of lifestyles probably has a profound and important influence on the outcome of aging. We cannot control our genetic composition, but we can decide, via our behavior, which risk factors we minimize. What makes this low heritability of longevity so striking is the comparatively high heritability of

322 Chapter 8 Genetic and Social Aspects of Aging in Humans many common diseases. For example, the h2 of asthma is 60%, of insulin-independent diabetes mellitus is 70%, of blood pressure is 40–70%, of bone mineral density is 60–80%, of obesity is 50– 90%, and of osteoarthritis is 50–70%. How may we explain the difference in h2 between longevity and disease? In disease, it is usually clear that a small number of genes are involved in predisposing one to develop the pathology. In longevity, however, it may well be that what we inherit is the absence of any predisposing genes for illness. In this scenario, the presence of illness requires fewer genes than the presence of longevity, and so the heritability of longevity might be expected to be lower. There is one exception to this rule that tests the case. Certain centenarian families are believed to have one or more pro-longevity genes, and there is a high h2 for longevity in these particular families (see below). And so the general rule might be that high h2 is expected when a small number of genes are required for the expression of the syndrome, but a low h2 is expected when it is the absence of many genes that is required. Gavrilov and Gavrilova (1997) undertook a different type of genealogical study. They used high-quality genealogical data from the Russian nobility of the 17th to 19th centuries and examined the longevity of the offspring as a function of the parental age at reproduction. They found that the mother’s age at reproduction was not correlated with the longevity of either her sons or her daughters. The father’s age at reproduction, however, was significantly correlated with the longevity of his daughters but not his sons. The daughters of older men (50–59 years old) showed almost a 4-year decrease in their mean life span, from 64.6 years to 60.8 years. The mechanisms involved are not clear, yet a change of such magnitude likely has many biological and social effects. Given the increase in the divorce rate since the 1800s and the probable increase in the number of second families started by older men, this sex-specific age effect may act as a countervailing force against the long-term trend of increasing longevity. In addition, they detected a pattern of familial resemblance in life span between long-lived parents and their children, indicating an increased life span heritability for the

progeny of longer-lived parents (Gavrilov and Gavrilova 2001). This finding is consistent with independent studies done on centenarians (see below). A similar sort of genealogical study was done using a historical data set involving 8861 individuals from the British aristocracy covering the period from 700 to 1875 (Westendorp and Kirkwood 2001). These researchers also found evidence for a correlation of female longevity with that of their fathers. However, Westendorp and Kirkwood also found evidence that a daughter’s longevity was dependent on that of their mothers, and to some extent on that of their spouse. A man’s longevity was more heavily dependent on that of his spouse. Thus, despite their substantial differences in time, place, and execution, all the genealogical studies agree in showing a moderate genetic component to human longevity. It would be interesting to redo some of these genealogical studies in another century or two, by which time the life span will have undoubtedly increased, and see if the heritability values have decreased as we have learned and employed more effective nongenetic interventions.

8.3 Population Studies Raymond Pearl pioneered various study procedures in his efforts to refine the simple genealogical surveys. In one study, he and his colleagues examined the longevity of offspring born to individuals who lived to a very advanced age (90 or more years). The data from this study were compiled and analyzed by Abbot et al. (1974) and Murphy (1978). Again, they show a striking positive relationship between the longevities of parents and offspring (table 8.2). However, the data also show, in opposition to the results of Jalavisto (1951), a greater effect of the father than the mother on the longevity of offspring. The insurance companies realized that it would be beneficial to their future profitability if they could determine whether there was a significant positive relationship between parental

8.3 Population Studies

323

Table 8.2 Mean Life Span of Children Who Had One Very Long-Lived Parent Age of non-proband parent at death Sex of proband

parenta

Male Female

Sex of Child

< 60

60–80

>80

Male Female Male Female

67.6 73.8 67.0 73.0

71.4 74.1 69.3 73.5

73.2 77.2 70.9 73.3

Source: after Murphy (1978). aAll

proband parents lived to at least 90 years.

and offspring longevities. Several different studies, summarized by Cohen (1964), have been conducted to address this question. The data from these investigations generally, but not consistently, are compatible with the conclusion that heredity counts—that a good predictor of your life span might be your parents’ life spans. It is, of course, this early insight into the economic value of knowing an individual’s family history that has led to discussions of the privacy of personal genetic and medical information. Sons are physiologically more demanding to produce than daughters, and there are some data showing that mothers who have more sons have a significantly shorted life span, while mothers who give birth predominantly to daughters have a nonsignificant increase in life span (Helle et al. 2002). It should be noted that these longevity changes appear to show up only in women who have seven or more sons (or daughters); thus the relevance of these data for modern life is tenuous at best. Female life span in preindustrial societies was dependent not only on parental inheritance, but on the number and sex of children and on the social effects of a gender-based family structure. The data are inconclusive as to whether there is a general reproductive cost (i.e., is there an inverse relationship between female longevity and number of births regardless of the children’s gender?). Data showing a reproductive cost in the British aristocracy has been reported by Westendorp and Kirkwood (1999), but no such evidence was detected in the Russian aristocracy by Gavrilov and Gavrilova (1999). In addition, there are data involving the relationship between repro-

duction and longevity of 16th–17th-century French Canadian women, a population that did not use contraception and had not undergone the demographic transition and in which fecundity was honored (Le Bourg et al. 1993). There was no evidence of any trade off between female fertility and longevity. In contrast, a comparison of European aristocratic and rural individuals from the 18th and 19th centuries showed an apparent reproductive cost in that the more fecund rural women lived shorter lives than the less fecund aristocratic women (Korpelainen 2000). However, other data made clear that the two groups differed in their reproductive patterns, as the aristocrats underwent fertility selection, while the rural women underwent mortality selection. Because the two groups were living substantially different types of lives as a result of the significant environmental differences between them, it is probably not wise to directly compare them. The most prudent way to interpret the data might be to conclude that humans alter their reproductive and survival characteristics in response to broad environmental factors. The penetrating effects of those environmental differences make it difficult to directly compare populations living under substantially different conditions. All the studies cited here suffer from some sort of methodological or statistical shortcoming, as has been pointed out in some detail by Cohen (1964); no single study is entirely reliable. Yet the broad trend is clear enough: Parental age has some effect on the length of life of the offspring. The results appear to be more striking for sons than for daughters. This finding has some practical implications, not the least of which is that persons

324 Chapter 8 Genetic and Social Aspects of Aging in Humans born to older parents may need to be more closely screened for their health problems at older ages. The results also have implications for the current trend of delaying childbearing in the developed societies (Gavrilov and Gavrilova 2003).

8.4 Twin Studies To an incautious geneticist, the data patterns of these several genealogical and population studies immediately suggest polygenic and additive inheritance; that is, a large number of genes, each with a small but cumulative effect, could be involved. This conclusion would be premature, however, because not all familial effects are necessarily genetic. For example, there is probably an almost perfect correlation between the language the parents speak and the language their child speaks, yet to argue that the high parent– child correlations are evidence of genetic transmission of this trait would be foolish. Many of the lifestyle choices we make that could affect our longevity, we likely learned at our mother’s knee. A classic method of sorting out such familial environmental and genetic effects is to study twins. Kallman (1957) collected the histories of 1739 pairs of twins in which at least one twin lived to age 60. Kallman examined both the effects of parental life span on the longevity of twins and the similarity of life span between identical and fraternal twins. Although this study also had some statistical shortcomings, the data point in the same direction as the results of the genealogical and population studies discussed earlier. There was a good positive correlation between parental life span and progeny life span. More interesting, however, is the demonstration that there was less difference in longevities for identical twins (36.0 months difference) than for fraternal twins of the same sex (74.6 months) or of different sexes (106.0 months). A similar pattern of longevities was observed among Danish twins (McGue et al. 1993). In other words, the possession of an identical genome by twins brings with

it a much more nearly identical life span than is characteristic of other siblings. This conclusion is supported by the observation that identical twins have a somewhat higher heritability of longevity (with values of h2 ranging from 0.33 [McGue et al. 1993] to 0.50 [Yashin and Iachine 1995b]). Recall that in the general population, h2 ranges from 0.10 to 0.33 (Mayer 1991). In addition, monozygotic (identical) twins appear to show a statistical association between middle-aged and senescent deaths (McGue et al. 1993), as the life expectancy of one twin was significantly less than predicted if his or her co-twin died in middle age. The behavioral and psychological findings also suggest that identical twins undergo a qualitatively similar, if not identical, pattern of aging as they grow older (Bank and Jarvik 1979; Jarvik 1979). This last finding is entirely consistent with the results of several more recent twin studies, all of which show that genes play a dominant role in personality development— with heritabilities for a general cognitive factor of 81% early in life, decreasing to 51% in later life (Finkel et al. 1995; Plomin et al. 1994). Together these results suggest that heredity is a significant factor in determining the human life span. The apparently low magnitude (about a 2- to 5-year increase in offspring life span for every 10-year increase in parental life span) might reflect the possibility that what we inherit from our parents is not a tendency to long life but the absence of a tendency to short life. The identity of the patterns of aging in identical twins, apart from any considerations of the correlations in familial longevities, suggests that the manner in which our individual behavior and physiology change and adjust with age may be under more of a positive genetic control than we had previously thought possible. Our genes give us not just the absence of deleterious conditions, but also the unique temporal reaction patterns that identify us as surely as our fingerprints do (see figure 3.18). Most human studies rely heavily on the length of life as an indication of the influence of genes on life span. One important conclusion afforded by twin studies is that our reliance on life span measure-

8.5 Sex Differences and Longevity

ments alone may seriously qualitatively underestimate the role that genes play. A better estimate would perhaps pay more attention to quantitative and qualitative biomarker measurements, as exemplified by the work of Manton et al. (1995), in which the specific physiological variables affecting the functional capacities of individuals were identified, and these variables underlay the late age-specific mortality (see section 8.11 below). Knowing which of these variables are under strong genetic control and which are subject to environmental modulation would likely facilitate rational and targeted interventions.

100000

1.3

1.2

1.5

2.9

2.7

325

8.5 Sex Differences and Longevity In the past, men may have proclaimed themselves the stronger sex and convinced many people of the truth of this proposition. But they could fool neither the geriatric physicians, the bulk of whose patients was composed of women, nor the demographers, who counted the survivors. In fact, mortality rates are higher for males than for females through every part of the life cycle (figure 8.3). At conception, it has been estimated that the male:female sex ratio may be

2.2

1.8

1.8

1.8

1.6

1.5

10000

1000

100

10 0.5

10

30

50

70

85

Age (years) Figure 8.3 Sex-specific mortality rates in the United States in 1976. The numbers across the top of the graph are the male/female death rations for each age group (e.g., a ratio of 2.9 means that 290 males in that group died for every 100 females). (After Hazzard 1986b.)

326 Chapter 8 Genetic and Social Aspects of Aging in Humans as high as 170:100. The excess male mortality clearly is responsible for reducing this ratio to about 130:100 at the first trimester and then to about 106:100 at birth. It has been suggested that women live longer than men not because old women are healthier than old men, but because younger men are not as healthy as younger women. Some support for this point of view is found in the fact that the flu pandemic of 1918 had a major long-term impact on the sex differential: The epidemic killed more men than women. The surviving men were healthier, however, and the sex differential in longevity disappeared for more than a decade (Noymer and Garenne 2000). A similar pattern happened with tuberculosis, and there was clearly an interaction effect in that men with tuberculosis were more susceptible to the flu. The return of the sex differential is thought to be related to an increase in male heart attacks due to earlier increases in smoking rates among males. Another interesting but still unexplained observation is that, in general, women have more illnesses than men, but their disorders are less likely to be fatal (Holden 1987b). Perhaps this finding is related to the observation that the immune response of women is quantitatively greater than that of men. This observation has been seen in a variety of other species as well and suggests

that females can combat disease more efficiently than males. Hormones affect the immune system (see chapters 5 and 12), and it has been shown that treating female mice with male hormones reduces their immune response. Postpubertal castration of mentally deficient men indicated that eunuchs live slightly longer than normal controls, possibly because of the reduction of their initial mortality rate (IMR) in a Gompertz analysis (Hamilton and Mestler 1969). One possible explanation for this effect is the inhibitory effect of high levels of sex steroids on functions related to longevity (see chapter 6). The data in figure 8.3 are taken from a developed country. It is fair to ask if all human populations show this same gender difference in longevity or whether it is a particular attribute of only certain populations. Some data suggest that this gender difference is a general human trait. Regardless of the actual mean life span, Gavrilov and Gavrilova (1991) found that males lived longer than females in only 6 out of 157 populations. However, D. E. W. Smith (1993) has shown that the actual size of the gender gap varies widely among human populations. In developing countries, the gender gap varies from about 4 to 10 years (table 8.3), but the size of the gap varies for widely different and usually societyspecific reasons. Among developed nations, Japan

Table 8.3 Sex-specific Differences in Mean Life Expectancy at Birth Country

Male

Female

Gender gap (female–male)

Greece Japan Iceland United States France Finland Former Soviet Union Guatemala Pakistan Bangladesh Nepal

72.1 74.8 74.0 71.4 71.0 70.1 62.9 55.1 59.0 54.9 50.9

76.3 80.4 80.2 78.3 79.2 78.5 72.7 59.4 59.2 54.2 48.1

4.2 5.6 6.2 6.9 8.2 9.8 9.8 4.3 0.2 –0.2 –2.8

Source: from table 5.1 of D. E. W. Smith (1993).

8.5 Sex Differences and Longevity

nomic groups live longer than the males in their cohort. So considering all the data, it is reasonable to conclude that the gender gap is a consistent but modifiable human trait. Incidentally, this gender gap is not a general law of biology. In some species, females live longer than males, in others the reverse occurs, and in some there is no difference between the two (Gavrilov and Gavrilova 1991). The gender gap in each species might well have its own particular causation and consequent explanation. Gender-specific mortality has been attributed to all sorts of biological phenomena, not the least of which is the fact that males have only one X chromosome while females have two. However, the situation is not as simple as it might seem, since one X chromosome is normally inactivated in all human females. Thus both sexes have the same number of functional chromosomes. More recently, attention has focused on physiological factors. It has been known for a long time that women have lower indices of coronary artery disease than do men. An examination of the epidemiological and clinical evidence led Hazzard (1986a) to conclude that a major cause of the increased mortality of younger males can be attributed to the clinical consequences of coronary artery disease. In addition, the limited data of Ornish (1993) suggests that lifestyle changes

had the smallest gender gap, presumably because both men and women live so long there. (More recent data show that in 1996 the United Kingdom had the smallest gap at 4.9 years Gjonca et al. 1999.) The former Soviet Union had the largest gender gap, but apparently because both sexes (especially males) are showing a dramatic decrease in mean life span there. The small gender gap in Greece seems to stem from the fact that women there do not appear to live as long as women in other developed countries do, and the identical gender gap in Guatemala seems to stem from the fact that the majority of people dies prematurely there. The cases in which the gender gap disappears or even reverses itself, societies are characterized by high rates of premature mortality and/or a devaluation of female children. In the United States the gender gap has approximately doubled since the turn of the 20th century, suggesting that women have benefited more than men from the advances in living standards (D. E. W. Smith 1993). Within the United States, the data suggest that most different ethnic groups exhibit the gender gap even as their longevities change in response to society wide changes (table 8.4). Note that 22 of the 24 possible comparisons in table 8.4 show that women of different ethnic and (presumably) socioeco-

Table 8.4 Mean Age at Death of Different Ethnic Groups at LT50 and LT75 over Three Decades in California Female Group White African American Hispanic Asian/others

LT50 LT75

327

Male

1970

1980

1990

1970

1980

1990

83.4 87.8

85.9 89.6

87.0 91.3

78.2 83.7

79.9 85.7

81.1 86.4

LT50

73.9

78.0

81.0

68.3

72.0

73.4

LT75

81.2

84.4

87.2

76.7

78.9

80.5

LT50 LT75

76.1 83.3

79.1 85.6

81.2 87.4

72.7 80.2

72.8 80.8

71.1 80.7

LT50

77.9

81.7

82.5

78.4

79.5

80.0

LT75

84.2

87.6

88.3

84.2

84.2

84.7

Source: data from tables 3 and 4 of Go et al. (1995).

328 Chapter 8 Genetic and Social Aspects of Aging in Humans may lead to a faster regression of coronary artery disease in females than in males, which implies the possibility of a hormonal basis. One interesting aspect of estrogens is that their cardioprotective effect may stem from their ability to serve as antioxidants in vitro (RuizLarrea et al. 1997). If this possibility is correct, then it would imply that males and females differ in their innate levels of endogenous oxidative stress. Assays designed to test this hypothesis yielded provocative data. Using the amount of peroxide generated by mitochondria as an indicator of the reactive oxygen species (ROS) leaking out of the organelle (see chapter 9), Borras et al. (2003) showed that peroxide generation is higher in liver and brain mitochondria taken from male rats than from female rats (table 8.5). As one might expect, males exhibited higher levels of oxidative damage to their mitochondrial DNA than females. Moreover, ovariectomy increased the peroxide production in sterilized females to male levels, and this effect was reversed by estrogen replacement therapy. It would have been interesting to note if estrogens would have had a protective effect on male mitochondria as well. Mitochondria are known to contain an estrogen receptor (Chen et al. 2004). Estrogen exerts a protective effect on cells subsequently exposed to toxic agents, possibly by the indirect activation of the atheroprotective prostaglandin PGI2 (Egan et al. 2004) and/or possibly by stabilizing the antiapoptotic proteins that initiate cell death (Nilsen and Brinton 2004; also see chapter 10). It has been hypothesized that estrogen may be one component in mitochondrial signaling to the nucleus (Felty and Roy 2005). Female rats have significantly higher levels of various antioxidant enzymes than do males, and this indicates that the lower oxidative damage levels in females is likely due both to their lower ROS production and to their higher expression and activity of various antioxidant enzymes relative to males. The longevity difference between the sexes may have its basis in this estrogen-dependent modulation of ROS production and antioxidant enzymes. In a quantitative estimate of the factors contributing to the sexual difference in mortality, Waldron (1987) estimated that about 18% of the

sex differential in total mortality is due to sexspecific hormonal effects, which may represent an intrinsic and perhaps unchangeable risk to the male sex, although it must be recognized that dietary supplementation might protect against mitochondrial dysfunction (Miquel 2002; also see Liv et al. 2002). However, Waldron (1987) also estimated that at least 55% of the sex differential in total mortality can be attributed to destructive behaviors such as smoking, as well as to accidents due to reckless use of guns, cars, and alcohol. This is supported by studies showing that female longevity seems to be less dependent on genetics than is male longevity, and that sociocultural conditions allow female centenarians to practice a healthier life style and live under more favorable environmental conditions than do men (Franceschi et al. 2000). Whether such sexually characteristic behaviors are the product of hormonal influences is a moot point. What is important is that much of the sex differential in total mortality may represent modifiable behaviors.

8.6 Ethnic and Social Differences In recent years much attention has focused on the measurable differences between different ethnic groups with respect to various social and cultural activities. The age-adjusted mortality rates show large racial and ethnic differences, almost all of which can be attributed to social and economic factors. But the important question is whether there is a significant difference in the manner in which different ethnic groups age. The data of tables 5.1 and 8.4 suggest that the answer is no. The different ethnic groups represented in these tables seem to age in a more or less identical manner, at least as judged by the demographic and survival data. All groups except Hispanic and Asian men (see table 8.5) showed substantial increases in mean longevity during the past three decades. The failure of these males to emulate their sisters and wives can probably be attributed to the high-risk behaviors of young males (see figure 8.3). By 1990, the differences between like sexes of any two groups seemed to be approach-

8.6 Ethnic and Social Difference

329

Table 8.5 Estrogen Protects Mitochondria from ROS Stress

Assay H2O2 production mtDNA damage

Males

Females (intact)

Females + ovariectomy

Females + ovariectomy + estrogen

1.00 1.00

0.75 0.30

1.1 —

0.5 —

Source: redrawn from data presented in Borras et al. (2003). Note: all data expressed relative to males.

ing the magnitude of the difference between the sexes of any one group (see table 8.4), suggesting that the intergroup differences are not intrinsic but may be due largely to socioeconomic and cultural factors. There are some minor differences that may be intrinsic, such as a greater sensitivity of African Americans to renal disease or a greater life expectancy for older (85 years) African-American men and women relative to whites of comparable age, but these differences do not contradict the interpretation of the data of table 8.4 that the variation in patterns of aging within any single group are comparable to the variation in patterns of aging between groups. There are socioeconomic differences within and between different groups. It is a general finding, exemplified by the data of tables 5.1 and 8.4, that survival curves are to a large extent a function of social conditions. That is, an individual’s life span is the result of a complex interaction between his or her genome and the environment, a finding obtained in numerous population and twin studies (e.g., Finkel et al. 1995). Improving the environment is the fastest and most efficient method of increasing mean life span, as implied by the data of table 8.4. And degrading the environment, as is the case in much of eastern Europe and the former Soviet Union and developing countries (see table 8.3), is an effective way to shorten our lives. Keep in mind that genes do not operate in a vacuum; they interact with the environment in complex and still poorly understood ways. The adoption of a deterministic social policy is not justified by the available genetic evidence. There may be varied and debatable reasons for adopting a particular public policy, but one certain reason for

rejecting such a policy would be if it contradicts biological knowledge. One can subdivide a population by identifying characteristics other than race and see whether longevity is affected by that variable. Income is a major factor. There is a twofold difference in the age-adjusted male death rates between the richest and poorest residency areas such that mortality rate is related to family income up to a level of about $20,000, beyond which mortality rate changes very little with income (Sorlie et al. 1992). Harmful behaviors such as smoking are more prevalent in low-income groups, and protective behaviors such as moderation in the use of alcohol are more prevalent in high-income groups (D. E. W. Smith 1993). It is reasonable to speculate that harmful and beneficial dietary practices, respectively, follow the same pattern. Dietary behaviors, of course, directly alter the blood glucose, insulin, and insulinlike growth factor-1 levels. I discussed in chapter 6 that these factors alter ISP activity, and hence the gene states of the body. These observations suggest that the effect of income on longevity is due in part to the physiological implications of dysfunctional behaviors and cannot be ascribed entirely to the shortage of money per se. Educational level is another variable that affects longevity. The progression in educational level attained from primary to secondary to postsecondary is associated with a significant reduction in mortality rate for both blacks and whites and for both men and women (Christenson and Johnson 1995). This finding is consistent with studies showing a direct correlation between performance on IQ tests as a child and adult longevity (Whalley and Deary 2001).

330 Chapter 8 Genetic and Social Aspects of Aging in Humans Another factor is marital status. Married people live longer than those never married, divorced, separated, and widowed people at every age over 20 years, and mortality rates are as much as 50% higher for the unmarried groups (D. E. W. Smith 1993). Furthermore, one study has suggested that one effect of parental divorce may be decreased longevity among the children (Schwartz et al. 1995). Presumably, parental divorce alters children’s behaviors and eventually their physiology. “Lifestyle” seems to affect longevity. Both Mormons and Seventh-Day Adventists follow more or less closely a religious-based way of life that proscribes certain behaviors (smoking, drinking alcohol, consuming caffeine) and encourages certain other behaviors (strong family life and education for both groups and vegetarian diet for Seventh-Day Adventists). When compared to the general U.S. population, both of these groups show increased longevity and decreased mortality from many common age-related diseases. All in all, these different factors may exert their effect on longevity and mortality via the differential physiological effects of behaviors that have a strong socioeconomic bias. Himes (1994) made an observation that supports this suggestion. As we have seen, the human life span has increased as a result of the elimination of premature deaths. An examination of the age patterns and causes of death in three different developed societies (Sweden, Japan, and the United States) indicates that Sweden and Japan have similar age patterns of mortality, which are, however, quite different from the pattern in the United States, which shows higher mortality at younger ages. A separate study compared sexrelated differences in mortality for two developed European nations, Denmark and France (Gjonca et al. 1999). Denmark has relatively low male but high female adult mortality, whereas France has relatively high male but low female adult mortality. All five countries have similar life span values but different age- and sex-specific mortality patterns. Different national groups appear to achieve a long life span in different ways, and this difference may be related to the cultural and ethnic homogeneity of the society. In a heteroge-

neous society such as the United States, the various factors and their behavioral correlates may have differential effects on population subgroups sufficient to skew the general response of the population as a whole relative to that of a more homogeneous society.

8.7 Superlongevity: Myth or Reality? There are reports that human beings, somewhere in the world, have discovered the secret of longevity. This idea appeals to a certain wishful side of our natures, especially since these “fountains of youth” are most often located among unsophisticated peoples leading a simple rural life in the Ecuadoran Andes, northern Pakistan, or parts of former Soviet Georgia. Several decades ago, reports began to reach the Western world that there were disproportionately high numbers of centenarians in these areas, and that some of them had reached the fantastic ages of 150 years or more. For example, reports from Caucasia claimed that Shirali Muslimov was the oldest man in the world when he died in 1973 at the alleged age of 168, and that Khfaf Lasuria was the oldest woman in the world when she died in 1975 at the alleged age of 142 (Pitskhelauri 1982). Alas, objective investigations such as those of the expatriate Russian gerontologist Zhores Medvedev (1986), now at the National Institute for Medical Research in London, have shown by analysis of the official Soviet census data that the maximum life span in these areas is probably no more than the expected 115 years or so that is characteristic of the human species. For example, the mortality rates in these areas are higher than in other parts of the Soviet Union until the ages of 79. Only after that point in the life span can one discern the lower mortality rates expected of a long-lived population. The geographic distribution of the centenarians is spotty; they are absent or very rare in many areas and yet relatively common (six or more) in some villages. Such villages become famous for the extraordinary ages of their select inhabitants. As Medvedev has pointed out, these particular villages, rather than the more

8.8 Centenarians

realistic statistical figures for the whole region, provide the main contribution to the public awareness of this superlongevity phenomenon. There are other statistical and methodological reasons for doubting the accuracy of the reported ages, not the least of which is the fact that there are twice as many super-long-lived men as women, a ratio completely opposite from that found in all other age groups of the population. Either something extraordinary happens to old men in Caucasia, or something extraordinary happens to the data. The tales probably had their origin in a little pious fraud. After all, in a culture where age is esteemed and respected, adding a few years surreptitiously to your real age is obviously a winning strategy. In the absence of documentary records and given a parent with the same name, the exaggeration probably wasn’t very difficult. Leaf (1984) demonstrated that a similar explanation applies to the Ecuadoran and Pakistani reports: the social and economic benefits of being extraordinarily old have unduly influenced people’s memories of their longevity. These rural Edens do not possess the secret of the fountain of youth. We should not judge these people too harshly, especially since many of us accede to the demands of our own culture and surreptitiously remove some years from our real age with the help of hair dye and faulty memory. In fact, Leaf (1984) pointed out that the raw and unverified data on the number of centenarians in the U.S. census for 1970 overstate the real (verified) numbers by a factor of 17, a factor more or less similar to that seen in the popularized overestimates of these three other societies. A large part of Jack Benny’s long-ago fame as a radio comedian was based on this age-denying behavior, to say nothing of the profits of present-day cosmetic firms. We each react to aging in the manner that our culture teaches us.

8.8 Centenarians Society has always regarded centenarians as having, in a sense, won the game of life. By living so long, they embody the luck that most of us think

331

we want for ourselves. And so kings and presidents send them congratulations on their birthdays, and reporters ask them to share with us their secret of longevity. And their secrets are multivaried and contradictory—do drink wine, don’t drink wine, do exercise, don’t exercise, do smoke, don’t smoke, and so on—and teach us little except the vagaries of life. But in recent years several organized biogerontological investigations into the incidence, geneaology, physiology, and genetics of centenarians have been conducted, and these studies are adding some substance to our understanding of these slow-aging individuals. Centenarians are rare, even in today’s longlived societies of northern Europe or Japan (see table 15.1). In Japan, for example, on January 1, 2000, 13,036 men and women had been verified as being 100+ years of age. Since there were 1,420,534 births in Japan in the year 1900, this means that only 0.92% of that birth cohort had survived for at least one century (Robine and Saito, 2003). The mean longevity in these societies is about 85 years. So, although it is fairly common for an average person leading a healthy life to live for nine decades, the ability to live another two decades or so past that average mark is such a rare event that fewer than 1 in a 100 individuals will attain it. The so-called supercenternarians, or people who have reached the age of 110 years or more, are even more extraordinary and rare: There are only about 153 such individuals known to survive in all the 18 major developed countries (Robine and Vaupel 2001). The important thing to note about these numbers is that they are increasing. Over the past 160 years, the maximum age at death, the number of confirmed centenarians, and the life expectancy have been rising in developed societies, a topic I explored in chapter 2 (Oeppen and Vaupel 2002). However, the fact that there are historical records of centenarians or other very old humans living in the past tells us that centenarians are not a new occurrence in human history but that their frequency increases as our environment becomes less dangerous and our population size increases. Given that the heritability of longevity is rather low in humans and other primates (~10–30%), then it might seem reasonable to conclude that

332 Chapter 8 Genetic and Social Aspects of Aging in Humans being able to survive for at least one century is largely a matter of luck. However, the New England Centenarian Project has uncovered at least four families that clearly demonstrate a segregation of genetic factors for extreme longevity over a three- to four-generation span (Perls and Terry 2003). These pedigrees are not based on the sporadic appearance of one or two extremely old (90+ years) individuals. In one family, for example, two long-lived individuals married and had 13 children, 8 (2 males and 6 females) of whom were 90+ years at death. Given that becoming a centenarian requires a 8- to 17-fold better survival probability than found in normallived females and males, it is obvious that the occurrence of eight extremely old children in one sibship cannot be explained by chance alone but must reflect the segregation of some genetic factors. This supposition is supported by two other facts. One is that siblings of centenarians have an enhanced survival relative to normal-lived individuals, their death rate being about half that of the normal cohort (Perls et al. 2002). That is a huge decrease, on the order of the effect of adhering to a caloric restriction dietary regime. A similar but not as extreme situation may be observed with the parents of centenarians. If genetic factors for longevity are segregating in families, then one would expect parents and siblings to have some of these same factors as their targeted centenarian and for this to be reflected in their life spans. This expectation is upheld by the data. Second is the fact that a particular polymorphism involving a certain region on chromosome 4 has been identified as being present in the centerarians and their long-lived sibs but absent in normal-lived individuals (Perls et al. 2003b). The identity of the gene(s) involved is not yet known. Centenarians appear to age slowly: 90% of them were independently functioning at the age of 90 years. But a study of the morbidity patterns among them suggests that they may be sorted into three groups with different strategies of living long, depending on whether and when they suffered from a typical senescence-related disease (Evert et al. 2003). The first group are the “survivors,” individuals who had a diagnosis of an age-

associated diseases before the age of 80 and survived it. Twenty-four percent of males and 43% of females fit into this category. The second group are the “delayers,” individuals who did not have a diagnosis of an age-associated diseases until after the age of 80. Forty-four percent of males and 42% of females fit into this class. Both of these groups would likely be the beneficiaries of increasingly better medical and general environments. Note that most centenarians fit into one of these two groups. Also note that more women than men were diagnosed with disease at some time. The third group are the “escapers,” individuals who attained their 100th year without the diagnosis of any common age-related disease. Thirty-two percent of male and 15% of female subjects were in this group. These would be the people with particularly interesting genetic or physiological attributes. Note that significantly more men than women are in this nondisease group. This fits with the observation that more women than men are centenarians (ratio ~10:1) but that the men are healthier. The continued study of this third group of centenarians will likely yield insights into the mechanisms that allow one to have a slow rate of loss of function without ever crossing the line from health to disease. We do not know the genetic or physiological mechanisms operative in centenarians, but there are some interesting reports indicating the involvement of well-known mechanisms. First, studies comparing adults of various ages with centenarians on various aspects of body composition and insulin action found that centenarians had a lower body mass index, lower levels of body fat, lower plasma triglycerides, and lower indicators of oxidative stess (Barbieri et al. 2003; Paolisso et al. 1997). Centenarians also had a higher insulin sensitivity than that of much younger adults (i.e., their insulin resistance was much lower). Although the centenarians had low absolute levels of insulinlike growth factor (IGF-1), their levels of active IGF-1 (i.e., not bound to a transport protein) was higher than that of aged adults (75–99 years) but lower than that of younger adults (<50 years). Thus, the possibility exists that these centenarians have somehow preserved their IGF-1 and insulin

8.9 Segmental Progerias and Premature Aging Syndromes

homeostatic mechanisms, allowing them to retain their insulin sensitivity and escape the derangement of their metabolic control mechanisms and its consequences. If so, this would independently implicate the growth hormone/IGF-1 axis and the insulin/glucose homeostatic processes as potential mechanisms in the extended longevity of at least some centenarians. Another study showed a significantly higher frequency of a particular mitochondrial mutation in some centenarians relative to normal-lived people (Zhang et al. 2003). The mutation involves an alteration of the mitochondrial replication control site. It is plausible, although unproven, that an accelerated mtDNA replication might somehow compensate for various age-related insults to mitochondrial function (see chapter 10). What is more likely is that the known age-related decreases in mitochondrial function may be related to the age-related increases in insulin resistance. A comparison of healthy, lean, young and old adults revealed that the approximate 40% reduction in mitochondrial oxidative phosphorylation activity was highly correlated with increased triglycerides in the muscle and liver (Petersen et al. 2003). The insulin resistance observed in the old adults was confined mostly to skeletal muscle. The approximate 45% increase in the lipid content of the muscle was not due to changes in lipid metabolism per se but rather was correlated with increased insulin resistance. And the insulin resistance was significantly correlated with the rates of muscle ATP synthesis (i.e., mitochondrial function). Taken together, the data suggest that insulin resistance might be related to the increased intramuscular lipid content, which might result from the age-related changes in mitochondrial function. Although this latter study did not deal with centenarians, the other studies discussed above point out that healthy centenarians have retained their insulin sensitivity. For the moment, it is reasonable to conclude that the difference between healthy normal-lived individuals and healthy centenarians is that the latter have maintained their mitochondrial function and insulin homeostasis, while the former have not. Understanding the reasons for this differential functioning will be important. These findings

333

suggest that centenarians live long because they are a lot better than the rest of us at preserving key integrative metabolic functions (particularly those involving the ISP) which have been independently demonstrated to be of major importance in longevity regulation (see table 7.12).

8.9 Segmental Progerias and Premature Aging Syndromes If life span had a simple genetic basis, one would expect to find a strong correlation between parental and offspring life spans. The studies discussed above do not show this result. The human genetic data are consistent with these parent–offspring studies. Although more than 3000 specific human genetic conditions are known, no single mutation appears to lead to a significantly longer life span. Various single-gene genetic traits do, however, have a segmental feature characteristic of accelerated biological aging. “Segmental” means that the pathologies of these traits are limited to one or a few organ systems. Thus, they mimic the aging process, but only in part. George Martin (1978) combed through the human gene catalogue and identified at least 162 single-gene defects and numerous chromosome disorders, each of which displays a segmental aspect of this accelerated-aging phenotype. Of these, Down’s syndrome, progeria, and Werner’s syndrome have been most often viewed the most useful caricatures of normal aging. Good summary descriptions of these conditions may be found in Hasty et al. (2003). Down’s syndrome is caused by a chromosomal disorder, most often an extra copy of chromosome 21. Patients with this syndrome are characterized by (1) delay in the rate of normal development, (2) failure to achieve full development, and (3) more rapid onset of apparent aging, with various organ systems degenerating earlier than normal. Patients with this syndrome have a shortened life expectancy (only 8% survive to age 40) and suffer from premature graying and hair loss, increased tissue lipofuscin, increased neoplasms, variable adipose tissue distribution, amyloidosis, increased

334 Chapter 8 Genetic and Social Aspects of Aging in Humans autoimmunity, hypogonadism, degenerative vascular disease, and cataracts. In addition, they suffer from a precocious dementia, which appears to be accompanied by neuropathological changes indistinguishable from that of SDAT (senile disease of the Alzheimer’s type, or Alzheimer’s disease; see chapter 5). The gene responsible for the amyloid precursor protein (APP) composing the neuritic plaques characteristic of SDAT and of Down’s syndrome has been mapped to chromosome 21, the one causing the chromosome imbalance. This result first engendered a lot of speculative excitement that the two diseases might have the same dysfunction. In fact, patients with trisomy 21 overproduce APP from birth and show diffuse APP plaques as early as 12 years of age, long before they develop other SDAT-type lesions. However, continued research has shown that the story is more complex. As described in chapter 5, it has been shown that an increased production of APP, particularly the more harmful Ab-42 peptides, may be brought about by mutations on chromosome 21, on chromosome 14, or on chromosome 1; the age of onset of Ab-42 plaque deposition is accelerated by the ApoE4 gene on chromosome 19 but is delayed by the ApoE2 allele of the same gene, as discussed in the next section. Thus, overlapping but different genetic mechanisms seem to govern the production of these plaques in patients with Down’s syndrome and with SDAT. This observation reinforces the viewpoint that similar if not identical phenotypes may be produced by different processes. Down’s syndrome may be brought about via subtle changes in timing of the developmental events induced by the chromosomal imbalance, particularly involving other genes on the chromosome that are expressed in the facial and neural structures usually involved in Down’s syndrome (Hopkin 1995). If so, the etiological relationship of this syndrome to the normal aging process is not clear. Progeria, or Hutchinson-Guilford syndrome, is a rare genetic disease probably caused by the spontaneous appearance in one parent’s germline of a dominant mutation. Patients with this condition have striking clinical features that superfi-

cially resemble premature aging, but actually share no basic mechanisms. Patients usually appear normal at birth but begin to lose their hair and subcutaneous fat beginning sometime in the first year. Growth slows and finally ceases. These children characteristically attain the height of an average 5-year-old and the weight of an average 3-year-old. The skin becomes thinner, making visible the superficial veins. “Age spots” appear over the body. Bone mass is resorbed. Sexual development is limited. The progeric head, with its beaked nose and underdeveloped jaw, is characteristic. Patients have normal to above-normal intelligence. No neurofibrillary tangles appear in the central nervous system. The median age at death is 12 years. The cause of death is almost always very severe coronary artery disease. It is unlikely that this disease represents an acceleration of normal aging, so it is best viewed as a pathology with segmental applications to various tissue functions, the loss of which brings about an apparent—but not necessarily mechanistic— resemblance to aging. This view is supported by the recent finding that progeric children suffer from lamin-A defects. Abnormal lamin-A causes nuclear envelope interruptions, partial chromosome extrusion, and the silencing of genes located on such extrusions (Fossel 2004). Perhaps the great variety of phenotypic defects noted in these children arises from the possibility that different extrusions will result in the silencing of different genes, any or all of which can initiate a cascade leading to loss of function. Werner’s syndrome is an autosomal recessive trait that is sometimes called progeria of the adult. Patients generally appear normal during childhood but cease growth during the teenage years. Premature graying of the hair and baldness occur, as does skin and muscular atrophy, hypogonadism, poor wound healing, atherosclerosis, osteoporosis, soft-tissue calcification, juvenile cataracts, and a tendency toward diabetes. The median age at death is 47 years. Death appears to result from complications involving the cardiovascular system or from malignancies. Basic research at the cellular level has revealed that cells from patients with either Werner’s syndrome or progeria exhibit particular defects in connective

8.10 The Genetics of Human Senescence

tissue metabolism. Furthermore, their fibroblasts (a type of connective tissue cell) have a significantly shorter life span potential in vitro than do normal fibroblasts (see chapter 11). They also have a significant increase in chromosome abnormalities relative to the controls. The gene responsible for Werner’s syndrome was localized on the short arm of chromosome 8 and has been precisely located and identified as an WRN gene which belongs to the RecQ family of DNA helicases (Yu et al. 1997). Helicases are one of a class of enzymes involved in unwinding DNA in preparation for any one of various different DNA-specific activities, such as replication, repair, or recombination. This particular helicase is found predominantly within the nucleolus. At least four different mutations of this gene are found in patients with Werner’s syndrome, and they give rise to a variety of cell and clinical abnormalities (G. Martin and Oshima, 2000). Many of the abnormal WRN proteins possess helicase activity but have a markedly impaired expression (Fossell 2004). This gives rise to the possibility that the varied phenotypic effects might be traceable to reduced helicase activity in different tissues. Derangements in any one of the helicase-dependent activities could cause the cells to function poorly, and thus precipitate a premature aging syndrome. Certainly, much of the pathology of Werner’s syndrome can be understood in principle as the result of a failure of DNA repair or replication. I discussed in chapter 7 the important role of genetic stability as an essential component of extended longevity. The segmental progerias discussed above (as well as several other syndromes not discussed here; see Hasty et al. 2003) each represent unusual types of genetic instability, brought on by particular defects. These diseases validate the absolute necessity for genetic stability and reinforce the view of genetic stability as an important public aging mechanism. However, each of the segmental progerias appear to be affecting this public mechanism of aging in various private ways that are not likely to lead to any widely applicable mechanisms, except to show how particular defects in DNA repair might give rise to particular phenotypes.

335

8.10 The Genetics of Human Senescence As the preceding discussion of the segmental progerias indicates, all the known mutants that affect longevity in humans do so by reducing life span. The impact of other genetic factors on the incidence of age-associated diseases and disabilities is becoming more clearly defined. These genetic factors might act either by extending longevity in the individuals who carry them or by promoting disease and/or premature death in the individuals who carry them. Given the nature of medical research, more is known about the latter category than the former, so our account may be unbalanced. What we do know suggests that the eventual explanation of the genetics of human aging will include a complicated series of physiological and evolutionary interactions.

8.10.1 The ApoE Gene As an example of genetic factors that promote disease, let’s consider the apolipoprotein E (ApoE) gene. This gene encodes a protein synthesized in the liver, brain, lungs, spleen, kidneys, and macrophages. The ApoE gene is located on chromosome 19 at position q13.2. Human populations are polymorphic for three common alleles (E2, E3, and E4), which encode three major ApoE isoforms with different molecular and physiological characteristics (see Kamboh 1995 for references). When the frequencies of the three different ApoE alleles were measured in centenarians and in noncentenarians, the results showed a decided shortage of people bearing the ApoE4 allele among the centenarians (5.2%) compared to the normal controls (11.2%), coupled with an excess of the ApoE2 allele among the centenarians (12.8%) relative to the normal controls (6.8%; Schacter et al. 1994). The E2 and E4 alleles seem to have a positive and negative effect, respectively, on longevity. We have some insight into the nature of these effects. Individuals homozygous for ApoE4 alleles are significantly more likely to develop late-onset Alzheimer’s disease than are those homozygous

336 Chapter 8 Genetic and Social Aspects of Aging in Humans for ApoE3 alleles. Individuals homozygous for ApoE2 alleles are even less likely to develop Alzheimer’s disease, and heterozygous individuals demonstrate intermediate risks. The rare individuals who are completely deficient in the protein (because of a null mutation of some sort) are at very high risk of atherosclerosis and hyperlipidemia. Note that the presence of the ApoE4 alleles does not necessarily mean that the individual will develop Alzheimer’s disease, nor does its absence offer complete protection, for this allele is neither absolutely necessary nor sufficient for the expression of the disease. Thus, the presence of a functional ApoE protein is necessary to avoid very early death, but the particular isoform(s) present each have different probabilities of bringing about different effects in late-life morbidity for the individuals who carry them. Individuals who carry the ApoE4 allele are strongly predisposed to developing late-onset Alzheimer’s disease and are also likely to respond poorly to current therapies (Poirier et al. 1995). The latter observation opens the possibility of using genetic identity to subdivide a heterogeneous disease group into more homogeneous subsets of patients who can respond differentially to various therapeutic interventions. In addition to its role in neural impairment, the ApoE4 allele appears to be involved in the etiology of ischemic heart disease, or atherosclerosis, a leading agerelated disease and cause of death. The frequency of the ApoE4 allele varies among different populations, ranging from a low of 0.07 in Oriental populations to a high of 0.37 in New Guineans (Kamboh 1995), and these frequencies are inversely proportional to the frequency of ischemic heart disease observed in the different groups. In addition, fewer older individuals carry the ApoE4 allele than younger individuals, suggesting that fewer bearers of this allele survive to old age (van Bockxmeer 1994). How can one protein be involved in two syndromes as disparate as neural degeneration and atherosclerosis? A plausible, although unproven, scenario follows. The ApoE protein participates in all three known pathways involved in lipoprotein metabolism: transport of dietary lipid from the intestine to the liver, from the liver to extra-

hepatic cells, and from extrahepatic cells back to the liver. In this process, the ApoE protein binds to the surface of different types of cholesterol-rich lipoprotein particles and causes their uptake by liver cells, which have receptors on their surface that recognize and bind the ApoE protein. It has been suggested that the major difference among the three isoforms (E2, E3, and E4) is their differing number of cysteine residues. In fact, the three proteins are defined by whether they have cysteine or arginine at codons 112 and 158. The ApoE2 protein has cysteine at both sites, the ApoE3 protein has cysteine at codon 112 and arginine at codon 158, and the ApoE4 protein has arginine at both sites. The lack of cysteine in the ApoE4 protein and its consequent inability to participate in disulfide bond formation might result in a protein that has an altered binding capacity for a variety of intra- and extracellular components, and this condition may give rise to different disease syndromes, depending on the organ system involved (Kamboh 1995; Schacter et al. 1994). The lack of disulfide binding in ApoE4 causes it to fail to bind to a particular microtubule protein (tau; see chapter 5), and the unbound tau protein is consequently hyperphosphorylated and gives rise to the neurofibrillary tangles characteristic of Alzheimer’s disease. ApoE2 and ApoE3, in contrast, can bind to the tau protein and thereby prevent or slow the formation of tangles. The same lack of disulfide binding capacity of the ApoE4 allele for lipoprotein particles has been put forth as a plausible explanation for its association with atherosclerosis and ischemic heart disease (Kamboh 1995). Indeed, it has been shown that Alzheimer’s patients that have the ApoE4 allele also have abnormally high levels of cholesterol (Giubilei et al. 1990). The three ApoE alleles were shown to differ in their antioxidant activity and thus in their ability to protect cells against oxidative insults (Miyata and Smith 1996). As might be suspected, the antioxidant ranking of the three alleles is E2 > E3 > E4. The differential antioxidant activity probably is also due to the differential ability of the three resulting proteins to bind metals (which have pro-oxidant activities) and to remove them from the cell’s environment.

8.10 The Genetics of Human Senescence

8.10.2 Genetics of Hypertension High blood pressure is a risk factor intimately involved in the development of cardiovascular disease, diabetes, and the metabolic syndrome (see chapter 5). High blood pressure is a major public health problem; it affects about 25% of the adult population in developed societies. High blood pressure is under genetic control in rats and mice, and these model organisms have been well studied to uncover the molecular and genetic mechanisms underlying hypertension. This work has coincided with and informed the ongoing clinical research on the human condition. Much of the increased survival of middle-aged and older humans (figure 2.14) is due to the development of drugs that aid in controlling blood pressure. Hypertension is a classic example of a phenotype that depends both on a small number of genes with major effects plus a very large number of genes with minor effects, both interacting with a diverse spectrum of environmental factors. It may thus serve as a paradigm of the complex physiology needed to maintain normal functioning in this essential system and of the difficulty in controlling the system by pharmecutical means alone. A full discussion of this topic is presented in chapter 9 as a model with which to understand the genetics of senescence. The reader should incorporate that material into the present discussion as appropriate.

8.10.3 Other Genes I have outlined what appears to be a plausible and unitary (but still unproven) explanation for the role of the ApoE4 allele in two seemingly disparate diseases. (But the reader will know from the material presented in chapter 5 that this hypothetical explanation, even if true, is only one part of the complete story for each of these diseases.) The apparent underlying simplicity is masked by the complexity of the pleiotropic interactions stemming from this one protein. The fact that the presence of a particular allele does not guarantee the presence or absence of a particular syndrome suggests that there are probably substantial inter-

337

actions taking place between gene and environment, resulting in phenotypic plasticity. Such interactions could, if defined, lead to potential interventions. Similarly complex stories probably underlie the involvement of other genes, such as the angiotensin-converting enzyme (ACE) gene or the vitamin D receptor gene, both of which are associated with the promotion of age-related disease morbidity and mortality. Such studies will continue to reveal much about the biological basis and the nature of the genetic contribution to middle- and late-life morbidity and mortality and their consequent effects on life expectancy and individual longevity. In the ApoE story, we can tentatively conclude that defective protein-binding capacity leads to tissue-specific cascades of grave and abnormal consequences. Note that there is no gene for short life per se, but rather a gene that simply affects protein binding. All the rest is the result of a pleiotropic cascade. Perhaps most diseases will reduce eventually to similarly simple explanations. Charlesworth (1996) has pointed out that the progerias and the ApoE polymorphism can be viewed as excellent examples of Medawar’s concept of mutation accumulation and William’s notion of antagonisic pleiotropy. These are two of the major conceptual mechanisms underlying the evolutionary theory of aging. This does not necessarily mean that genes that enhance longevity do not exist in the human species. Such genetic conditions would be much less obvious in the population. A child dying of progeria or a man dying of heart disease at the age of 35 is much more noticeable and commented on as being caused by something out of the ordinary than is an elderly person living to an extraordinarily old age, particularly before the advent of careful population statistics to serve as an objective marker. We have discussed above that some centenarians have a particular region on chromosome 4 that seems to enhance their longevity relative to the general population. In addition, certain centenarians and nonagenarians were shown to have a statistically significant association with certain alleles at the HLA (human leukocyte system A) locus, which is responsible for much of our immunological defenses (see chapter 12;

338 Chapter 8 Genetic and Social Aspects of Aging in Humans Takata et al. 1987). There is evidence in human nonagenarians and centenarians of an excess frequency of some HLA alleles, but the data also indicate that there are important ethnic or geographic differences. For example, elderly Chinese in the Shanghai area seem to have an excess of the HLA-A9 allele and a deficiency of the HLAA30, Cw3, Cw6, and C27 alleles, all relative to the middle-aged control group (Ma et al. 1997). However, elderly French Caucasians showed an excess of the HLA-Cw1 allele in females and the HLA-Cw7 allele in males (Proust et al. 1982), while other French familial nonagenarians show an excess of the HLA-A30 and HLA-DR11 alleles (Henon et al. 1999). Thus, there are mostly differences between the two groups but some similarities as well. It seems reasonable that certain HLA halplotypes are associated with extended longevity, but more work will be needed to positively identify the particular alleles that seem to vary in different groups. Such alleles might be considered longevityenhancing genes. Other longevity-enhancing genes probably exist, but they have not yet been identified and characterized, since they likely require similar sorts of statistical and biochemical clues to be recognized. In addition, there is no reason to believe that humans are fundamentally different in this respect from laboratory animals known to have such genes. It is reasonable to believe that the human genome contains a diversity of genes that affect life span in both positive and negative directions. Another piece of evidence that points in this direction are the factors affecting the age at death of people afflicted with Huntingdon’s disease. Huntingdon’s disease is a rare autosomal dominant mutation that causes deterioration of certain neurons in the basal ganglia and brings about a neurological conditions affecting both movement and cognition. The disease characteristically has a mean age of death at midlife (30–50 years), and this observation is used as support for the mutation accumulation theory of aging (see chapter 4). However, there is a strong correlation between the age at death of individuals with the disease and the age at death of their nonafflicted sibs, such that patients from long-lived families tend

to die late in life, and patients from shorter-lived families tend to die earlier in life (Finch 1990). It appears as if the familial factors affect the age at onset and not the duration of the disease. The implication is that the familial factors regulate the age of onset of loss of function in both the afflicted and control sibs. These factors may be genetic, environmental, or both (see chapter 9 for a model of the onset of senescence). G. Martin (1978) concluded from his survey of genetic disorders affecting life span that there are a large number of genes at which variation could modulate aging. However, many of these genes might be expected to be involved only indirectly, and not causally. As discussed in chapter 4, the approach taken by Cutler (1975) and by Sacher (1975) to derive the rate of evolution of the modern human life span led them to conclude that longevity has evolved rapidly by means of relatively few mutational alterations, most if not all of which are thought to operate at the regulatory-gene level and not at the structuralgene level. Certainly the data of chapter 7 can be interpreted as leaning in that direction.

8.10.4 Future Studies and Interventions Until now, genetic investigations into human aging have been restricted to clinical or statistical studies, for obvious ethical and empirical reasons. However, a revolution has taken place in human genetics since the 1970s. In 1968, the chromosomal location of only one gene was known with any certainty. In 1978, the chromosomal location of about 120 genes had been painstakingly deduced (McKusick and Ruddle 1978). By 1988, more than 1600 genetic markers had been mapped (G. Martin and Turker 1988), and a complete linkage map of the human genome had been published (Donis-Keller et al. 1987). The human genome was sequenced in 2001, as were those of a number of other organisms, including all of our model organisms. (We have many fewer genes than was once thought [30,000–35,000] but many more splicing variants and other interactions to yield about 100,000 functionally different proteins.) The ability to

8.10 The Genetics of Human Senescence

compare genomes has led to a sea change in how we study genetics and cell biology. Enough progress has been made in mapping the genome that we can now associate particular enhancedlongevity pedigrees with particular DNA regions and with their gene products, adding a powerful tool to the analysis of longevity in humans, as was pointed out in the genetic studies on centenarians discussed earlier. One procedure used to analyze longevity is quantitative trait loci (QTL) mapping. If one has a dense array of detectable markers spread across each of the chromosomes, then one can make the appropriate crosses (or in the case of humans, assay family members who differ in the trait of interest) and determine which sets of genomic markers are preferentially associated with the phenotype of interest. The genes of interest lie somewhere between, and are linked to, the nonrandomly assorting markers. The most convenient markers are known DNA noncoding sequences as obtained from the sequenced human genome. At some point, our increased genetic knowledge will allow us to shift from a statistical and general approach to the genetics of human aging, to a genetic and causal approach. Thus, instead of calculating heritability values for longevity on the basis of the analysis of long- and short-lived populations, we will be able to identify the chromosomal regions—and eventually the genetic loci—involved in the trait of interest. This qualitative shift in our knowledge is one of the outcomes that justifies the cost of the Human Genome Project. Such procedures allow us to identify genes of medical interest in individual humans and model organisms, as well as genes of interest in commercially important organisms such as pigs, cattle, and trees. The qualitative advances in our knowledge brought about by the application of these procedures to humans promises to be inextricably entwined with some serious ethical and social problems. Reflect on the ApoE protein described earlier. Should people with the ApoE4 allele be refused health insurance? Should they be denied long-term and expensive education because they might not live long enough to make it worthwhile? Or should they be shunned in romance by prospective

339

spouses? These and other difficult questions are being addressed at both institutional and personal levels by both scientists and lay people alike. It would be useful to have some guidelines in place before we are once again surprised by the unexpected social and personal impact of our quest for knowledge (see chapter 15 for a more detailed discussion of bioethical issues). Once the genes that are believed to be mechanistically involved in a particular aspect of human aging are identified and sequenced, we can be sure that the desire to use these genes in some sort of intervention strategy will not be long in coming. In fact, the discussions have already begun (see chapter 15; Dykes 1996). The potential type of intervention most often discussed initially is somatic gene therapy: the introduction of normal genes into all or some of the somatic cells of an individual in order either to replace genetically defective functions or to alter pathological disease processes. Somatic gene therapy does not alter the heritability of the underlying mutation; such a change would involve germline gene therapy, which is not currently practiced in humans. Somatic gene therapy for humans is in its infancy, with more failures than successes to date. It does not seem to be practical except for certain monogenic diseases. Stem cell therapy, as discussed in chapter 6, might be a better approach Nonetheless, much effort and hope have been and continue to be invested in it, and it is worthy of at least a brief discussion here. The diseases currently targeted for somatic gene therapy include certain types of immunodeficiency, hypercholesterolemia, hemophilia, cystic fibrosis, and muscular dystrophy. All of these are monogenic conditions. None of these conditions is intimately involved in the aging process. This raises a question of strategy and priorities. If we assume that the goal of biomedical research is to increase the mean life expectancy of the population, then somatic gene therapy should be targeted on heart disease, cancer, stroke, and other conditions that kill middle-aged people. If, however, we assume that the goal of biomedical research is to maintain the health and decrease the portion of the life span during which aging people are disabled, then somatic gene therapy should be

340 Chapter 8 Genetic and Social Aspects of Aging in Humans focused on alleviating Alzheimer’s and other neurological diseases, arthritis, diabetes, osteoporosis, and immunosenescence, which kill or immobilize older people. The question is academic at the moment, since none of the genes for these disorders is yet characterized in sufficient detail, nor are the roles of such genes understood well enough to construct rational gene therapies. A more important consideration is that the probable polygenic nature of many of these conditions guarantees that effective somatic gene therapy will be some time in coming. A much more likely outcome of these genetic approaches for these multigenic syndromes will be the development of intelligently designed drugs which will interact with the genes or gene products affected in these conditions and retard or reverse the loss of function associated with the syndrome. I pointed out in chapter 6 that the thrust of research with the model organisms has been to develop pharmecutical interventions. The discussion in chapter 15 of the social and ethical implications of intervening in the aging process is predicated on the assumption that the first practical large-scale antiaging intervention will involve targeted drugs with particular effects.

8.11 Physiological and Other Predictors: Conclusions from Longitudinal Studies Although expensive and tedious, a relatively large number of human longitudinal studies have been done during the past half century or so. Together they provide a substantial database against which various generalizations can be tested. Unfortunately, the very real differences in their structure and methods make it impossible for databases of these studies to be combined, so each study must be understood on its own terms. Yet their results are generally compatible. I briefly examine certain selected results of the Framingham, Duke I, and Baltimore Longitudinal Study on Aging (BLSA) studies, as well as some early data from the Alameda County study. A wealth of data indicate the change in a particular physiological variable with age. But these

variables do not operate in a vacuum. They interact with other physiological variables over time, and some are greatly influenced by gender. The data from the Framingham study were analyzed by Manton et al. (1995b) and used to construct a more realistic model of the gender changes in several physiological variables and how they affect functional capacity and mortality. Some of these variables change in complex ways (figure 8.4). For example, what does the alteration in cholesterol levels with age say about the activity or changing roles of the ApoE alleles in these individuals? Cutting through such idle speculations, Manton et al. (1995) used these observed data to construct mathematical models of the optimum values of the 10 variables studied that would yield the lowest mortality, highest life expectancy, and highest degree of function (table 8.6). There are two important points to be garnered from this table. First, the optimum values for males and females are often quite different, and the difference between the observed and optimum values translates into apparently substantial amounts of unrealized life expectancy. Second, if this analysis is correct, there is a huge amount of unrealized life expectancy that we miss out on because our physiology is not optimized. The observed mean life spans in their study were about 73.7 years for men and about 79.6 years for women, when the optimized mean life spans should have been 111 and 103.3 years respectively. There is at least a 23–37 year penalty for not maintaining an optimal physiological fitness. Before one dismisses this as statistical fancy, remember that calorie-restricted animals maintain an optimal gene expression state and physiology and thus live about 40% longer than controls, a value consistent with the calculations in table 8.6. The extra 10-year longevity of some Seventh Day Adventists points in this same direction. In addition, Manton et al. (1995b) concluded that the individual’s functional capacity explains more of the age dependency of mortality than do risk factors. I pick up this point again in chapter 13. A complete description of the important physiological factors seemingly would yield a complex pattern of interactions that change with age and that have extensive pleiotropic effects on the functional status of the individual.

160

Vital capacity (CL/m2)

(a)

Male Female

140

(b)

120 100 80 60 40

30 40 50 60 70 80 90 100 110 120 130

(c)

Male Female

(d)

240 220 200

280

180 30 40 50 60 70 80 90 100 110 120 130

341

Male Female

270 260 250 240 230 30 40 50 60 70 80 90 100 110 120 130 48

Hematocrit (%)

Cholesterol (mg/dl)

260

Body mass index (10 x kg/m2)

8.11 Physiological and Other Predictors: Conclusions from Longitudinal Studies

Male Female

47 46 45 44

43 30 40 50 60 70 80 90 100 110 120 130

Age (years)

Age (years)

Figure 8.4 The predicted gender-specific ideal age trajectories of four risk factors calculated using risk factor dynamics and mortality equations estimated from the 34-year follow-up data from the Framingham Heart Study. The predicted risk factors are not constant but change throughout life. The risk factor may be similar (a) or different (b–d) in males and females. (After Manton et al. 1995.)

Table 8.6 Values Producing Lowest Mortality in Gender-Specific Mortality Functions Male Factor Pulse pressure (mm/Hg) Diastolic blood pressure (mm/Hg) Body mass index (kg/m2) Cholesterol (mg/dl) Blood glucose (mg/%) Hematocrit (%) Vital capacity (l3/m2) Smoking (cigarettes/day) LVH (%) Heart rate (per min) Life expectancy remaining at age 30 years Source: after Manton et al. (1995b).

Female

Observed means at 30

Optimal values

Observed means at 30

Optimal values

48.5 82.0 25.4 212.3 79.5 47.1 136.8 15.0 2.0 72.4

31.0 75.8 24.2 176.8 84.1 46.7 159.7 0.0 0.0 81.0

43.8 76.6 23.3 195.2 77.9 42.2 114.0 8.0 1.0 49.6

46.8 78.0 26.8 221.7 124.4 44.6 121.6 0.0 0.0 55.5

43.7

81.0

49.6

73.2

342 Chapter 8 Genetic and Social Aspects of Aging in Humans trates the potential value of this approach toward constructing a physiological index of aging. The first Duke study revealed a general stability over time in an individual’s activities and traits. We change less than we might fear or our teachers might hope. This study also made clear that many individuals do not show the functional decline in particular traits that were predicted by crosssectional data; thus, the results of this study emphasize the value of the longitudinal approach. In brief, the results show that in that particular study population, the strongest predictors of longevity were physical health, not smoking, work satisfaction, and happiness (Palmore 1982). Epel et al. (2004) demonstrated that women who were under high levels of psychological stress had significantly shorter telomeres than did nonstressed controls. Telomere length is known to have important effects on the cell’s functioning (see chapter 12). Our mind and body are connected, and some of the mechanisms described in chapter 13 may well play a major role. Finally, the results from the Alameda County study (Guralnik and Kaplan 1989) suggest that

The BLSA has performed an extensive number of physiological, cognitive, and psychological tests on hundreds of subjects over a 30-year period. The results to date have been thoroughly summarized (Shock et al. 1984). Most of the subjects are still alive, so many of the conclusions are still open-ended. One of the goals of this study is the construction of physiological indices of aging. An example of such an endeavor is provided in a report by Tobin (1981) in which he chose four physiological variables considered to be clinically important measures of individual health. Age-adjusted standard scores were developed for each test such that a score of 50 equaled the mean value of the physiological variable for any age group. These scores were then compared between study subjects who were still living (n = 860) and those who had died (n = 162). As figure 8.5 shows, the individuals who were destined to die soon had displayed lower-than-normal scores in three out of four of these physiological functions. An expanded version of this biomarker index is presented in figures 3.18–3.20 and illus-

52.5 <0.001

<0.001

<0.01

L

L

NS

50.0

T score

47.5

45.0

42.5

L

D

D

D

L

D

40.0

BP

FEV 1.0

CCR

[G]120

Figure 8.5 The relationship of performance level of four physiological tests to survival on subjects in the Baltimore Longitudinal Study of Aging. Shaded bars (D) represent results for subjects who have died; open bars (L) represent results for the survivors. The T score is an age-adjusted scoring system that allows for the relative comparison of the different tests. For the first three tests, T scores of survivors were significantly different from those of nonsurvivors. BP, systolic blood pressure; FEV 1.0, forced expiratory volume in 1 second; CCR, creatinine clearance; [G]120, glucose concentration 2 hours after administration of oral glucose. Values across the top represent probabilities (NS = not significant). (After Tobin 1981.)

8.11 Physiological and Other Predictors: Conclusions from Longitudinal Studies

the following variables can function as statistically significant predictors of healthy aging: race, higher family income, absence of hypertension, absence of arthritis, absence of back pain, being a nonsmoker, having normal weight, and drinking only moderate amounts of alcohol. Several of these variables have been statistically implicated in other studies, as described earlier in the section on ethnic and social differences. How does one interpret these different findings? The BSLA and Framingham results document the role of physiological factors, while the Alameda County and Duke I studies appear to

343

involve an interesting mix of physiological and behavioral factors. There is no contradiction between these disparate conclusions, for the neuroendocrine–immune system together translates physiological factors into behavioral traits and vice versa. The same factors are affecting each of us, but in different ways. Some of these differences might be due to real physiological differences between the sexes; others might reflect more of a sociocultural influence on our perceived sex roles. Men and women appear to live differently. No wonder it proves so puzzling when we live together.

This page intentionally left blank

Mortality Function

Part IV What Is the Mechanistic Basis of Aging and Senescence? 95 68 59 50

Age

This page intentionally left blank

9

Mechanisms Underlying the Transition from Health to Senescence

9.1 The Role of Theory in Gerontology Changes that lead to an increased or enhanced level of functioning of the organism are often defined as “development,” whereas changes that lead to decreased functioning are often defined as “aging.” Yet a descriptive account of frog embryology gives us no insight into the nature of the mechanisms regulating and controlling each phase of the frog’s development, and neither will it enhance our knowledge of aging. We cannot be sure that we understand the mechanisms of aging until we have a fairly good idea of what we are trying to explain. Historically, however, gerontology was characterized, even by its practitioners (Hayflick 1985), as lacking a strong database even while it abounded in theories to explain the fundamentals of biological aging. This situation is in the process of changing. The rapid pace of knowledge over the past decade has allowed us to develop a knowledge of facts and of theory simultaneously, each aspect building on the other. Without the ability to make testable predictions, it is operationally impossible to make a scientific judgment as to which one of several theories most closely describes reality. For example, the atomic theory in one guise or another has, since the time of the Roman poet Lucretius 2000 years ago, played a role in the history of chemistry. Yet plausible as this theory might have sounded to alchemists and chemists alike, not until 200 years ago were the first precise descriptions, measurements, and definitions of observed chemical changes made. This reliance on obser-

vation and empirical modifications of theory led to the development of the periodic table of the elements by Dmitry Mendeleyev in 1869, an innovation that allowed chemists to predict the existence and properties of hitherto unknown elements. Today we take it for granted that chemists can manipulate molecules, yet this mastery rests on the predictive ability of their theories. Biological gerontology is a young discipline and has only within the past several decades been able to attract the serious interest of large numbers of scientists. The popular interest in aging mechanisms and interventions likely has come about as a result of this increased scientific interest and activity. The reasons for the previous neglect are not clear, but they probably have a lot to do with the older scientific-cultural idea that aging was not a fundamentally interesting biological process. This perception probably stems from the fact that one of the central questions in biology for the past century has been the unraveling of development—of how an egg transforms into an adult. (The other question was that of deciphering the mechanisms involved in the process of evolution.) Aging was probably assumed to be an uninteresting sort of eroding away of the end product of development. But our concepts have now changed, in part because of demographic data and in part because we are beginning to understand that the failure to maintain function is an inherent—and potentially modifiable—part of the evolutionary design of organisms. Recent gerontological theory shows signs of becoming more predictive, using evolutionary

347

348 Chapter 9 Mechanisms Underlying the Transition from Health to Senescence theory to provide a unifying conceptual framework for the field and integrating it into the mainstream of biological thought, as shown in chapter 7. Certainly, the response to the publication in 1991 of Michael Rose’s book Evolutionary Biology of Aging marked the general acceptance of this point of view. The disposable-soma theory (see chapter 4) views aging and longevity as the inevitable outcome of an evolutionarily derived equilibrium between the amount of resources devoted to maintenance and repair and the amount devoted to reproduction. This theory subsumes many other observations of a valid but limited nature, each having to do with detailed biochemical mechanisms leading to cell, tissue, and/or organism failure, while simultaneously organizing them into a framework consistent with the fundamental biological principles of evolution and cell biology. At the very least, this theoretical synthesis allows us to see order where little was apparent before and to provide a coherent explanatory basis for both the similarity and variability of aging across different species. Quantitative computer simulations of the theory yield believable and interesting results, thus building confidence in the correctness of its assumptions (Kowald and Kirkwood 1994, 1996; see also chapter 14). Another integrative theoretical step has been the abandonment of the dichotomy between “pathological” and “normal” aging. It was long customary to consider the study of age-related diseases as a distracting hindrance to the study of “real” aging. But an extension of the evolutionary theory of aging led Bellamy (1988), Holliday (1995), and others to consider that the common age-related diseases are not just random occurrences of physiological breakdown but are the systemic failures that highlight the weak points of the evolved anatomical and physiological design of the organism. Thus, studying the etiology of these diseases allows us to identify these evolutionary weak points and to characterize the mechanisms involved in them. And adoption of this point of view allows us to integrate into modern biogerontology all of the huge host of medical studies bearing on these age-related pathologies, with a concomitant increase in our

detailed knowledge of the processes underlying the loss of function characteristic of aging. This does not mean that we must abandon gerontology and study only the care of the ill and aged. The province of geriatrics is not included in our curriculum. But geriatrics is organically connected to gerontology via the study of the etiology of age-related systemic failures. This connection also means that the mechanisms underlying aging are no longer different from those usually encountered in the rest of biology, but are continuous with them. This widening of our conceptual views and the increase in our detailed knowledge should give second thoughts to skeptics persuaded that the problem of aging is insoluble and that the field is not yet ripe for investigation. The adoption of an evolutionary point of view is not limited to gerontology; a Darwinian interpretation of disease was put forth and generally accepted (Neese and Williams 1994). In the past few chapters, I have discussed the mechanisms that determine longevity (see table 7.12 for review). These discussions showed that longevity-determination mechanisms control the level of somatic maintenance and thus indirectly determine the length of the “health span.” But then we undergo senescence, and gradually lose that high level of functioning. In the next few chapters, I discuss various senescent mechanisms and categorize them as either stochastic or systemic. By “stochastic” I mean a mechanism based on the occurrence of single random events, such as gene mutation. The term “systemic” here denotes an explanation based on the occurrence of a hierarchical cascade of interconnected events. By this I mean an interrelated series of biological processes linked by a web of feedback signals that often yield a progressively unstable positive feedback cycle. I most emphatically do not mean an explanation based on the idea of determinative and sequential gene action designed to produce aging. Such an explanation would constitute an adaptive theory of aging (that we age because there is a particular reason for us to age), and, as we saw in chapter 4, aging has evolved not because it is adaptive but because the force of natural selection declines with age (we age because there is no reason not to age). “Systemic” simply

9.2 Genetic Architecture of Senescence

means interlocking or cascade mechanisms that can give rise to one or more common aging phenotypes. There is an obvious temptation to view these two terms (i.e. stochastic and systemic) as denoting opposing viewpoints, but that would be a mistake because they are complementary, as the next section makes clear.

349

pathway (ISP) and how these interactions determine the age of onset of an animal model of Huntington’s disease. I then draw the two sets of findings together into a common mechanism of longevity regulation and senescence onset.

9.2.1 Population-level Studies

9.2 Genetic Architecture of Senescence The loss of function that underlies senescence mostly (but not entirely) involves the direct and indirect actions of oxidative damage, and I shall develop the evidence to support that statement in chapters 10–13. The question is why the longevity determinant mechanisms discussed in chapter 7 repress cellular damage for some period of time and then give way to allow the onset of senescence and the subsequent loss of function. Life span can be viewed as consisting of a health span, in which we maintain function, followed by a senescent span, in which we lose function (figure 1.7). So the question can be more precisely rephrased to ask how and why we transition from the health span to the senescent span phases of our life. We would also want to understand, at least in principle, why the senescent processes do not affect all individuals in the same way, or why different tissues and organs of a given individual lose function in their own tissue-specific manner. Some clues to these difficult questions have been obtained by independent studies, done at either the population level or the molecular level, regarding the mechanisms underlying the onset of specific diseases. I examine the genetic architecture described by these studies at both levels to find some common organization that will answer these questions. At the population level, I examine the hypertension literature as a case in point, with the goal of describing the genetic architecture of the disorder and examining the implications of that architecture. At the molecular level, I examine the genetic literature and describe the genetic architecture arising from the interactions of members of the stress resistant/molecular chaperone family with the insulinlike signaling

Figure 3.13 shows that most individuals display an increase in blood pressure as they grow older. It is widely recognized that essential hypertension is under considerable genetic influence. Monozygotic twins have a greater concordance in blood pressure that do dizygotic twins (Feinleib et al. 1977), and there is greater concordance of blood pressure within families than between families (Longini et al. 1984). The concordance within families is not due to shared environments because adoption studies show greater concordance between biological siblings than between adoptive siblings (Biron et al. 1976; Rice et al. 1989). This familial or heredity component can be viewed as comprising two somewhat different sets of genes: first, some few (~17) single genes that exert a moderately large effect, and second, a large number of recessive genes (~300–600) that have a small individual effect but when combined by inbreeding can account for a large amount of the total variation in blood pressure within a population. Evidence to support this statement is as follows: Large effects: These are basically mutations in eight different genes that cause Mendelian forms of hypertension (see figure 9.1) and mutations in nine other genes that cause Mendelian forms of hypotension (Lifton et al. 2001). Many of these Mendelian genes act early in life rather than later. Small effects: Examination of isolated village populations on the Adriatic islands of Croatia revealed a tendency toward inbreeding in each village, coupled with a tendency toward low outbreeding levels between villages (Rudan et al. 2003). Examination of about 2700 adults showed a strong correlation between the prevalence of hypertension and the coefficient of inbreeding. This translates to a rise in systolic blood pressure

350 Chapter 9 Mechanisms Underlying the Transition from Health to Senescence

Angiotensinogen Renin + Gitelman Na syndrome CI-

AI ACE AII

DCT

AII receptor

17a -hydroxylase deficiency 11ß-hydroxylase DOC deficiency 21-hydroxylase deficiency Dominant PHA1 MR Hypertension exacerbated by Cortisol pregnancy

GRA Aldosterone Aldosterone Synthase Deficiency PT

Liddle Syndrome Na+ Recessive PHA1 K+

Bartter Na+ syndrome 2CIType 1 K+ Bartter syndrome Type 2

K+

CI-

Bartter syndrome Type 3

11ß-HSD2 AME Cortisone CCT

TAL

Figure 9.1 Mutations altering blood pressure in humans, illustrated against a diagram of a nephron. The molecular pathways mediating NaCl reabsorption in individual renal cells in the thick ascending limb of the loop of Henle (TAL), distal convoluted tubule (DCT), and the cortical collecting tubule (CCT) are indicated, along with the pathway of the renin-angiotensin system, the major regulator of renal salt reabsorption. Both hypo- and hypertensive inherited diseases affecting these pathways are indicated. AI, angiotensin I; ACE, angiotensin converting enzyme; AII, angiotensin II (AII); MR, mineralocorticoid receptor; GRA, glucocorticoid-remediable aldosteronism; PHA1, pseudohypoaldosteronism, type-1; AME, apparent mineralocorticoid excess; 11 bHSD2, 11b-hydroxysteroid dehydrogenase-2; DOC, deoxycorticosterone; and PT, proximal tubule. (After Lifton et al. 2001.)

of 20 mm Hg in the offspring of first-cousin marriages, which is similar to other published studies. In the Adriatic populations studied by Rudan et al. (2003), about 36% of the hypertension variance is attributable to inbreeding (i.e., homozygosity of the small-effect genes). Altogether, the 8–16 quantitative trait loci with the largest effect account for about 25% of the variance, and the other small-effect quantitative trait loci account for the remaining 75% of the variance. All inherited and acquired forms of hypertension share an increased net salt balance as the major contributory factor in the pathogenesis of hypertension (figure 9.2). Blood pressure regulation begins with kidney function. An increase in renal salt reabsorption leads to an increase in

blood volume and thus to an increase in cardiac output. Increasing the volume of blood pumped through the vascular system leads to an increase in blood pressure, particularly if other conditions have decreased the elasticity of the vasculature. Optimal functioning of the individual requires the avoidance of both sustained high and low blood pressures. It does not, however, only involve the kidney’s renin–angiotensin–aldosterone axis, for there is pleiotropic control of blood pressure exerted throughout the body: Natriutetic peptides are produced by the brain and heart in response to high pressure in these organs; the kinin–kallikrein system of the liver and kidney affect vascular tone and salt handling; the adrenergic receptor system influences heart rate and vascular tone; and the blood vessels secrete fac-

9.2 Genetic Architecture of Senescence

351

Renal salt reabsorption

Intravascular volume

Volume delivery to heart Renal salt reabsorption Cardiac output

Autoregulation of blood flow

NI systemic vascular resistance

Blood pressure

Systemic vascular resistance

Blood pressure

Figure 9.2 A fnal common pathway for the pathogenesis of hypertension. All inherited and acquired forms of hypertension share increased net salt balance as an inciting factor. Increased intravascular volume and volume delivery to the heart augment cardiac output and therefore blood pressure. The resulting tissue perfusion exceeds metabolic demand, leading to autoregulation of blood flow via increased vasoconstriction, resulting in a steady-state hemodynamic pattern of elevated blood pressure with increased systemic vascular resistance and normal cardiac output. Nl, normal. (After Lifton et al. 2001.)

tors that promote dilation or contraction of the vasculature. These findings suggest that the age-related increase in blood pressure brought about by the recessive elements are likely to affect this same pathway, but with a tissue-specific effect. Thus, we may conclude that the genetic architecture of hypertension is structured so that only one major regulatory pathway is involved and that only a small part of the incidence of hypertension is accounted for by large-effect Mendelian genes. The bulk of the effect is mediated by many recessive genes with modest individual effects. These individual effects are often brought about by single nucleotide polymorphisms (SNPs), which are inherited variations at a single site within a gene and which have functional outcomes. For example, a linked pair of SNPs that are more frequent in hypertensives than normotensive individuals involves a change from methionine at position 235 of the angiotensin (AGT) gene (235M) to threonine (235T) coupled with a change at the -6 position of the AGT promoter from G(-6G) to A (-6A). The -6A/235T polymorphism results in a

higher plasma level of AGT and of Angiotensin II. Blood pressure increases in the absence of comparable alterations in the various inactivating and feedback mechanisms. As another example, the D allele (presumably due to an SNP) of the angiotensin converting enzyme (ACE) gene is found in individuals with a significant increased risk of onset and progression of diabetic neurophathy in type I diabetes (Takahashi and Smithies 2004). Computer simulations of the system based on existing data provide an in silico method of identifying new control points and devising new experiments to critically test theoretical interventions. The blood pressure system interacts with environmental factors that can significantly affect the body’s ability to maintain a normal blood pressure. One example is the increased glucose–protein crosslinks (discussed in chapter 10) and the resulting decreased elasticity of the blood vessels, particularly the arteries. This relationship provides the entry point by which environmental and lifestyle factors such as diet and exercise may interact with an individual’s genome to produce an increased (or decreased) risk of developing hypertension and

352 Chapter 9 Mechanisms Underlying the Transition from Health to Senescence the cascade of debiltating changes leading to the expression of the metabolic sydrome and the concomittant loss of function and independence. If the same individual carries higher risk SNPs in both the ApoE gene (discussed in chapter 8) and the AGT or acetylcholinesterase genes, then it is not difficult to understand how an environmental effect such as unhealthy nutrition could have widespread and pleiotropic effects on a variety of physiological systems. It is likely that the expression of recessive effect genes increases as age increases, and that this component of the genetic architecture accounts for much of the age or tissue-related increase in blood pressure (see figure 3.12).

9.2.2 Molecular-level Studies In all model organisms, as discussed in chapter 7, the repression of the ISP results in the activation of the FOXO gene and the activation or repression of a whole suite of downstream stress-response genes under its control. In addition, thermal stress induces the gene coding for heat shock factor-1 (HSF 1) which then induces a number of known heat shock proteins (HSPs) with a molecular chaperone function. In C. elegans, RNAi techniques (RNA interference; a method of silencing specific genes) showed that the down-regulation of HSF1 suppressed the longevity of normal–lived wild type animals as well as that of the long-lived ISP mutants (table 9.1; Morley and Morimoto 2004).

Ubiquitous overexpression of the hsf-1 gene leads to a higher HSP level in all cells of the animal at normal temperatures and to an extended longevity as well. The extended longevity seems to be most dependent on increased HSP levels in neurons and muscle. Thus, even in the absence of any overt heat shock or other stress, the HSPs must be playing an important role in both normal- and long-lived animals at normal temperatures. Despite their name, the HSPs are evidently capable of dealing with non–temperature–dependent stressors. Other evidence shows that DAF 16 and HSF 1 act independently but cooperatively in upregulating the transcriptional activity of certain hsp genes, particularly certain of the small HSPs (Walker and Lithgow 2003). Their cooperative action leads to a greater thermal resistance and increased longevity than would either pathway alone. The HSPs induced by both DAF 16 and HSF 1 maintain cell function by protecting cellular proteins from unfolding or refolding into abnormal shapes or from forming protein aggregates that harm the cell (see figure 10.2). There is some evidence that directly ties the regulation of longevity with the onset of disease. The human neurodegenerative syndrome known as Huntington’s disease exerts its harmful effects by allowing specific protein aggregates to form, collect, and eventually kill the cell. The mutant gene has an expanded repeat region that codes for a polyglutamine (polyQ) tract in part of the protein. In an animal model of this disease, abnormal aggregation of the proteins depends on the

Table 9.1 Cooperative Effect of Insulinlike Signaling Pathway (ISP) and Heat Shock Factor (HSF) on Longevity of C. elegans Treatment N2 Control (normal-lived) N2; HSF-1 RNAi N2; DAF-16 RNAi age-1, Control (long-lived) age-1; HSF-1 RNAi DAF-2, Control (long-lived) DAF-2; HSF-1 RNAi

Mean life span (days) 23.2 17.9 16.9 32.1 18.5 34.3 17.8

Presumed effect ⇓ HSPs ⇓ ADS ⇓ HSPs ⇓ ADS

Source: from Morley and Morimoto (2004). Note: HSPs, heat shock proteins; ADS, antoxidant defense system.

% of Control 100 77 73 100 58 100 52

9.2 Genetic Architecture of Senescence

353

ous stress-resistance molecules (table 9.2). Increasing the intracellular level of the protein aggregates accelerated the onset of senescence and diminished life span. As the protection conferred on the cell by these HSPs/ISP diminishes, the possibility of cell damage increases, and with it the loss of function characteristic of senescence. Figure 9.4 presents a model of a cell’s transition from the health span phase to the senescent phase, based on the overwhelming of its cellular defenses by accumulated unrepaired damage. The evidence to support his model is drawn from several authors, including Walker and Lithgow (2003) and Morley and Morimoto (2004). The cell’s major defenses include (but may not be limited to) the various stress resistance proteins such as the antioxidant enzymes (e.g., CuZnSOD) and scavengers (e.g., GSH) as well as the various heat shock

number of contiguous glutamine residues contained in the protein, with a critical threshold at about 40 glutamine residues (Q40; figure 9.3). This threshold is fairly sensitive to quantitative changes in polyQ, since Q29 animals show very little effect, Q33/Q35 animals show an intermediate effect, and Q40/Q89 animals show a rapid loss of motility relative to controls. The time it takes to attain this critical threshold is dependent on HSF-1 and ISP activities. In C. elegans, the levels of HSF-1 and consequently the HSP concentrations effectively regulate the aggregation of these abnormal proteins within cells and thus affect the organism’s longevity (Morley et al. 2002; Morley and Morimoto 2004). In addition, the long-lived age 1 mutants have a very slow accumulation of these protein aggregates, presumably as a result of high intrinsic levels of vari-

Influence of Age and PolyQ# on Toxicity

200

aggregates

Q0 150

Q29 Q33 Q35

100

Q40 Q82

50

0 0

2

4

6 8 age (days)

10

12

14

100 Q29 Q33

80

% survival

Q35 Q40

60

Q82 40

20

0 0

2

4 6 age (days)

8

10

Figure 9.3 An animal model of Huntingdon’s disease has been created in C. elegans by inserting transgenes with differing numbers of repeating glutamine residues, which cause the mutant effect. Comparison of the effect of varied Q (polyglutamine) numbers on neurodegeneration reveal a critical threshold at Q40. This implies that the cells defenses are saturated by the accumulated damage at that point. (After Morley et al. 2002.)

354 Chapter 9 Mechanisms Underlying the Transition from Health to Senescence Table 9.2 Polyglutamine Aggregation Is Inhibited by ISP Activity % Of normal-lived control aggregation at Polyglutamine level

Genotype

Day 1

Day 2

Q40 Q82

age-1; RNAi age-1, control age-1; RNAi daf-16: RNAi age-1; daf-16 RNAi

47 55 69 95 98

38 57 71 98 98

± ± ± ± ±

7 5 5 6 4

± ± ± ± ±

6 4 3 8 6

Source: from Morley et al. (2002).

Major Effect Genes

proteins (HSPs). These, along with the upstream genes regulating the expression of these stress resistance genes (e.g., HSF, the ISP genes, etc.), constitute the highly conserved (or public) major genes which exhibit a Mendelian phenotype but which collectively account for only a small portion of the variance normally observed in the phenotype. These proteins guard against oxidative stressinduced damage to cell components, facilitate proper protein folding, and ensure abnormal proteins are compartmentalized and recycled. Their

glucose

DAF-2

thermal stress ?

DAF-16

Sod-3

Antioxidant defense

Minor Effect Genes

effectiveness is impaired both by the gradual accumulation of unrepaired oxidative and protein damage, and by the loss of connectivity of the numerous modifier (or minor) genes which act in concert with the major genes so as to enhance the effectiveness of the stress resistance proteins in each cell. These modifier genes have small individual effects on the stress resistance phenotype but collectively account for most of the variance in the phenotype. Most of them are not conserved and so constitute the private genetic variations that

HSF-1

shsp

hsp-70

Maintain cell function by protecting proteins from damage &/or misfolding

Modifier genes lose their connectivity and decrease stress resistance in a tissue-and age dependent manner. Unrepaired damage accumulates, further destroying the connectivity of the gene interaction network

Senescence Begins As Cellular Protection Fades

Figure 9.4 A model of a cell’s transition from the health span phase to the senescent phase, based on the overwhelming of its cellular defenses by accumulated unrepaired damage. See text for details.

9.2 Genetic Architecture of Senescence

account for the highly individual nature of aging. When the accumulated unrepaired damage reaches a threshold at which it saturates the cell’s stress resistance genes, then any further accumulation of such damage will push the cell off this unstable equilibrium and into a positive feedback cascade in which each increment of damage causes a loss of function. To the extent that such damage causes a degradation of the gene interaction pathway, and thus a degrading of the connectivity of the major and minor genes, then to that extent each decrease in connectivity results in more unrepaired damage. Intra- and inter-specific differences in the length of the health span may reside primarily in the depth and effectiveness (i.e., functional capacity) of the cellular defenses, as well as in the differences in the accumulated damage rate. The evidence also shows that activation or repression of individual hsp genes is not as effective as is their multiple regulation by HSF-1 and DAF 16, indicating the importance of networks of chaperones. I discuss gene networks in chapter 14. In effect, this molecular analysis suggests that longevity regulation depends on the ISP/HSF major regulatory pathways, whereas the onset of senescence depends on the levels of stress resistance and of unrepaired damage in specific cells and tissues throughout the body. It is likely that shorter lived animals have lower effective levels of the downstream stress resistance and repair activities, as well as higher levels of intrinsic damage, even during their health span, than do longer lived animals. The progressive increase in the basal load on the various homeostatic mechanisms diminishes the cell’s ability to cope with new stresses, but the diminishment to some critical threshold at which the onset of senescence takes place requires more time to occur in longer lived animals than in shorter lived ones. This might account for the significant differences in health span between short-lived and long-lived animals.

355

much larger number of QTL, each with a small effect. Mutations in the large-effect QTL led to Mendelian diseases, while increasing homozygosity in the small-effect QTL led to decreased function (i.e., increased probability of hypertension). How is this view of the genetic architecture of senescence related to the view obtained from the molecular model? The large-effect QTL of the population model probably correspond to the major genes of the ISP and/or the HSF 1 pathways and to their major downstream effector genes (e.g., the various hsp genes). Mutants in any of these limited number of genes have a large effect and are inherited in a Mendelian fashion. The downstream genes in the molecular model likely correspond to the small-effect QTL of the population model. Such genes may correspond to as-yet unidentified modifier genes that interact in an unknown manner with the various major hsp genes or proteins and modify their effect or which yield tissue-, age-, or environment-specific effects on the threshold levels above which loss of function occurs. The SNPs described in the hypertension story may have their homologues here in modifiers of the various hsp genes. Taken together, these two different sets of literature data strongly suggest that both population and molecular geneticists are describing the same genetic architecture, albeit in their own specific terminology. That jargon should not blind us to the fundamental similarities. In both cases, longevity regulation relates to the loss of function and the onset of cell and organism senescence and leads to an integrated viewpoint (figure 9.4). We studied longevity regulation mechanisms in the prior chapters; we will study the mechanisms of senescence in the next four chapters. An understanding of both processes is essential to an integrated understanding of aging.

9.2.4 Age-dependent Changes in the Genetic Architecture 9.2.3 Integrating the Two Models The population model depends on the existence of a relatively small number of quantitative trait luci (QTL) each with a large effect and some

Demographers use a quadratic equation to describe the age-specific death rate (or hazard rate, in their terminology). In the analysis by Manton (1999), depicted in figure 9.5, the mortality

Mortality Function

356 Chapter 9 Mechanisms Underlying the Transition from Health to Senescence

95 68 59 50 Age Figure 9.5 The output of a linear quadratic equation used to assess changes in the mortality function and their dependence on age. The relative flatness of the curve at age 50 indicates that most of the variables contributing to mortality at that age are known. The increasing slopes of the curve at later ages indicates that an increasing proportion of the variables affecting mortality at those ages are not known. One interpretation of these data is that it is the major genes are the principal variables in premature mortality; the minor genes are the principal variables in late-life mortality. This interpretation is consistent with the concept that loss of connectivity in the gene interaction networks connecting the minor and major genes is responsible for the observed loss of function characteristic of senescence. (After Manton 1999.)

function becomes steeper as age increases from 50 to 95 years because there is a risk factor–age interaction that increases mortality risk. At age 50, this interaction is known so well that the quadratic mortality function is almost flat, indicating that there are not too many unknown factors involved in mortality determination. In the case of hypertension, we might say that it is mostly the monogenetic effects plus the environmental components (e.g., salt intake) that determine hypertension. However, at age 95, the curve is very steep, indicating that many unknown factors are involved. In the case of hypertension, we can speculate that these may be the altered expression of some of the recessive effect QTL and/or SNPs. If we could successfully identify these factors in each individual and then use appropriate biological processes to counteract their effects, the hazard curve at age 95 would be rendered less steep and would more closely resemble the age 50 curve. This is exactly

the promise inherent in “personalized medicine,” whereby analysis of an individual’s SNPs would allow the identification of the particular risk factors early in life, thus allowing the deployment of individualized defenses to forestall their contribution to the age-related rise in mortality. Such a personalized small-effect strategy may be a useful complement to the more global anti-aging interventions, such as caloric restriction mimics, discussed in chapter 7. If the small-effect component of hypertension or the molecular chaperone response may be modulated in this manner, then it appears as if the reduction in expected mortality is brought about by the retention of optimal gene function, particularly in the minor-effect genes. The gene expression network paradigm discussed in chapter 14 is the most likely context within which the major gene–minor gene architecture common to both descriptions operates and is embedded. The loss of optimal function in such a

9.3 An Integrated Theory of Aging and Senescence

genetic architecture may arise from the loss of connectivity in the gene expression/interaction network (see, e.g., figure 13.4). The logical conclusion of this common senescent pathology is that its genetic architecture is both complex and plastic. It allows for major and minor genetic and environmental effects. Senescence consists in losing the ability, bit by bit, to decrease net salt balance or to maintain chaperone activity at some critical level. The number of genes involved, to say nothing of their different allelic states and epistatic interactions, allows us to understand why senescence is such an individual process. If we assume that other physiological processes have a similar genetic architecture (but one involving different sets of genes), then we have a working hypothesis as to why different tissues within the same individual may show quite different rates of senescence. Maintaining the functional state of the minor genes involved appears to be one way to delay the onset of senescence, a conclusion that was brought out by the data of chapter 7 and one that will be further reinforced by the data presented in the following chapters.

9.3 An Integrated Theory of Aging and Senescence 9.3.1 What Causes the Transition from Health to Senescence? Applying evolutionary principles allows us to understand that the life span of an organism is shaped by its genes but that the details of its senescence will flow from the interaction over time of those gene products with the organism’s internal and external environment. The discussion of longevity-determination mechanisms provided some insight into the processes responsible for maintaining the high functional levels characteristic of youthful healthiness. The senescent mechanisms I discuss in the next section provide more insight into the processes that degrade and destroy that high level of functioning and bring about the more obvious signs of aging. How can we reconcile the presence in

357

our bodies of these two diametrically opposed mechanisms? At the end of chapter 1, a bare-bones conceptual model of aging was presented based on the concept that the life span of an organism is composed of the health span and the senescent span. Figure 9.6 presents the outlines of an integrated theory of aging and provides a conceptual framework on which to incorporate the various facts and concepts. We each attain a state of maximum functional ability early in life, characterized by the processes and states indicated in the figure. Whether we maintain that state for a long or short time depends on the nature and activity of our genetically based longevity-determinant mechanisms and their interaction with our environments, as follows. In the health span, highly conserved (public) longevity determinant mechanisms preferentially operate so as to continue optimal gene expression and cell function during this phase. The genes involved in these processes constitute the “major genes” often described by population geneticists, and which often have Mendelian phenotypes but constitute only a small proportion of the variance in the phenotype. In the transition phase, the accumulated unrepaired damage to the cell eventually overwhelms the cells stress resistance mechanisms (ADS plus HSPs) which operate both independently and cooperatively as shown (see figure 9.4 for details). The senescent phase is characterized by the progressive degradation of the gene interaction networks by various types of public and private mechanisms. Loss of function characteristic of senescence coincides with the loss of specific connectivities. The senescent mechanisms described in the following chapters are the primary mechanisms contributing to this degradation. The gene interaction networks have characteristic connectivities and characteristic (public) failure modes. These latter give rise to the most common types of systemic failures, and these constitute the common diseases (e.g., the most common failure modes of the cells’ regulatory systems). The private mechanisms give rise to unique or rare individual modes of failure. Taken together, these private and public degradation modes constitute the “minor genes” described by population geneticists, and

358 Chapter 9 Mechanisms Underlying the Transition from Health to Senescence

HEALTH SPAN

+

LIFE SPAN = TRANSITION +

SENESCENT SPAN

Longevity Determinant Mechanisms

Altered Balance of Cell’s Defenses Due to Accumulated Damage

Degradation of Gene/Protein Interaction Networks

Homeostatic Ability Sensitive & Reliable

Abnormal Proteins Aggregate, Exceed Chaperone Capacity; Positive Feedback ⇒ Lowered Cell Function

Cell’s Regulatory Ability Decreases, Tissue/Systemic Functions Deteriorate Feedback Cascades Ruin Homeostatic Ability, Critical Thresholds Passed

Low ISP Levels High Antiox Levels High Repairs Levels

Damaged Cells Survive, Apoptosis decreases, Tumors increase

High OxDam Level High Inflammation Level Low Stress Resistance

Figure 9.6 An update of the life span model presented in figure 1.7. This version sums up the data of the intervening chapters which provides evidence for the existence of a ‘health span’ ‘senescent span’, coupled by a shorter period of transition from health to senescence. See text for details.

which often have individually small additive effects but which together account for most of the variance in the phenotype. Aging is a cell level phenomenon. The presence of senescent cells within a functioning tissue degrades the collective functioning of that tissue and makes it less capable of maintaining homeostasis. When sufficient cells cascade below some critical threshold, tissue or organ failure ensues. The pathways involved have been discussed (table 7.12), and the outcome of their continued operation is to maintain integrative and repair mechanisms at a high level of function during the health span. One can postulate that much of the difference between a short-lived mouse and a long-lived human resides in the depth and diversity of cellular defense mechanisms, as exemplified by their major (large-effect) genes, and this would involve both quantitative as well as qualitative differences in these defenses. These would be part of the species-specific genetic heritage common to all individual mice or humans. Eventually these pathways lose their effectiveness as the level of unrepaired damages increase to some critical level, as discussed in chapters 10–14. The increased stochastic damage brings about the loss of regulatory functions both within and between cells. Bad things begin to happen, and the only available interventions are those that retard (but

do not necessarily reverse) the loss of function characteristic of the senescent span. Eventually the functional level of the system falls below that minimum threshold compatible with life, and we die. The foregoing sequence adequately describes the aging process of most but not all individuals. The phenomenon of epigenetic stratification (see figure 10.11) suggests that some small proportion of the population will undergo connectivity changes that will actually increase or at least maintain their functional level. Perhaps some centenarians whose longevity was not inherited are an example of such positive changes. In any event, it seems that much of the difference between short-, normal-, and long-lived members of any one species likely resides in the heterozygosity (SNPs) of the numerous minor (modifier) genes that indirectly affect the activity of the major genes.

9.4 An Overview of Senescent Mechanisms There are several different ways to organize the different types of senescent mechanisms. Hayflick (1985) grouped them according to their level of action, writing of organ-based, physiology-based, and genome-based theories. Hart and Turturro

9.4 An Overview of Senescent Mechanisms

(1983) wrote of cell-based, organ-system–based, population-based, and integrative theories. Esposito (1987) wrote of causal, systematic, and evolutionary theories. In addition, theories of senescence could be sorted according to whether the causes each one postulates are supposed to arise from systemic cascade processes or from stochastic and random processes. There are probably other ways to sort the theories and mechanisms of senescence. None of these systems is fully satisfactory, in large part because the complexity of the aging processes suggests that the mechanisms involved can be sorted into more than one valid scheme. However, the origin of the change and its level of action both appear to be reasonable pegs from which to hang our descriptions. I use a dual classification scheme. I consider whether the theories suggest that particular effects are exerted within all or most cells (intracellular theories) or whether they are exerted mostly on the structural components and/or regulatory mechanisms linking groups of different cells (intercellular theories). In addition, I consider whether the effects postulated by each theory are conjectured to take place accidentally (stochastic theories) or are the result of the hierarchical feedback cascades characteristic of the species (systemic theories). As we will see, there are classification difficulties even with this simple scheme. Finally, I introduce the evolution in thought that now leads us to construct a smaller number of

much more inclusive theories, such as the stress theory of aging (Parsons 1995). Table 9.3 presents a classification of the 14 senescent mechanisms that I examine in some detail in chapters 10–13. These mechanisms have been classified according to how they fit best into the dual criteria of the location of the changes and the nature of their effects. These criteria are useful empirical guides; they are not exhaustive theoretical principles. Accordingly, a particular theory can just as logically be assigned to another cell. For example, I have classified DNA damage mechanisms as stochastic, but one might logically counter that the amount of damage is a function of the species’ DNA repair ability and that these mechanisms appear to be genetically determined and therefore systemic. Or one might suggest that the free-radical or oxidative-damage mechanisms can be viewed equally well as affecting the biochemical processes within each individual cell (and therefore should be classified as intracellular), or as affecting the regulatory processes that affect the metabolic relationships of various cells, tissues, and organs (and therefore should be classified as intercellular). The fact that this exercise reveals how difficult it is to unequivocally file mechanisms into pigeonholes suggests that the mechanisms are larger than the somewhat arbitrary compartments into which we are forcing them. But the classification scheme does have the advantage of allow-

Table 9.3 Classification of Senescent Mechanisms Origin of Change Level at which effect is executed

Stochastic

Systemic

Intracellular

Altered proteins Somatic mutations DNA damage and repair Error catastrophe Dysdifferentiation Free Radicals Waste accumulation

Metabolic mechanisms Genetic mechanisms Signaling mechanisms

Post-translational protein changes Wear and Tear

— Neuroendocrine Immunological

Intercellular Extracellular Multicellular

359

360 Chapter 9 Mechanisms Underlying the Transition from Health to Senescence ing us to engage in a simple linear discussion of the several mechanisms, postponing our integration of them until the end. This integration will involve the concept that the body normally is maintained by a network of processes that operate in parallel with one another and together provide a homeostatic system, as illustrated in figure 9.6. Table 9.4 summarizes these mechanisms and their present status. The original cell-based mechanisms viewed aging as the gradual wearing down of the somatic cells to the point at which they could not adequately discharge their functions. This failure was viewed as leading directly to the functional decrements characteristic of aging. In the early 20th century, the experiments of Alexis Carrel (1912) were thought to demonstrate that somatic cells grown in vitro are immortal, so the earlier wearand-tear theories fell into disfavor. Hayflick and Moorhead (1961) corrected Carrel’s work and demonstrated that normal cells grown in culture can undergo only a limited number of mitotic divisions. This observation inspired the development of numerous genetically oriented, cell-based theories of aging—theories that use either stochastic or systemic mechanisms to explain the observed impairment of cell function. The metabolically based mechanisms postulate intrinsic changes in metabolic functions as the source of the cell’s loss of function. And the conceptual foundations of the various genetic mechanisms of senescence also have their origins in the concept of autonomous systemic gene action within each cell of the organism. The homeostatic, integrative, or intercellular mechanisms are all based on the assumption that there are one or more pacemakers of aging—particular tissues or organs that initiate the onset of the events characteristic of senescence. (The term “pacemaker” is derived by analogy with the pacemaker cells of the heart, which initiate each heartbeat and are responsible for its rhythmicity.) In principle, such initiation could stem from either a positive or a negative signal. In either case, Shock’s empirical observations (Shock et al. 1984) that aging is characterized primarily by the body’s failure to maintain homeostatic equilibrium among different organ systems seems to lend much weight to this class of mechanisms. The highly integrated state of our neural, hormonal,

and immunological functions (see chapter 5) is certainly consistent with (but does not prove) this point of view. Earlier I identified a major paradox of gerontological theory: If one postulates stochastic causes, then one must account for the apparent predictability and systemic aspects of aging; yet if one postulates systemic causes, one must also account for the extraordinary variability in the course of aging among different individuals. It has been apparent for some time that certain theories of senescence are less different from one another than our classification scheme might suggest. For example, the rate-of-living theory was recognized as more or less congruent with the metabolic theories, and these latter theories have been recognized as being restatements of the freeradical theory. The free-radical theory has been recognized as one component of a generalized stress theory of aging. Many of the theories could be organized in a hierarchical scheme. In addition, it has been recognized that the mechanisms postulated by these different theories do not take place separately but rather interact within the organism. Thus, the free-radical and the post translational protein change, or glycation, interacts in a synergistic manner such that oxidative damage enhances the probability of glycation taking place (because the processes described in figure 10.1 are accelerated by free radicals; Kristal and Yu 1992). Thus, consideration of the interaction of different mechanisms acting at different organizational levels within the organism leads us inevitably to an integrative outcome. Chapter 14 presents a more detailed description of these and other integrative models that have been proposed in the past few years. In the meantime, it will be beneficial to keep in mind a dynamic view of the cell: It depends on the proper functioning of a dynamic gene interaction network, protected from oxidative damage by the effects of a dynamic antioxidant network and an integrated but complex system of repair enzymes, and integrated into the operation of all other cells by the neuroendocrine system, which faithfully lets each cell know the state of the external environment and thus what general energetic strategy it should adopt.

9.4 An Overview of Senescent Mechanisms

361

Table 9.4 Overview of Senescent Mechanisms Theory

Major Premise and Current Status

Altered proteins

Time-dependent, post-translational change in molecule brings about conformational change and alters enzyme activity. This affects cell’s efficiency. Occurrence in structural proteins leads to decreased elasticity and thus to functional physiological changes. Known major role.

Somatic mutation

Somatic mutations alters genetic information and decrease cell’s efficiency to subvital level. Simplistic interpretations no longer viable, but this may be the proximate result of free radical damage and thus of the occurrence of age-related neoplasms and other types of damage. Obviously related to the DNA damage/repair mechanisms described below. Known major role.

DNA damage and DNA repair

Cell contains various mechanisms that repair constantly occurring DNA damage. The repair efficiency is positively correlated with life span and decreases with age. Obviously related to somatic mutation mechanism described above. Known major role.

Error catastrophe

Faulty transcriptional and/or translational processes decrease cell’s efficiency to subvital level. Disproven.

Dysdifferentiation

Faulty gene activation-repression mechanisms result in cell’s synthesizing unnecessary proteins, possibly changing its differentiated state, and thus decreasing cell’s efficiency to subvital level. Known minor role.

Free radicals

Longevity is inversely proportional to extent of oxidative damage and directly proportional to antioxidant defense activity. Known major role.

Waste accumulation

Waste products of metabolism accumulate in cell and depress cell’s efficiency to subvital level if not removed from cell or diluted by cell division. Unlikely in its original form. Updated versions involving the accumulation of misfolded protein aggregates (e.g., Huntington’s disease) are more accurately described as involving post-translational protein changes, as below.

Post-translational protein changes

Time-dependent chemical cross-linking of important macromolecules (e.g., collagen) impairs tissue function and decreases organism’s efficiency to subvital level. Aggregation of important proteins plays a role in various age-related neurodegenerative diseases. Proven major role.

Wear and tear

Ordinary insults and injuries of daily living accommodate and decrease organism’s efficiency to subvital level. Proven minor role in restricted cases (e.g., loss of teeth leading to starvation), but modern reformulations are part of other theories.

Metabolic mechanisms

Longevity is inversely proportional to metabolic rate. Disproven in its orginal form but reformulated into the mitochondrial free radical theory and that reformulation appears to be correct. Proven major role.

Genetic mechanisms

Changes in gene expression cause senescent changes in cells. Multiple mechanisms suggested. May be general or specific changes. May function at intra-or intercellular level. Proven major role. (continued)

362 Chapter 9 Mechanisms Underlying the Transition from Health to Senescence Table 9.4 (continued) Theory

Major Premise and Current Status

Signaling mechanisms

Failure of intracellular signaling processes is an inextricable part of other mechanisms, particularly the genetic and metabolic mechanisms listed above. Apoptosis, for example, is induced by both extra and intracellular signals. Failure to induce or repress apoptosis probably responsible for a variety of diseases, although its role in nonpathological aging changes not clear. Failure of intercellular signaling mechanisms is a major component of the neuroendocrine and immunological mechanisms summarized below. Proven major role as an integral component of other processes

Neuroendocrine

Failure of cells with specific integrative functions brings about homeostatic failure of the organism, leading to senescence and death. Proven for female reproductive aging and other specialized cases. Effect is dose dependent. Known to effect changes in immune function. Proven major role.

Immunological

Life span is dependent on types of particular immune system genes present, certain alleles extending and others shortening longevity. These genes are thought to regulate a wide variety of basic processes, including phagocytosis of foreign, infected, or damaged cells as well as regulation of neuroendocrine system. Failure of these feedback mechanisms decrease organism’s efficiency to subvital level. Proven major role.

Much of the evidence I consider in the following chapters is correlative. There are good practical reasons for this, but the convenience comes at a price. Such data can suggest only that two variables are linked; the data cannot prove the causal nature of the relationship. Another complication arises from the nature of much of the genetic evidence. The absence of a factor (as in a null mutation) often results in abnormal pathology and a drastically shortened life span, but such evidence cannot be construed as proving that the factor is therefore sufficient to extend life span. Not many mutations result in the overexpression of the factor of interest, but this is exactly the condition needed for us to determine if a surplus of the factor will result in a slowing of the mortality rate doubling time and an extension of the life span. Many variables may be

necessary for life but may not be sufficient by themselves to bring about an extended life span. The final estimation of any senescence mechanism will depend in part on the acuity of analysis of the data and in part on how well data obtained from different levels of analysis are integrated. Finally, an explanation such as presented in figure 9.6 should also account for the failure of a complex, self-repairing organism to overcome the structural and functional defects that characterize aging. We will solve this conundrum only by an unwavering reliance on empirical observations and on critical testing of our concepts. The answers we develop should be consistent with evolutionary data, bearing in mind Dobzhansky’s famous (because it is accurate) statement (1973): “Nothing in biology makes sense except in the light of evolution.”

10

Stochastic Theories of Aging

10.1 Can Predictable Events Be Brought about by Stochastic Processes? There is a predictable regularity in the aging process, in the sense that a certain pattern of describable physiological changes occur with advancing age. It may seem improbable that a process as orderly as aging appears to be is due to random, or stochastic, factors. Yet this idea is the basis of theories of senescence postulating that the deterioration associated with old age is due to the accumulation of random molecular damage. Such faulty macromolecules could accumulate through two somewhat different mechanisms: failure to repair stochastic damage or stochastically caused error in macromolecular synthesis. In either case, information vital to the cell is lost, and the progressive accumulation of faulty macromolecules eventually reaches the point at which some or all cells of an organism are so metabolically crippled that systemic death results. Although these theories provide a mechanistic explanation for the observed age-related declines, two additional assumptions must be made if one is to rationalize how stochastic damage brings on systemic changes. The first assumption is that the cell or tissue or organism in question has specific classes or types of molecules that are particularly sensitive to certain types of damage. The second assumption is that long-lived species are better able to tolerate molecular damage than are short-lived species; in other words, long-lived species have either a better repair system or greater functional redundancy than do short-lived

species. Given these assumptions, one could equally classify such information-loss theories as systemic, for they view organisms as formulated to fail in a predictable and species-specific manner as a result of exposure to random insults. This discussion of the hidden assumptions highlights the complexities of the processes that are implicit even in what superficially appears to be a simple concept of aging.

10.2 Stochastically Based Theories 10.2.1 Wear-and-Tear Theory The wear-and-tear theories of aging are probably the oldest precursors of the concept of failure to repair. These theories persist probably because they are unconsciously reinforced by our everyday observations. All organisms are constantly exposed to infections, wounds, and injuries that are likely to cause minor damage to cells and tissues and organs. Such structural erosion and minor injuries might contribute incidentally to an age-related decline in functional efficiency. August Weismann (1891b) thought this gradual wearing down of the somatic cells as a result of use was the major cause of aging. Today we would strenuously disagree and could present three reasons for doing so. First, animals raised in an environment that protects them from minor insults and pathologies not only still age but also fail to show any improvement in their maximum life span. Second, many of the

363

364 Chapter 10 Stochastic Theories of Aging minor damages postulated by the wear-and-tear hypothesis are time-dependent changes only, and although they can certainly increase the probability of death for any individual, they cannot logically serve as a causal mechanism of the aging process. A lost tooth does not initiate aging. Finally, and most important, the theory is outdated. Advances in our knowledge of cell and molecular biology have generated the need to explain cellular and organismic aging in more precise terms. The modern reformulations of the failure-torepair hypothesis more convincingly explain particular aspects of biological aging than does the unmodified wear-and-tear hypothesis. For example, alterations in the texture of the cartilage of joints (see chapter 5) might be ascribed to wear and tear, but pigeonholing those changes gives us no conceptual framework of how to understand the process better or how to intervene in it. However, thinking of this alteration in cartilage texture as being, in part, the result of age-related changes in the expression of the genes that code for proteoglycan may help guide our thinking to a deeper level of understanding. As we shift from considering gross anatomical levels of organization to considering a more cellular and molecular level, discarding the wear-and-tear idea for more mechanistic explanations will be increasingly supported by the data. However, to the extent that these modern theories contain some aspect of the wear-and-tear idea, one could view all of the intracellular stochastic theories as conceptual descendants of this original idea.

10.2.2 Post-translational Protein Modification Theory One variation of the failure-to-repair class of theories is the idea that the time-dependent accumulation of irreparable chemical modifications in important macromolecules prevents the affected tissues from functioning normally and thereby is a plausible mechanism of aging. This idea was first put forth by Björksten and Champion (1942) and independently by several others. Such post-translational modifications are

known to occur in both proteins and nucleic acids, so their effects could be widespread. It has been known for quite some time that extensive age-related chemical changes leading to cross-linking of once separate fibrils take place in connective-tissue components such as collagen and elastin. These changes also appear to be associated with well-described functional decrements in skin and other tissues (see chapter 5). In fact, the previous discussion of cardiovascular aging suggests that one of the few intrinsic agingrelated changes in that system is increased stiffness of connective fibers in the arteries, causing an increase in systolic blood pressure, leading to other pleiotropic effects on the cardiovascular system. Several different techniques have been devised to show that the properties of connective tissue fibers such as collagen change with age; studies using these various techniques have been reviewed by Sell and Monnier (1995). Because one-third of the protein content of a typical mammal is collagen, the age-related changes that take place in this family of molecules have intimate effects on every aspect of our being. The aging of connective tissue changes our shape, size, nimbleness, and our ability to live independently. Collagen isolated from old rodent and human donors is more difficult to digest enzymatically than is collagen from young donors, probably because of cross-linking between the different collagen strands. Tail tendons of rats and mice contract after being heated and denatured in a warm salt solution. The rate of contraction, as well as the amount of weight needed to inhibit this contraction, increases with age. Such observations can be understood as the result of the greater stiffness and structural complexity imparted to the collagen fibers by the cross-linking. The fact that denaturation continues even in collagen fibrils removed from the organism and stored in vitro (but in air) suggests that collagen cross-linking is an intrinsic oxygen- and timedependent process. In situ, however, the process can be significantly modulated by the physiological state of the animal (e.g., diabetic or nondiabetic). Comparison of rodent strains with different longevities shows that collagen tends to age faster in the shorter lived strain than in the

10.2 Stochastically Based Theories

365

Perhaps the best known process that has been shown to give rise to cross-linked proteins in vivo is the nonenzymatic and irreversible reaction of proteins with glucose to form advanced glycosylation end products (AGEs; figure 10.1; Cerami 1985; Cerami et al. 1987; Vlassara et al. 1988). It may be best to view AGEs as constituting a post-translational modification of collagen by sugar. This modification begins with the nonenzymatic condensation of a sugar aldehyde by means of a free amino group of a protein to yield a Schiff base. This base then is rearranged to yield the more stable Amadori product, which can react further with other such proteins to form the cross-linked AGEs. Despite certain technical concerns, it is generally accepted that the levels of AGEs increase significantly with age in humans and in experimental animals (Sell and Monnier 1995). Type I diabetics appear to have significantly higher amounts of AGEs than would be expected on the basis of their chronological age alone, implying that elevated blood glucose levels accelerate the rate of production of crosslinked AGEs. In contrast, excepting diabetes, the

longer lived strain, and collagen condition generally seems to correlate well with either the animal’s chronological age or the strain’s maximum life span, but not with its mean life span (Harrison et al. 1978). There are several different chemical types of cross-links. One type consists of intermolecular cross-links between lysine residues in different collagen helices. These cross-links likely serve as precursors to even more complex cross-links among the several molecules, and some of these derivatives are among the most abundant species in human skin. Not all cross-links increase in number with age. Reducible cross-links such as those involving certain histidine residues are most numerous in the fetus or young organism and decrease thereafter. Nonreducible cross-links, which are formed by oxidation and include much of the fluorescent material seen in collagen in situ tend to increase significantly with age in most organisms tested (Sell and Monnier 1995). Other forms of collagen cross-links may arise as a side effect of the noxious products formed as a result of lipid peroxidation.

Glucose

CH2OH

N

(98%) Lactone

CH—(CHOH)4—CH2OH (2%) Aldehyde +

Extracellular proteins (collagen) Vascular proteins Lens crystallins

Slow Spontaneous Protein

Protein

NH2

N

C—R

Schiff Base Protein

Cross-linked Protein

N—C—R

Advanced glycosylation end products

N—C

R

Amadori rearrangements

R N Protein

Figure 10.1 Processes that lead to glucose-based cross-linking of individual protein molecules. See text for discussion. (Drawing by G. T. Baker III.)

366 Chapter 10 Stochastic Theories of Aging available evidence suggests that these age-related changes in long-lived proteins predispose one to, rather than cause, age-related diseases (Sell and Monnier 1995). Nonetheless, this predisposition likely arises because the old and damaged tissue is now more vulnerable to trivial insults than is young and undamaged tissue. AGE structures can be recognized and destroyed by macrophages as part of a process that also induces neighboring tissue cells to replace the destroyed structures with non–cross-linked molecules. This process of normal tissue remodeling is thought to become unbalanced with time, resulting in the age-related accumulation of crosslinked proteins and accompanying functional deficits. Many age-related disorders that involve collagen cross-linking occur at a younger age in diabetics, and Cerami (1985) has suggested that the reason is the chronic high glucose concentrations of diabetics. Moreover, there is evidence (see chapter 6) that extrinsic factors such as caloric restriction or exercise can impede the crosslinking process in the extracellular collagen fibers. Thus the external and internal environments can modulate the rate of connective-tissue aging. If AGE-dependent cross linking plays a major role in the pathological stiffening characteristic of diabetes and aging, agents that inhibit that process would be useful. A new class of thiazolium-based derivatives that can break established AGE crosslinks has been developed, and the most promising one (ALT-711) was initially tested on dogs. After 1 month of oral administration, healthy older dogs showed a significant (~40%) decrease in left ventricular stiffness (see chapter 5), such that their end-diastolic pressure/volume profile moved closer to (but still significantly different from) that of healthy younger dogs (Asif et al. 2000). Similar types of improvements were noted in various tissues of both normal and diabetic rats (Vasan et al. 2003). Phase 2 human clinical trials were recently completed. Older people with stiffened cardiovasculature taking ALT-711 had a significant reduction in arterial pulse pressure and an increase in large artery compliance compared to placebo controls (Vasan et al. 2003). Other phase 2 clinical trials are underway to determine if ALT711 might provide a useful therapy to directly

improve left ventricular diastolic function. These preliminary data demonstrate the proof of concept that AGE-dependent cross-links have a deleterious physiological effect that can be reversed by means of specific cross-link breakers. The connective tissue that holds us together and gives us shape contains molecules other than collagen; the most prominent are elastins and proteoglycans. Similar sorts of age-related posttranslational modifications seem to be taking place in these molecules as well, although there are major differences in the protein structure and chemistry involved, to say nothing of the difficult technical problems in analyzing them. I will not explore those aspects further (see Sell and Monnier 1995), except to note that the general precepts developed for collagen appear to apply to these other molecules as well. Similar nonenzymatic cross-linking reactions take place between glucose and intracellular proteins such as hemoglobin. Of particular interest, food-restricted rats show a statistically significant slower increase in their level of glycosylated hemoglobin than do ad libitum-fed controls (Masoro et al. 1989). Other important intracellular macromolecules, in particular DNA, are also susceptible to glucose-mediated cross-linking (Bucala et al. 1984). This process might underlie the DNA damage that brings about age-related changes in gene expression discussed later in this chapter. However, in contrast to the extracellular collagen molecules, genetically important macromolecules such as DNA possess an efficient repair system, and any understanding of genetically based senescent changes must take this into account.

10.2.3 Altered Protein Theory I discussed the altered protein theory in chapter 4 when I pointed out that work with the nematode had first shown that various purified enzymes obtained from older animals appeared to have been altered in some of their immunological and thermal-stability properties (see tables 4.5 and 4.6). An abundance of evidence shows that these alterations do not involve errors in the sequence of their component amino acids, nor do

10.2 Stochastically Based Theories

they involve the types of covalent changes that would accompany the chemical modification and cross-linking of the preexisting protein (Rothstein 1983). Because the enzymes are made correctly, and because they are not irreversibly chemically altered thereafter, the only other reasonable explanation must be a conformational change in the old enzyme—in other words, a change in its shape. The evidence shows that the old enzyme has undergone a shape change that can be reversed by denaturation and renaturation (Sharma and Rothstein 1980). Similar procedures done with other altered enzymes (e.g., phosphoglycerate kinase) in other species (e.g., the rat) have yielded comparable data and comparable conclusions (Gafni and Cook 1988; Hardt and Rothstein 1985; Yuh and Gafni 1987). Presumably, the altered enzymes are long-lived molecules (i.e., have a low turnover rate) and reside in the cell for such a long time that they are subtly denatured by the cytoplasmic environment. The regularity of this denaturation is such that at least two proteins (a-crystallin and glyceraldehyde-3phosphate dehydrogenase), which undergo characteristic alterations, are being used as potential biomarkers of aging in an ongoing mouse study (A. Gafni, personal communication). Comparisons of enzymes purified from young and old animals show that they may also vary in their carbonyl content, with the older animals having a higher level of these oxidized protein products. Berlett and Stadtman (1997) have calculated that the oxidized protein content in an old animal might represent 30–50% of the total protein content. In addition, the catalytic activity of many enzymes decreases by 25–50% in older animals. Because the actual amount of protein usually remains unchanged, the level of inactive enzyme agrees roughly with the level of oxidized protein. In Berlett and Stadtman’s view, oxidative modification is a unifying concept that allows us to understand the alterations of proteins during aging. Altered proteins, then, can be viewed as representing a special case of the cross-linking concept, involving conformational changes that are either reversible (use hydrogen bonds to change shape) or irreversible (use co-

367

valent bonds to attach extraneous groups). It is important to note that nonenzymatic proteins (e.g., transcription factors) also undergo conformational changes, with deleterious effects on the cell (Heydari et al. 1993). Protein processing appears to slow down with age, apparently because of unknown changes in the cytoplasmic pathways of degradation, and this underlies the age-related increase in the level of altered proteins (Van Remmen et al. 1995). The existence of these altered proteins provides excellent support for the concept of molecular aging.

10.2.4 Somatic Mutation and DNA Damage Theories The idea that chromosomal abnormalities might underlie the aging process is a more specific version of the failure-to-repair concept. This idea dates back to the influential paper by Szilard (1959) in which he postulated that speciesspecific life spans might be the result of speciesspecific rates of random hits that inactivate large chromosome regions or even entire chromosomes. This concept flowed out of the earlier work done on the effects of radiation on cells. Szilard’s suggestion renewed interest in the genome as a controlling factor in the aging process. Of course, mutations affecting the germline cells (the sperm and ova) might well result in abnormal offspring, but they could have nothing to do with the aging process in the parent, since only the parents’ somatic cells age. Hence, this group of theories focuses on damage done to the genomes of somatic cells only. It seems reasonable that processes that destroy the integrity of the somatic-cell DNA would also cause a loss of function in the affected cells and tissues. DNA may suffer two basic types of alterations: mutations and damage. The two are not synonomous. Mutations are changes in the polynucleotide sequence such that the standard AT or CG base pairs are deleted, added, substituted, or rearranged. These mutational changes often affect, sometimes seriously, the information coded into that portion of the genome. For example, the difference between a normal hemoglobin

368 Chapter 10 Stochastic Theories of Aging and an abnormal sickle-cell hemoglobin is due entirely to the effects of a single substitution of an adenine for a thymine at the 17th base position in the beta-hemoglobin gene that encodes that protein. Regardless of the type or extent of the mutational alterations, and regardless of its effects on the informational content of the genome, the affected DNA molecule retains its characteristic double-helical structure and still consists of an uninterrupted sequence of nucleotide pairs. It still looks like a normal molecule. DNA damage, on the other hand, refers to any of many chemical alterations in the double-helical structure of the molecule. The damage may be caused by either exogenous or endogenous sources; the latter are perhaps more important. These alterations produce structural irregularities that interrupt, modify, or break the double helix. Examples of such irregularities include but are not limited to pyrimidine dimers, apurinic sites, single-strand breaks, adducts, and covalent crosslinking of the DNA strands to one another or to other molecules. Even if none of these forms of damage alters the informational content of the DNA molecule, they do structurally interrupt the DNA molecule. The effects of mutations and DNA damage are similar but not identical, and the processes by which they manifest their results are different. Both can interfere with gene expression; on this basis they have been independently proposed as possible mechanisms of aging. I discuss them below as independent theories. 10.2.4.1 Somatic mutations

If mutational mechanisms play a role in the aging process, they could arise through a germline mutation in the parent that would then give rise to an offspring in which all body cells contained a somatic mutation. In fact, one of the basic theories of aging is the mutation accumulation theory first put forth by Medawar (1946, 1952; see chapter 4), and this theory has an impressive amount of empirical support. The early realization that several inherited pathologies believed to mimic certain segmental aspects of aging, such as Down’s syndrome, are caused by chromosome abnormali-

ties contributed to a restored interest in genetic causes. I review the evidence for this theory beginning with the very large and visible genome alterations and then move to the molecular level. Jacobs and colleagues (Jacobs and CourtBrown 1966) conducted the first investigation of the effect of aging on human chromosome number and found a significant increase in the number of lymphocytes that exhibited visible chromosomal abnormalities as a function of age. Later studies verified these initial observations. Similar sorts of chromosome abnormalities have also been observed among “old” (late-passage) cells grown in vitro (Saksela and Moorhead 1963). Jarvik (1988) suggested that cells of individuals afflicted with Alzheimer’s disease might have chromosome abnormalities. Chromosomes from older humans appear to be more fragile than those from younger individuals; the rate of aminopterin-induced breakage is higher in older chromosomes than in younger ones (Esposito et al. 1989). In most, but not all, human embryos aneuploidy is lethal. This lethality stems from the fact that the loss of one or more chromosomes usually alters a cell’s functions so severely that it is no longer viable. Thus, it might appear plausible that such stochastically caused chromosome damage causes the observed age-related declines in function. However, the frequency of aneuploid cells is low even in older healthy people. For example, the most common abnormality found in lymphocytes is the loss of a chromosome. In one study, such affected cells increased from a frequency of 3% in youths aged 5–14 years to more than 9% in people 65 years old and older. Though the threefold increase is certain, it is not clear whether that difference is sufficient to explain the difference in functioning between a 5 year old and 65 year old. Moreover, the missing chromosomes often involved the X chromosome, an interesting observation because the loss of one sex chromosome often has no deleterious effect in females, and such hypodiploidy is most often found in females (Schneider 1985). In addition, such chromosome loss is not necessarily seen in other body cells (Finch 1990). The data suggest that chromosome loss does not play a substantial role in bringing about the aging phenotype in normal individuals.

10.2 Stochastically Based Theories

A more recent study showed that abnormally low levels of a particular mammalian protein (BubR1) involved in one of the cell’s mitotic checkpoints allows the progressive development of aneuploidy in a variety of dividing tissues (Baker et al. 2004). Normal mice show no such decline in their BubR1 levels nor in the efficiency of their mitotic checkpoints. One-day-old mice with the hypomorphic protein showed no aneuploid mitotic figures. But the frequency increased gradually so that 12-month-old (middle-aged) mice had a 33% frequency of aneuploid mitotic figures. The progressive failure of their mitotic checkpoint allowed for the accumulation of abnormal cells, and the presence of these abnormal cells is correlated with the premature aging and short life span characteristic of these hypomorphic BubR1 mice. Taken together, these results suggest that the very low level of aneuploid cells in normal individuals has no observable effect on the aging processes, but abnormal genetic (or environmental) situations that allow the accumulation of cells with abnormal chromosome numbers might trigger cell senescence and organismal aging. The findings imply the existence of some threshold above which tissue function is adversely affected. Somatic chromosome lesions can be most easily recognized in dividing cells, and this fact made such cell studies a popular experimental approach. Curtis (1963) and Curtis and Miller (1971) demonstrated an inverse relationship between the rate of increase of chromosomal lesions in regenerating liver cells and the species-specific life span of the three species (mouse, guinea pig, and dog). G. Martin et al. (1985) examined the chromosomes from enzymatically dispersed kidney cells shortly after they were added to culture (early-passage cells) obtained from young (8 months) and very old (40 months) mice. Their results showed that the frequency of a variety of chromosomal lesions is substantially elevated in the older animals. I discussed Werner’s disease in chapter 8. Cells taken from individuals afflicted with this segmental mimic of aging have a limited life span in culture (G. Martin et al. 1965). The reason for this shortened in vitro life span may be related

369

to the fact that the chromosomes of these cells are extraordinarily unstable and undergo particular chromosome rearrangements and/or loss (Salk et al. 1981). In turn, the chromosome instability is probably a result of mutations in the normal WRN gene, which is a DNA helicase type of DNA repair gene (G. Martin and Oshima 2000; Yu et al. 1996). Studies on other normal tissues in vivo have revealed that many tissues show no evidence of visible chromosome damage, yet they still age. Visible chromosome abnormalities are not widespread, so it is difficult to entertain them as potential causative factors of aging. In fact, a review of this topic concluded that the chromosomal aberrations must be considered the effect, not the cause, of aging (Sen et al.1987). Not all chromosome damage need result in visible abnormalities (Ames et al. 1993; De Flora et al. 1996). Many, perhaps most, somatic mutations involve molecular alterations of DNA that are sufficient to irreversibly alter the information coded therein but that do not affect the structure. Such a mutant gene would be incapable of producing a normal gene product. Szilard proposed in 1959 that the accumulation of such somatic mutations constitutes the elementary step in the aging process. This hypothesis was based on an analogy with the known effects of radiation. Indeed, H. J. Muller received the Nobel prize because he demonstrated that exposure of experimental animals to X-rays induces both germline and somatic mutations. How could one detect such mutations and critically test the hypothesis? There are several viable approaches, four of which I discuss here. First, one could attempt to detect the abnormal gene products directly and see if there is an age-related increase in their frequency among all members of the test population. A clever utilization of existing data was adopted by Popp and colleagues (1976). The human alpha- and betahemoglobin chains have been completely sequenced, and their amino acid composition is known. The amino acid isoleucine is not normally encoded genetically in either of these polypeptide chains. Inclusion of isoleucine into the hemoglobin molecule thus would be indicative of

370 Chapter 10 Stochastic Theories of Aging a synthesis probably due to a somatic mutation in a few cells of that organism. Isoleucine assays were done on the hemoglobins of two groups of individuals: those accidentally exposed to radiation (from an atomic bomb test) as youths and those who had never been exposured to radiation. The unexposed individuals showed no significant age-related increase in isoleucine content. The exposed individuals showed significantly higher levels of isoleucine content in their hemoglobin. These data suggest that somatic mutations are readily induced by high levels of irradiation but do not increase with age, as demanded by the theory. Clark et al. (1963) took another approach to testing the somatic mutation theory by looking at different forms of the wasp Habrobracon. Their experiment is instructive because it illustrates the principle that modern biogerontology is data based; thus the apparent result of an experiment may be completely overturned if important facts were inadvertently underappreciated. The truth really does lie in the details. The wasps used in this experiment had an unusual genetic system that allows the construction of males that are either haploid (n) or diploid (2n); that is, they are thought to differ only in the number of chromosome sets each contains. This fact allowed Clark et al. to set up a critical test of the theory. If the accumulation of somatic mutations is the cause of aging, the haploid animals should age faster, and have a shorter mean and maximum life span, than the diploid animals. This relationship should hold whether or not both sets of animals have been irradiated. This situation would arise because a somatic mutation occurring in the haploid organism would inactivate the animal’s only copy of that gene, but would still leave a backup copy of the gene in the diploid animal. The results were striking. The nonirradiated haploid males had the same life span as that of the nonirradiated diploid males. However, the irradiated haploid males had a significantly shortened life span compared to the irradiated diploid males. These classic data have long been interpreted by many as directly contradicting the predictions of the somatic mutation theory. Woodruff and Thompson (2003) reviewed the Clark et al. study and pointed out two im-

portant facts that strongly argue against the traditional interpretation. First, many, perhaps even most, tissues in wasps have cells with extra chromosomes. Thus, some large number of the wasp’s somatic cells are neither haploid nor diploid but are polyploid. Equivalent normal life spans would be expected in such animals, and that is what is seen. Their second objection to the experiment is the fact that diploid male wasps are sterile, and changes in reproduction are known to affect the rate of aging (see chapters 4 and 7). These objections are serious enough for us to reclassify this classic Habrobracon experiment as not being robust enough to critically test the somatic mutation hypothesis. Woodruff and Nikitin reinvestigated the somatic mutation theory using the technique of P-element insertional mutagenesis (Nikitin and Woodruff 1995; Woodruff and Nikitin 1995). This technique has an advantage over radiation in that it allows the experimenter to induce defined and controllable numbers of random, single mutations. Woodruff and Nikitin found convincing evidence that the induction of somatic genetic damage can reduce life span in Drosophila but that the effect is dependent on the particular species of fly involved and the type of transposable element used. The literature suggests that even in the heterozygous condition, recessive mutations can reduce life span by 1–2%, and synergistic interactions between different deleterious mutations have the potential to further increase this effect to a statistically significant level. This possibility was tested by Woodruff and Thompson (2003), and their data indicate that the accumulation of undetected somatic mutations could, under special circumstances, reduce longevity by as much as 22% over a period of 16 generations in Drosophila. These data suggest a direct relationship between increased somatic genetic damage and the age of onset of senescence. Thus, the empirical data are mixed, but the more recent information tends to support the idea that the accumulation of molecular-level mutations in somatic tissues plays a major role in bringing about an early aging syndrome. This conclusion is upheld by investigations of DNA damage and repair. The maintenance of genome integrity

10.2 Stochastically Based Theories

seems to be a major factor in understanding the molecular basis of senescence, particularly in long-lived species such as humans. 10.2.4.2 DNA damage and repair

DNA damage is not a rare event in mammalian cells. The data in table 10.1 suggest that the number of damaging events is so high that without repair mechanisms, within a few years or so the affected cells would not be able to transcribe or to replicate their DNA accurately or completely. Measurements made by Ames et al. (1993) indicate that there are probably more than 10,000 oxidative hits of DNA per cell per day in humans, a value in general agreement with those of table 10.1. Fortunately, the specific DNA repair mechanisms within the cell are normally more than sufficient to repair the ongoing damage. For example, the typical neuron would lose in its lifetime about 3% of the total number of adenine or guanine (purine) bases in its DNA solely as a result of depurination were it not for the existence of an apurinic (AP) repair system that specifically attacks and repairs AP sites in variety of organisms (Gensler and Bernstein 1981). Each type of damage usually has a specific type of repair system that enzymatically recognizes and removes the damaged portion of the DNA and synthesizes a new patch using the nonaltered opposite strand as a template; these systems are reviewed in detail by Demple and Harrison (1994), Dusenbery and Smith (1996), and O’Connor (2000). However, these systems do not constitute the cell’s only level of repair.

371

The general repair mechanisms associated with recombination can also play a role in repairing non-recombinational damage such as interstrand cross-links and double-strand breaks. In addition, studies have shown that DNA repair is linked to its transcription; actively transcribed strands are repaired at a higher rate than are nontranscribed strands (Bohr and Anson 1995). This mechanism is an additional safeguard against damaged DNA generating damaged gene products. Because unrepaired lesions are potential mutation sites in replicating DNA, it is important that DNA damage be repaired before cell replication. Such repair is accomplished by a block in the cell cycle before DNA replication that involves activation of the p53 tumor-suppressor gene, as well as other antiproliferation genes (see figure 7.8). Extensive DNA damage can activate apoptosis, the intrinsic cell death program that eliminates potentially neoplastic cells, which I discuss further in chapter 12. As depicted in figure 10.2, cells and organisms normally exist in a dynamic equilibrium between the rate of DNA damage and the rate of DNA repair. Should the net rate of damage increase for any reason, then the balance will be reset to a new, perhaps dangerously high, level of DNA damage. The major predictions of the DNA damage theory of aging are that (1) there is a positive correlation between DNA repair ability and life span, and (2) a systemic, age-related shift in repair activity is a major factor underlying the appearance of the functional decrements characteristic of aging. The underlying assumption of the DNA damage theory is that all DNA damage is related to

Table 10.1 Damage and Repair Kinetics of Mammalian DNA Type of damage Pyrimidine dimers in skin (noonday/sun) Thymine glycols O6-methylguanine Depurination (cleave out A/G) Depyrimidation (cleave out T/C) Single-strand breaks in DNA Source: after Tice and Setlow (1985).

Rate of appearance (events/hour) 5 × 104 13 130 580 29 2300

Rate of repair (events/hour) 5 × 104 (normal cells) 5 × 103 (Xeroderma pigmentosum group C cells) 1 × 105 1 × 104–105 — — 2 × 105

372 Chapter 10 Stochastic Theories of Aging

Normal DNA

Damaged DNA Repair

Death mutation

Figure 10.2 The level of damaged DNA depends on the relative rates of input of damage and repair and/or cell death.

levels of DNA repair and/or to levels of functional performance. This relationship may be true in many cases, but it need not be true in all cases. For example, DNA damage might occur in a critical cell type, yet not be related either to an agerelated functional decline or to changes in the level of DNA repair. For a somatic mutation to give rise to cancer requires simultaneous changes in all the other systems that monitor and modulate cell growth, and so such a progression is a low-probability event. Even if the cell undergoes apoptosis as a consequence of the cell’s regulatory systems, the affected cell is usually embedded in a network of functional redundancy such that the cell’s loss of function and/or death would be compensated for by other members of the network. The loss of function of the particular cell might well impinge on the organism’s level of functional redundancy, but this would be noted only as a loss in cell number or as a decrease in future performance, neither of which is being measured. It would, however, show up as a gradual loss of homeostatic ability, and such loss is considered to be the essence of aging (see chapter 1). In mammals, the daily rate of DNA damage can be conveniently determined by an assay of the animal’s urine for DNA damage products. Ames (1994) has shown that the species-specific urinary output of thymine glycol and thymidine glycol is directly proportional to the species’ oxygen consumption: Humans have the lowest damage rates; mice have the highest. Because oxygen consumption and free-radical production are re-

lated (Sohal and Weindruch 1996), it seems likely that oxidative damage is the major cause of DNA damage. This hypothesis provides a plausible mechanism of DNA damage, but it does not address the question of DNA repair. Evidence supporting the prediction of a correlation between species-specific life spans and DNA repair ability was first put forth by Hart and Setlow (1974). Their interspecific data (figure 10.3) showed a striking correlation between the species-specific repair ability and the characteristic species life span. Subsequent work has generally (Cortopassi and Wang 1996), but not always (see Kato et al. 1980), tended to support this initial observation. The relationship appears in groups other than mammals as well; DNA repair systems have been found in representatives of all five kingdoms of living organisms. SmithSonneborn (1979) showed that the clonal life span of unicellular paramecium is reduced when these organisms are exposed to ultraviolet light,

Human

100

Elephant 60

Cow

30

Life span (years)

Endogenous or exogenous damage

10

6 Hamster Rat 3 Mouse Shrew 1

0

10

20 30 Grains/nucleus

40

50

Figure 10.3 The correlation between life span and the ability of fibroblasts from various mammalian species to conduct unscheduled DNA synthesis (that is, DNA repair), as represented by the average number of grains per nucleus. (Data from Hart and Setlow 1974, as redrawn by Tice and Setlow 1985.)

10.2 Stochastically Based Theories

Human

100

Elephant 50 Cow Life span (years)

an inducer of thymine dimerization in DNA. However, when UV irradiation is followed by photoreactivation, a process that reverses the dimerization, the clonal life span is not only restored but even increased. These results suggest that the DNA repair enzymes are induced by UV and are capable, once the excess dimers are removed by photoreactivation, of subsequently repairing other unspecified but accumulated DNA damage that limits the life span. This interpretation is supported by the positive and linear correlation between the number of AP sites in the paramecium genome and the clonal age of the population, indicating that the accumulation of DNA damage is age related (Holmes and Holmes 1986). Since we may regard the mitotic cells in our bodies as spatial or temporal clones, the findings on paramecium may have a wider relevance. Prevention and/or repair of DNA damage may result in an extended clonal age, which, if the cells involved play a critical role in the organism, may translate into an alteration of the aging phenotype. UV photoreactivation may not play a major role for cells that are embedded in the darkness of our bodies, but an analogous situation of inducible DNA repair may occur in multicellular forms as well. Drosophila contains an enzyme called recombination repair protein 1 that, when experimentally overexpressed, reduces the frequency of somatic mutations and repairs certain types of DNA damage induced by oxidative stress (Szakmary et al. 1996). One interesting set of interspecific data, depicted in figure 10.4, shows an inverse correlation between the species-specific life span and the cell’s ability to convert harmless substrates into DNA-damaging agents. One interpretation of this data set is that metabolic activities that ordinarily detoxify harmful chemicals can on occasion convert those chemicals into active substances that may bind to DNA and cause a pattern of agerelated damage. These data also suggest that this trend is evolutionarily selected against in longlived species. Such detoxification systems may be a pleiotropic component of a long life. A review of various interspecific studies indicates that a good but not excellent relationship between relative DNA repair activity and maximum life span

373

Rabbit 10 Guinea pig 5 Rat Mouse Hamster 1

1,000 1,200 400 800 600 200 Rate of benzo[a]pyrene conversion (pmol/4 hours)

0

Figure 10.4 The abilities of fibroblast cultures from various mammals to convert benzo[a]pyrene to a water-soluble form. This metabolic activation of the hydrocarbon gives rise to a potentially dangerous molecule. Note the inverse relationship between life span and the activation ability. (Data from Moore and Schwartz 1978, as redrawn by Tice and Setlow 1985.)

potential (Cortopassi and Wang 1996). Cortopassi and Wang suggested that DNA repair is a necessary but not a sufficient condition for long life, a view in agreement with past studies but one which perhaps should be extended by the more recent report of Symphorien and Woodruff (2003). One cogent objection that has been raised to interspecific studies that measure a particular variable against the species’ maximum life span is that maximum life span is the outcome of the annual rate of mortality and the annual rate of increase in the population, and as such it constitutes a poor measure of the extent of senescence present in the population (Promislow 1993). However, the interspecific studies are trying to correlate a particular variable (e.g., DNA repair) with the extent of senescence in the populations. The studies’ conclusions may still be correct; the point is that they often use an inaccurate measure of senescence, in part because the better measure (annual change in the age-specific rate, or qx) simply is not known for many species. Thus

374 Chapter 10 Stochastic Theories of Aging we need to accept the results of most interspecific studies with a measure of uncertainty. What evidence supports the second of the theory’s predictions—namely, that a systemic, age-related shift in repair activity is the major event underlying the appearance of the functional decrements characteristic of aging? Intraspecific experiments can best address this question. Early experiments done with human fibroblasts taken from different-aged individuals showed differences between young and old, suggesting that aging affects the DNA repair parameters measured (Goldstein 1971; Painter et al. 1973). Many later studies confirmed this early conclusion, but the reliance on correlative studies and the tendency to investigate specific DNA repair systems made it difficult to draw robust conclusions on the overall relationship between DNA repair and aging (e.g., Vijg and Knook, 1987). Measuring the number and types of tissue-specific mutations accumulated over time in a transgenic mouse model seemed to present a more informative approach. Dolle et al (1997) showed that in such a model, there was about threefold increase in the frequency of point mutations in mouse liver from infancy to old age, compared to about a twofold age-related increase in the brain. Additionally, the frequency of large deletion mutants was very low and only increased about twofold with age in the liver, while actually decreasing in the brain. The existence of tissue- or organ-specific DNA mutation rates may explain some of the apparently contradictory results of the prior literature (see table 10.2). For example, brain and

liver differ in their mitotic behavior, and this alone might account for the different behavior of deletion mutations in these two tissues. Also, there is no obvious way to ascertain the effect of these different levels of mutation accumulation on the function of the brain and/or liver, and without such a functional measure we really cannot say much about the effects of somatic mutations on aging. Finally, it has been suggested that the heterogeneity of results on overall DNA repair ability is misleading in the sense that the ability of a cell to function properly may depend on its ability to selectively repair only those genes that are essential to its survival and the accomplishment of its differentiated task. Thus, transcription-coupled DNA repair in differentiated cells would preferentially repair genes active in that tissue type but allow mutations in nonactive genes to accumulate. If so, then the measure of mutation accumulation may be functionally meaningless. A review of the studies above and other studies led Warner and Johnson (1997) to three conclusions: First, many DNA alterations increase with age; second, any specific DNA alteration seldom reaches what would be expected to be a critical level; and third, mutations require cell proliferation to be fixed in a population of cells. Even though the results of these decades of work may be debatable and not capable of proving the somatic mutation hypothesis, neither are they capable of disproving it. This situation arises in part because each of the studies focused on investigating one particular DNA repair mechanism, and so no one may have grasped the dynamics of

Table 10.2 Organ-Specific Age-related Decreases in DNA Repair Ability Age-related decreased repair ability at agea Type of DNA repair

18 Months

28 Months

Excision repair Single-strand break repair Double-strand break repair Gamma-irradiation repair

Spleen, lung No change No change Kidney, lung, testes, brain

Spleen, lung, liver, testes, kidney, heart Testes, brain No change Liver, kidney, lung, testes, duodenum, brain, muscle

Source: after Niedermuller (1985). aIn all cases, the value of the 9-month-old animal to repair damage done in each specific organ is taken as the standard, and the 18 month and 28 month are expressed relative to that standard.

10.2 Stochastically Based Theories

the overall system. There are more than 100 different types of oxidative DNA lesions alone, and we know that cells have evolved an intricate network of DNA damage repair pathways, each focusing on a particular type of lesion (see figure 7.8). It is unlikely that such an energydemanding conserved repair process would be maintained if it did not fill some important evolutionary task. Genetic studies afford better insight into the relationship between DNA repair and longevity. If DNA repair ability were essential to normal longevity, mutants that decrease or abolish DNA repair activity should decrease the animals’ life span, while mutants that increase DNA repair activity should increase their life span. The former relationship was first reported in Drosophila by Whitehead and Grigliatti (1993). These investigators used temperature-sensitive (ts) alleles of three mutagensensitive (mus) genes. Animals carrying such mutations are very sensitive to mutagens, usually because of defects in DNA repair pathways. The interesting attribute of ts mutants is that the affected gene product performs normally when the organism is raised at the permissive temperature (22°C) but is inactivated at the restrictive temperature (29°C). The authors found that the mus mutants raised and/or maintained at 29°C had shorter life spans than the controls, showing that the loss of normal DNA repair capacity leads to a reduction in longevity. A more rigorous experiment which tested both predictions was done by Symphorien and Woodruff (2003), in which they measured the

375

life spans of flies with variable numbers of a specific DNA repair gene (mei-41). Flies with no active copies of this gene had a significantly shortened life span relative to flies with one active copy. In a different experiment, flies with an extra copy of the mei-41 gene had a longer life span than did the normal fly. Although the latter differences were not large, they are in the right direction. DNA repair activity was not actually measured but was assumed to directly vary with gene dose. To my knowledge, this is the first experiment to show that life span of a multicellular animal can be both shortened by an apparent decrease in DNA repair activity and increased by an apparent increase in DNA repair activity. It supports the idea that genomic integrity as measured by DNA repair activity is both necessary and sufficient for long life. Observations showing that loss of DNA repair function leads to a premature aging syndrome have been made in various mammalian models of premature aging. Table 10.3 lists human segmental progeroid syndromes that accelerate some but not all aspects of usual aging. Many of these heritable mutations involve the inactivation of proteins that sense or repair DNA damage. These findings suggest that failure to maintain genome integrity underlies at least some accelerated aging phenotypes. In mice, genetic alterations involving nucleotide excision repair, transcription-coupled repair, and/or nonhomologous end-joining (NHEJ) of double-strand breaks can lead to signficant reductions in life span and other signs consistent with an accelerated aging (see table 10.4). In the KU86

Table 10.3 Genetic Defects Associated with Human Segmental Progeroid Syndromes Syndrome

Inheritance

Trichothiodystrophy Hutchinson-Guilford Cockayne Ataxia telangiectasia Werner’s Down’s Rothmund-Thomson Xeroderma pigmentosum

Autosomal recessive Unknown Autosomal recessive Autosomal recessive Autosomal recessive Trisomy 21, de novo Autosomal recessive Autosomal recessive

Mean life span ~10 ~13 ~20 ~20 ~50 ~60 Normal? (↑Cancer incidence)

Source: from data presented in de Boer et al. (2002) and in Hasty et al. (2003).

Genetic defect DNA repair, basal transcription Laminin A defect Transcription-couple DNA repair DNA damage signaling protein kinase DNA helicase, exonuclease Gene overdose? DNA helicase Nucleotide excision repair

376 Chapter 10 Stochastic Theories of Aging Table 10.4 Genetic Defects Associated with Accelerated Aging in Mice Alteration Phenotype

Wrn KO

Ku80/86 KO

Xpd HM

Life span Cellular senescence Sensitivity to carcinogens Skin atrophy Weight loss Genetic function altered

↓ ↑ ↑ No data No data DNA repair DNA helicase

↓ ↑ ↑ Early Early DNA repair NHEJ

↓ No data ↑ Early Early DNA repair DNA helicase

Ercc1 KO ↓ ↑ ↑ Early Early DNA repair NER endonucleae

p53 DM ↓ No data No data Early Early DNA repair

Source: adapted from table 2 of Hasty et al. (2003). Note: KO, knock-out; HM, hypomorphic mutant; DM, dominant active mutant.

mice, for example, the increase in the age-specific mortality rate associated with the onset of senescence occurs at about week 56 in the normal-lived control mice, but at about 3 weeks in the homozygous knock-out mutant mice (Vogel et al. 1999). Mice with defects in any of the several genes whose products are involved in NHEJ repair exhibit similar phenotypes, some of which strongly resemble those of rare human mutations (Hasty et al. 2003). Mice with a mutation in the XPD gene, which is homologous to the gene responsible for the human trichothiodystrophy syndrome (see table 10.3), also develop normally until adulthood but then show an accelerated onset of senesence (de Boer et al. 2002). There are seven distinct xervoderma pigmentosum (XP) genes, and XP mice carrying mutations in both the A and D genes show enhanced cellular sensitivity to oxidative stress. Oxidative stress is intimately involved in the senescent processes (chapter 7); hence the synergistic effects of the double mutants provide indirect evidence of causal links among oxidative stress, DNA damage, and the accelerated onset of senesence seen in these several mutants. I return to this topic in the discussion of the dysdifferentiation theory of senescence below. Much data suggest that damage to mitochondrial DNA plays an important role in some aspects of aging. I discuss the relationship between mitochondrial damage and aging in chapter 11, where it is considered under the rubric of the mitochondrial free-radical theory of senescence.

Damage to nuclear DNA likely contributes to the aging process, at least in certain tissues. The review of the data strongly suggests that the absence of DNA repair ability probably has a causal relationship to the expression of a shortened life span and accelerated senescence. The experiment of Symphorien and Woodruff (2003) shows that apparent overexpression of DNA repair activity may result in an extended longevity, but such a result seems to be more difficult to bring about as the model organism becomes more complex (see Sinclair 2002). This difficulty may be a consequence of the increased complexity of mammals relative to flies relative to yeast; or it may simply reflect the difficulty of getting most laboratory animals into an optimal low-stress state in which the alteration of this one variable can significantly alter life span. Either interpretation is informative, for it tells us that if and when a pharmecutical-based longevity extension becomes realistic, then it will probably not be possible for such an intervention to work other than in the context of a healthy organism in which all necessary systems are already at their optimal low-stress state.

10.2.5 Dysdifferentiation Theory It has been suggested that normal aging of an organism results because its component cells drift away from their proper state of differentiation.

10.2 Stochastically Based Theories

Differentiated cells are characterized by their ability to selectively repress the activity of genes not necessary to the survival of the cell and its particular functions. An alteration in the specificity of these activation–repression mechanisms, perhaps brought about by time-dependent changes in the cell’s internal milieu, could theoretically interfere with the cell’s finely tuned ability to carry out its specified function. Cutler (1985, 2003) coined the term “dysdifferentiation” to denote this process and suggested that the resulting lack of stringent gene control might result in the cell’s synthesizing proteins other than those characteristic of its differentiated state. The theory postulates that the mechanisms that result in altered gene expression patterns during senescence are identical to the processes responsible for the changes in gene expression that take place during development (Cutler 2003). It is unlikely that the mechanisms responsible for senescent changes are absolutely identical in detail to those responsible for developmental changes. In addition, the data of chapter 7 show that the genome is very stable. But as restated by Cutler (2003), the theory now contains an interesting kinetic element: the “dwell time” in the genome of an unrepaired somatic mutation, and I focus here on this aspect. Organisms with a low rate of DNA repair might allow a mutation to exist in the genome for a long time before repairing it. But if the mutation’s effect is to indirectly destabilize the gene interaction network that dynamically regulates the cell’s gene expression profile, then the initial perturbation of this network by the initial appearance of the mutated gene product might well shift it into a nonoptimal (i.e., nonyouthful) state. If this new state of the network is as stable as the original state, then it might not be possible for the cell to shift its gene expression pattern away from the nonoptimal state. This would be especially true if the new nonoptimal state allowed sufficient other damage to occur in other parts of the network. Even if the cell eventually repaired the somatic mutation which initiated this cascade, its network effects would continue to persist such that the organism would continue in its nonoptimal senescent state. Under these circum-

377

stances, organisms with a very active DNA repair system might have a lower average dwell time for somatic mutations than organisms with a less active DNA repair system. If the probability of any somatic mutation being able to destabilize the existing gene expression network is related to the length of time during which the altered gene or protein is present in the network, it follows that organisms with a short dwell time should have longer health spans and a later onset of senescence than organisms with a long dwell time. It also follows that the same phenotypic traits should be found in organisms with a low rate of somatic mutation production in the first place, relative to those with higher mutation rates. So we can best view the dysdifferentiation theory of senescence as arising from the intersection of the organism’s abilitiy to resist oxidative stress, its ability to repair genomic and/or DNA damage, and the stability of its gene interaction networks at that point in its health span. Like the error catastrophe theory, the dysdifferentiation theory has the virtue of making specific and testable predictions, for it postulates that cells from old donors will synthesize more proteins that are uncharacteristic of its particular differentiated state than will similar cells from young donors. (This is not the same thing as stating that the same tissue type taken from animals of different ages will display a different panorama of proteins.) Such changes might best be detected by metaplasia—or the presence in one tissue of differentiated cells or tissue-specific proteins that are usually characteristic of another tissue. It is not uncommon to find such small nodes of metaplastic tissue in organs of older individuals. The organs of the elderly often seem to exhibit transformation of related cell types, such as intestinal metaplasia in the stomach or gastric metaplasia in the small intestine. Although they are often benign, metaplasias are suspected to give rise to transformed cells under certain conditions. Hartman and Morgan (1985) showed that certain types of metaplasias may be induced in laboratory rats after treatment with known mutagens. These mutagens cause heritable changes in somatic cells. The treatment is believed to have given rise to somatic mutations, although no

378 Chapter 10 Stochastic Theories of Aging cytogenetic analyses were performed in this study. The induced mutations presumably inactivate the normal functioning of mature, normally differentiated cells. These mutations probably have little effect on the organism’s ability to function because of the normal reserve capacity of the remaining cells. However, mutations that are presumed to affect a stem cell and alter the final differentiation pathway of that cell and of all its descendants are likely the ones that give rise to the observed metaplasias. Thus a stem cell normally destined to give rise to gastric cells would be altered such that it would now give rise to intestinal cells. High numbers of metaplasias in any single tissue could disrupt the integrity of that tissue and lead to the physiological decline characteristic of aging. Some metaplasias might progress from a benign state to a cancerous state. Other cancers originate in a multistep sequence of specific somatic mutations. Both humans and mice have a high probability of developing cancers during their senescent phase. The general physiology of both mammals is so similar that we depend on mice to provide us with an accurate surrogate description of ourselves. But humans live some 30 times longer than mice (90 years vs. 3 years). Furthermore, human cells in culture are much more resistant to cancerous transformation than are mouse cells. It is now generally considered that part of the explanation for these differences is that long-lived animals such as humans have both a much more effective antioxidant defense system and more effective DNA repair system than do short-lived mice. There is, for example, a highly significant positive correlation (r = 0.84, P<.001) across 13 mammalian species between their species-characteristic maximum life span and the activity in younger individuals of a molecule that is an early and important responder to certain types of DNA damage, poly(ADPribose)polymerase (PARP; Burkle, 2000, 2001; Grube and Burkle, 1992). Humans have the highest PARP activity levels; rats have the lowest. Thus we produce significantly fewer reactive oxygen species molecules with mutagenic capability than do rodents, and we are probably more efficient at repairing any ensuing DNA damage

than are rodents. It is logical to conclude that humans probably have a shorter dwell time than do mice with regard to any somatic mutations produced. Considering the various theories of senescence against the background of gene interaction networks allows us to construct a more cohesive and dynamic mental image of how the cell might slip from a healthy state into a senescent state. Adding the elements of time and interactions allows us to see the relationships of different senescent damage processes with the cell’s various senescent defense mechanisms.

10.2.6 Error Catastrophe Theory The error catastrophe theory was originally developed by Orgel in 1963 and differs from the somatic mutation and DNA damage theories in that it postulates an error in information transfer at a site other than in the DNA. The basic idea behind this theory is that the ability of a cell to produce its normal complement of functional proteins depends not only on the correct genetic specification of the various polypeptide sequences, but also on the competence and fidelity of the protein-synthesizing apparatus. Even if the genome contains neither somatic mutations nor DNA damage and the organism has an accurate copy of the correct species-specific information for a particular protein, the organism will reap no benefit if it garbles the information in the process of translating it. Consider the general path of information flow in biological systems, as shown in figure 10.5. Errors in type I products yield a somewhat lower average efficiency in a particular aspect of metabolism or of structure but would leave no lasting trace once the product was turned over. An example of this situation might be a person in whom some hemoglobin had inadvertently suffered a one-time error and had been made with isoleucine instead of leucine. Such abnormal molecules probably would be less efficient, perhaps strikingly so, in binding oxygen and carbon dioxide than would normal molecules. However, the abnormal molecules would be cleared from the body when the old red blood cells that con-

10.2 Stochastically Based Theories

DNA

RNA

Protein

379

Molecules with structural and metabolic uses (Type I)

Molecules used in synthesis of nucleic acids (Type II) Figure 10.5 The flow of information from genome to gene products. Gene products are classified into one of two categories depending on whether they are involved in information replication. See text for discussion.

tained them were destroyed in about 90 days. They would have no lasting effect. Errors in type II products can be subdivided into two categories: those that lead to a complete loss of function (type IIa) and those that lead to reduced specificity of an information-handling enzyme (such as a DNA or RNA polymerase; type IIb). An error in the type IIa class is similar to the type I error in that it affects the efficiency of the information transfer process by reducing the number and kinds of molecules replicated. The extent of damage depends on which part of the translation process is affected, but such an error does not affect molecules that are synthesized. If the molecules that are not synthesized are absolutely required for life and cannot be substituted for by other molecules, and if their concentration in the cell falls below a minimum threshold, then cell death will likely ensue. Errors in type IIb products, giving rise to a reduced fidelity or specificity of translation, are very different; even one such error inevitably leads to an exponentially increasing error frequency. Many, if not all, of the molecules synthesized by defective type II enzymes are themselves defective. A single defective enzyme molecule in the cell amplifies its effects throughout the cell with every round of synthesis. These errors are transmissible and cumulative and might ultimately lead to what Orgel (1963) called an “error catastrophe.” An error catastrophe results when the error frequency reaches a value at which one of the processes necessary for the existence of a viable cell becomes critically inefficient. A vital threshold has been crossed, and the cell dies. If enough cells die, the result is a decrement of physiological function characteristic of aging.

This theory is both logical and reasonable, but its best virtue is that it makes a specific and testable prediction: Proteins obtained from cells of old people should exhibit a significantly higher frequency of errors than proteins extracted from cells of young people. Such protein populations are easily and accurately examined experimentally by means of a molecular separation technique, twodimensional polyacrylamide gel electrophoresis (2D-PAGE). This procedure sorts out the proteins by molecular weight and by electrical charge, depositing the isolated proteins on an x–y grid. It is very sensitive, allowing investigators to accurately detect even small numbers of proteins that contain a change in charge and/or molecular weight. Such changes are indicative of errors in synthesis. The 2D-PAGE method made the error catastrophe theory one of the few theories of aging that could be critically tested at the time it was put forth. The 2D-PAGE test has now been applied to a wide variety of different species and cell types, including E. coli, Drosophila, C. elegans, and human fibroblasts. The results are quite clear: There is no evidence of the predicted electrophoretic heterogeneity that is characteristic of synthesis errors in the proteins obtained from older people. Not only is there no difference in error rate between young and old, but all cells show remarkably good fidelity in synthesis, and all have error rates that are so low as to be almost undetectable. It is unlikely that such transcriptional or translational errors are one of the mechanisms responsible for aging and senescence (Rothstein 1987). However, this evidence does not necessarily rule out errors in the replication of the DNA itself. Murray and Holliday (1981) showed that

380 Chapter 10 Stochastic Theories of Aging certain DNA polymerases obtained from older cells have an increased error rate relative to those from younger cells. The loss in fidelity is less severe in calorically restricted mice than in ad libitum-fed animals (Srivastava et al. 1993). This observation is formally consistent with the error catastrophe theory, and Holliday (1984) skillfully supported it by wedding this theory to the disposable-soma theory (see chapter 4), whereby with aging the cell must divert an increased proportion of resources from reproduction to error correction.

10.2.7 Oxidative Damage Theory Except for those few organisms that are specially adapted to live under anaerobic conditions, all animals and plants require oxygen for the efficient production of energy. About 95% of total metabolic energy is produced in the mitochondria, the cellular organelles that combine the carbon molecules obtained from the digestion of our food with the oxygen obtained from breathing. If these reactions are blocked, we lose consciousness and die very quickly. Oxygen is essential to the energyproducing reactions that keep us alive. However, oxygen supplied at concentrations greater than that of normal air is toxic to many plants and animals. Exposing E. coli to pure oxygen causes them to stop growing immediately. Oxygen enhances the damaging effects of ioniz-

ing radiation on living cells: A threefold higher dose of radiation is needed to kill cells in a nitrogen atmosphere than is needed to kill the same cells in an oxygen atmosphere. Exposure of humans to pure oxygen for as little as 6 hours causes chest soreness in some individuals, and longer exposures eventually lead to irreversible damage to the alveoli of the lungs. There are many other examples of oxygen damage, for almost all tissues of many organisms are affected, as table 10.5 shows. The damaging effects of oxygen on organisms vary depending on the species, on the particular tissues examined, and on the individual’s age, physiological condition, and diet. Nonetheless, oxygen toxicity appears to be a general phenomenon and may reflect a fundamental biological mechanism. Ironically, the molecule on which we depend for life is also the molecule that can harm us. Various explanations have been put forward to explain oxygen toxicity. The widely accepted current explanation is that most of the damaging effects of oxygen are the result of cellular damage caused by free radicals. These highly reactive, naturally occurring chemicals are induced in the presence of oxygen and have been implicated in more than 60 disorders, including heart disease, cancer, and cataracts. They also appear to be one of the major factors responsible for the changes in body structure and function that are characteristic of aging and senescence. The initial presentation of this theory (Harman 1956) was based

Table 10.5 Effects of High Oxygen Concentrations on Animal Tissues Species

Exposure

Organ examined

Results

Rats

Pure O2 at 5 atm for 75 minutes

Heart

Rats Hamsters

Pure O2 at 0.33 atm for 3 days 70% O2 for 3–4 weeks

Liver Testes

Guinea pigs

70% O2 for 6–36 weeks

Monkeys

Pure O2 at 0.5 atm for up to 22 days

Bone marrow development Liver

Mitochondiral swelling followed by myofibril damage Mitochondrial damage Degeneration of seminiferous epithelum; cessation of sperm production Inhibition of red blood cells

Humans

Hyperbaric oxygen therapy

Ear

Proliferation and abnormality of smooth endoplasmic reticulum; decrease in glycogen content Hemorrhages of inner ear; deafness

10.2 Stochastically Based Theories

in large part on the observed inverse relationship between life span and metabolic rate for mammals and the inverse effect of temperature on the life span of poikilotherms, as described in chapters 7 and 11. The data that have been obtained since the theory was first stated are always plausible, usually convincing, and in some cases definitive. The oxidative-damage hypothesis is one of the major theories providing a probable and testable biological mechanism for the senescent process. There is a wealth of literature on this topic, which, along with the encyclopedic books by Halliwell and Gutteridge (1999) and Cutler and Rodriguez (2003), should be consulted for further information. A free radical is defined as any chemical species that has an odd number of electrons. Most chemical compounds have paired electrons and thus are only moderately reactive; they require specific starting conditions or enzymatic assistance to react chemically with another substance. Molecules that have an unpaired electron, however, are thermodynamically unstable; they are highly reactive as they seek to combine with another molecule to pair off their free electron. Free radicals are produced in the cell by various mechanisms, including exposure to toxic agents (such as oxygen, radiation, and pollutants or drugs) and enzymatic processes that produce and release free radicals in vivo. Common to all of these mechanisms is the ability to provide enough energy to break the covalent bond that holds the two atoms or molecules together. If A and B are two atoms covalently bonded, and : represents the pair of electrons constituting the bond, the result of normal breakage of this bond—that one atom receives both electrons— can be represented as follows: A:B → A:– + B+

(10.1)

The products are two ions, one with a positive charge (due to the absence of an electron) and one with a negative charge (due to the extra electron). This is the normal situation, for example, in pure water: H:O:H → OH:– + H+

(10.2)

381

The hydrogen and hydroxide ions are not very reactive. However, if the covalent bond is broken such that each atom receives one electron: A:B → A• + B•

(10.3)

then two free radicals have formed. If this process took place in water: H:O:H → •OH + H•

(10.4)

the resulting hydrogen and hydroxyl radicals would be quite reactive and dangerous. Free-radical reactions have three biologically important stages: initiation, propagation, and termination. The details of the initiation phase vary depending on the particular atom or molecule involved. For example, the mitochondria of the cell usually reduce oxygen (O2) in a series of one-electron steps. But oxygen has two reactive electrons. The result is the frequent formation of univalently reduced oxygen (O2•), or superoxide radical. This superoxide radical is capable of damaging biological structures, but usually it reacts with hydrogen peroxide (also formed normally in the cell) to yield a hydroxyl radical, a hydroxyl ion, and oxygen: O2• + H2O2 → •OH + OH– + O2 (10.5) In this particular situation, the hydroxyl radical is what enters the next or propagation phase and causes the cellular damage. No matter how it is initiated, however, once it is formed, the free radical (represented here as R·) can propagate itself indefinitely: R• + O2 → ROO•

(10.6)

ROO• + RH → ROOH + R• (which can recycle)

(10.7)

or it can undergo any of several other possible chemical rearrangements of existing molecules. The rearrangement depicted in reaction 10.6 above is similar to a lipid peroxidation, a process that causes great damage to cell membranes. Oxidation of proteins to yield a carbonyl protein takes place as follows:

382 Chapter 10 Stochastic Theories of Aging • OH + RCH2NH2 → H2O + R=• CHNH2

(10.8a)

O2 → O2• → H2O2 + •OH → H2O

R=• CHNH2 + O2 → R-CH(OO•)NH2

(10.8b)

R-CH(OO•)NH2 → R-CH=NH + H+ + O2–

(10.8c)

Cytochrome oxidase efficiently adds four electrons during energy generation in the mitochondria, but some of these toxic intermediates leak out of the inner mitochondrial membrane and can initiate their damaging propagation reactions in the mitochondrion (such as causing damage to mitochondrial DNA; see chapter 11) or perhaps in the membrane lipid components of the cytoplasm (causing lipid peroxidation). How are reactive oxygen species (ROS) produced as a consequence of the normal operation of the cell, and what kinds of defenses does the cell use against the damage they cause? It is important to realize that oxidants are not all bad; after all, our cells depend on the ability to oxidize and reduce various compounds in a controlled manner to extract chemical energy and use it to power our existence. The various reactive oxidizing and reducing substances in our bodies are normally kept in separate metabolic pools within the cell, separated by the necessity for precisely controlled enzymatic reactions to bring them together. But when cells are sick or injured, then these several barriers may be breached, the redox balance of the cell upset, and harmful amounts of ROS produced. Figure 10.6 shows a general scheme whereby an endogenous insult (high blood glucose levels) can initiate ROS production, which sets off a damage cascade using the chemical reactions described in equations 10.1– 10.9b above. The figure also indicates some of the levels at which various antioxidant enzymes or molecules can exert their effects, as outlined in equations 10.10–10.15 above. Insulin is a key metabolic hormone, and its normal operation is essential to the storage of triglycerides and glycogen, as well as the uptake of glucose by the skeletal muscle. If its production and/or uptake are not in balance, then either hyper- or hypoglycemia may result. It is known that even transient increases in blood sugar levels give rise to transient increases in ROS production in human cells, coupled with transient decreases in the cells’ normal levels of various antioxidant defense molecules. The healthy body can easily recover from transient increases in

It is now known that reactive nitrogen-based compounds can also form reactive radicals that can cause biological damage either directly or indirectly: RNO2 + 1 e– → RNO2 •

(10.9a)

RNO2 • + O2 → RNO2 + O2 • (10.9b) or 2 RNO2 • + 2H+ → R-N≡O + RNO + H2O

(10.9c)

This continuous cycle of cell damage and freeradical propagation can be terminated by various processes, which together constitute the termination phase: R• + R• → R:R

(10.10)

R• + ROO• → ROOR

(10.11)

2 ROO• → ROOR + O2

(10.12)

Antioxidant H + ROO• → Antioxidant• + ROOH

(10.13)

RNO2 • + GSH → R-N≡O + GS• + OH+

(10.14)

or

The last two categories of termination reaction are of great interest because they suggest that exogenous antioxidant molecules provide protection against the deleterious effects of free radicals. To sum up, the reaction of O2 with four electrons yields three high-energy free radicals and eventually a termination product such as water and can be visualized as follows:

(10.15)

10.2 Stochastically Based Theories

Glucose

383

NADPH oxidase

PKC O2

H2O2

CAT

SOD

GPX

Fe3+ OH

Fe2+

Protein Damage

Protein repair DNA Damage

DNA repair

Lipid Damage

Cell Injury & Death

GPX

Vit E

Vit C

Figure 10.6 A simple schematic representation of the interacting mechanisms of oxidative damage in a hypothetical cell. High glucose levels activate protein kinase C (PKC) and eventually induce reactive oxygen species production. The resulting superoxide or peroxide cause various types of cell damage if not terminated by various antioxidants (italics). (Redrawn from McCord 2003.)

blood glucose levels, but chronically high levels (as in diabetes) lead to extensive ROS production and cascading levels of damage to cell components. Recall from chapter 7 that the insulinlike signaling pathway is one of the body’s major longevityregulating mechanisms. The classic description of diabetes as leading to an accelerated aging is not that wrong. Multiple cellular defenses protect the cell against the continual oxidative and free-radical stress to which it is exposed. In general, these can be broadly categorized as compartmentalization, protective enzymes, and antioxidant molecules. Compartmentalization is the structural segrega-

tion of free-radical production sites and substances from other parts of the organism. Most oxidative metabolism, and thus most free-radical production, occurs in the mitochondria, which are structurally isolated from the rest of the cell. Furthermore, many of the naturally occurring defense mechanisms are concentrated about or in the mitochondria, although the cytoplasm and the extracellular fluids also contain antioxidant defenses (table 10.6). The net effect of this distribution is to create a layered defense concentrated near the sites at greatest risk. In addition, the cells are normally exposed to oxygen concentrations (1–20 mm) far below those present in air

384 Chapter 10 Stochastic Theories of Aging Table 10.6 Nature and Distribution of Cellular Defenses against Free Radicals Type

Location

Nonradical decomposition Catalase Glutathione Glutathione peroxidase Glutathione-S-transferase Quenching of active oxygen MnSOD CuZnSOD Lipophilic agents Vitamin E, carotenoids, flavonoids, ubiquinol, etc. Hydrophilic agents Vitamin C, uric acid, bilirubin, albumin, etc.

Cytoplasm, mitochondrial matrix space Cytoplasm, mitochondrial matrix space Cytoplasm Cytoplasm Mitochondrial inner membrane and matrix space Extracellular, cytoplasm Cytoplasmic and mitochondrial membranes Plasma, sera

Source: from Lippman (1983) and Camougrand and Rigoulet (2001).

(158 mm), which certainly lowers the rate of oxygen radical formation. There are many protective enzymes. Two of the more common ones are superoxide dismutase (SOD) and catalase (CAT), which often work together: SOD

2H+ + O2• + O2• → O2 + H2O2 (10.16) CAT

H2O2 + H2O2 → 2 H2O + O2

(10.17)

Another important antioxidant defense enzyme is glutathione peroxidase, which can reduce many peroxides besides the simple H2O2 (hydrogen proxide) molecule. Catalase can protect the cell only from H2O2. In addition to these several enzymes, some nonenzymatic molecules are capable of termination reactions with free radicals. The resulting oxidized antioxidant molecules are then either regenerated or excreted and replaced in the diet. Vitamin E (alpha-tocopherol) is one of the most important biological free-radical quenchers because of its ability to insert itself into the cell membranes and thereby protect the unsaturated fatty acids (which compose much of the membrane) from oxidative damage. Vitamin E is regenerated, at least in vitro, by reaction with ascorbic acid. Unfortunately, we absorb only small amounts of the vitamin E in our diet. However, continued

high levels of vitamin E in the diet does result in a slow buildup of the compound in various parts of the cell, thus circumventing the absorption limit to some extent. Studies with human endothelial cells in culture show that pretreatment of the culture media with alpha-tocopherol resulted in an increased in a vitamin E content and a dosedependent reduction in H2O2-induced lipid peroxidation (van Dam et al. 2003). The net result is lower levels of lipid damage. The in vivo effects of vitamin E show similar effects. For example, mitochondria from the hearts of rats fed vitamin E-deficient diets had 40-fold less vitamin E and were more susceptible to lipid peroxidation relative to rats fed a vitamin Esufficient diet (Paraidathathu et al. 1994). Longterm studies with humans have shown only minor side effects associated with a high-dose regime, and both animal and human studies have shown that vitamin E supplementation results in a significant reduction in various types of free radicals and various aspects of morbidities (Azzi et al. 2003; Halliwell and Gutteridge 1999; Jiang et al. 2003; VERIS 1998a,b). It has been suggested that vitamin E and its derivatives probably have cell regulatory functions in addition to their antioxidant functions (Packer et al. 2001; Visarius and Azzi 2003). Vitamin C, or ascorbic acid, is an effective scavenger of the superoxide radical in cells and has been shown to reduce lipid peroxidation in

10.2 Stochastically Based Theories

experimental animals. It may have a particularly important role in the lungs, where it is found in high concentrations in the extracellular fluid of the respiratory epithelium. Ascorbic acid is an excellent aqueous antioxidant that terminates the ROS reaction as follows: AscH– + X • → Asc • + XH

(10.16)

Ascorbic acid can also act as a pro-oxidant, especially when allowed to reduce ferric iron, which can then react with organic peroxides to form hydroxyl radicals. Many plant compounds contain effective antioxidants such as carotenoids and a wide variety of different phenolics, which may have multiple specific effects (Halliwell and Gutteridge, 1999). Many of them act as chain-breaking peroxylradical scavengers and as terminators of various free-radical reactions. The data of table 10.7 show that your mother’s advice to eat your fruits and vegetables was well founded, particularly since it has been shown that the serum overall antioxidant capacity values of humans increase after the consumption of strawberries, spinach, red wine (!) or vitamin C (Cao et al. 1998). Much data supports the view that higher serum antioxidant values translate into lower levels of oxidative damage.

Table 10.7 Antioxidant Capacity of Some Fruits and Vegetables Fruits

ORAC value

Vegetables

ORAC value

Prunes Raisins Dates Blueberries Strawberries Raspberries Plums Oranges Red grapes Kiwi

5800 2800 2700 2400 1500 1200 900 800 700 600

Kale Spinach Brussel sprouts Alfalfa sprouts Broccoli florets Beets Red bell peppers Onion Corn Eggplant

1800 1300 1000 900 900 800 700 500 400 400

Source: from http://hnrcwww.hnrc.tufts.edu/researchprograms/ USDALabResProgDes/Oracchrt.html. Note: ORAC is an assay for measuring the overall antioxidant capacity of a sample.

385

Finally, there is a large number of synthetic antioxidants, most of which were developed for industrial use as stabilizers of rubber or petroleum products. The BHT that one often sees on the ingredient list of packaged foods stands for butylated hydroxytoluene, one of the synthetic antioxidants widely used to prevent oxygen-based food spoilage. Some other synthetic antioxidants are ethoxyquinone and mercaptoethylamine. These are relatively nontoxic chain-breaking terminators of free-radical reactions. A thorough description of these compounds is given in Halliwell and Gutteridge (1989). Antioxidants do not work separately from one another, but rather they form a kinetically linked antioxidant network in which redox reactions link so that they eventually feed the electrons captured from ROS down the energetic path and out to water. A simplified version of such a network is pictured in figure 10.7. Aviram (2000) reviewed the field and concluded that oxidative damage to biomolecules have not yet been adequately validated as markers or indicators of the progression of any human disease. That statement appears to contradict much of the research data summarized here. Some of the discrepancy probably resides in the higher safety standards inherent in any recommendation for dietary supplementation of humans relative to the level of evidence required to draw a basic science conclusion. Some of it may stem from the incomplete knowledge of antioxidant usage or effects in many human studies. Some of it may well stem from the difficulty of obtaining accurate measures of a person’s actual antioxidant status; people do not accurately recount their diet, nor are the available laboratory tests capable of measuring simultaneously all aqueous and lipid-based antioxidant activities. These sorts of concerns are valid technical matters that need to be settled. But their resolution will not automatically lead to a change in the panel’s recommendations. I do not think that conclusion reflects an opinion that oxidative stress has little or nothing to do with aging and senescence. The reluctance of medical organizations to recommend antioxidant doses higher than the minimum needed to prevent overt illness represents a conceptual shift from disease management

386 Chapter 10 Stochastic Theories of Aging

Glucose (source) aqueous

lipid

G-6-P

NAD(P)*

GSSG

Asc+

-TocH

pentose shunt

NAD cycle

thiol cycle

vit. C cycle

vit. E cycle

6-P-G

NAD(P)H

GSH

AscH

-Toc+

H2O2

ROOH

H2 O

GPx

GR

ROO+

CAT

sinks

H2O + O2

SOD

GSSG

O2. Figure 10.7 A network view of antioxidant action. Reactive oxygen species generated oxidative damage (top right or bottom center) is repaired by vitamin E in a lipid compartment or by superoxide dismutase (SOD) in an aqueous compartment. These electrons are then transferred to water in an orderly fashion by connected interactions of various antioxidant cycles. Once oxidized, any antioxidant will not be capable of accepting other electrons until it is reduced. Thus, a continuous source of reducing molecules must be provided, usually from glucose via the pentose shunt. Note that electrons can flow from sources at either left or right to glutathione (GSH) and eventually to water which serves as the sink. Other and more complex network arrangements are likely in vivo.

to wellness management. Both tradition and law make this shift difficult for the medical profession and for medically based regulatory bodies such as the Food and Drug Administration. The reluctance to recommend high antioxidant doses likely foreshadows what the official response to some future pharmecutical anti-aging pharmecutical will be. On most days of our lives, we make important decisions on less than perfect data. Your decision of whether to take antioxidant supplements may be one of those occasions. Interested readers may wish to consult Jacob et al. (2003) for further details. All major components of the cell are damaged by oxygen-derived free radicals (table 10.8). The damage done is the net result of several complex variables, such as the types of free radicals present, their production rate, the structural integrity of the cells, and the activity of the several different antioxidant defense systems present in the organism. An age-related increase in cellular damage due to free radicals might be caused by agerelated alterations in any of these variables. Oxidant damge is a complex concept and cannot be

cleanly proven or disproven on the basis of a single experiment. Accordingly, several different types of evidence have been gathered in testing this theory. The interspecific, or evolutionary, approach is to examine the mean levels of various antioxidant defenses in different species and the mean

Table 10.8 Cell Components Damaged by Reactive O2 Species Component

Damage

Lipids

Peroxidation of unsaturated fatty acids in various membranes

Proteins

Oxidation of -SH containing groups, cross-linking, enzyme inactivation

Carbohydrates

Polysaccharide depolymerization, glycooxidation, glycoaldehyde production.

Nucleic acids

DNA damage include strand breaks, cross-linking, base hydroxylations, base excision

10.2 Stochastically Based Theories

levels of species-specific life spans to determine whether there is any statistical correlation between the two. There is a statistical correlation between species-specific lifetime energy expenditure and species-specific content of molecules such as SOD or other antioxidant defense system molecules (see Cutler 1984), although these correlations are much more robust when measuring mitochondrial MnSOD than when measuring cytoplasmic levels (see figure 11.05). This analysis is open to the same criticisms regarding the use of life span as a measure of senescence that apply to the interspecific DNA repair studies discussed earlier (Promislow 1993). An evolutionary comparison more interesting than simply comparing different mammals is to compare mammals and birds. As Holmes and Austad (1995a) have pointed out, birds live significantly longer than mammals of comparable size, despite having a much higher metabolic rate and blood glucose levels—two traits that are usually associated with accelerated senescence in mammals. This paradox suggested that birds must have specialized protective mechanisms against oxidative damage, and this suggestion turns out to be true. A biochemical investigation of heart mitochondria in rats (life span of ~ 4 years) and pigeons (~35 years) showed that the pigeon mitochondria leak only 30% of the level of free radicals that rat mitochondria leak (Herrero and Barja 1997a,b). This low level of free radicals probably translates into a lower level of oxidative damage. The decreased leakage is apparently due to a number of different changes in the pigeon relative to the rat, suggesting the incremental evolution of a more efficient energy production system. Another interspecific study that might escape Promislow’s (1993) criticism is that of Agarwal and Sohal (1996). Their comparison of five species showed an inverse relationship between the brain’s susceptibility to X-ray–induced oxidative damage and the maximum life span potential. In other words, the tissues of the longer lived species appear to be less susceptible to oxidative damage. Agarwal and Sohal also measured the same damage parameter at two different ages within one species and found an age-related increase in oxidative damage to proteins, suggest-

387

ing that the parameter was measuring a senescent change and not simply a change due to a variable other than aging. It would have been better to take these measurements in all five species tested, but the point is proven in principle. There are several different types of intraspecific approaches. One strategy consists of administering antioxidants to experimental animals and then determining whether the treatment has any effect on their life span. The addition and withdrawal of vitamin E to and from the diet of a nematode yielded results in excellent agreement with the predictions of the theory, as both the mean and the maximum life span shifted in accordance with the exogenous vitamin E supplied (Balin 1983). Vitamin E supplementation in the diet of paramecium resulted in a significantly increased clonal life span, regardless of whether the life span was measured chronologically (in days) or functionally (in number of fissions; Thomas and Nyberg 1988). Some interesting data were obtained by vitamin E supplementation of the diet of the banana fruit fly, Zapronius paravittiger (Kakkar et al. 1996). Every dose tested increased longevity above the normal control value. The optimum dose for extension of longevity in either sex was 5 mg/ml of food. Higher doses yielded intermediate median and maximum life spans; doses greater than 25 mg/ml yielded longevities that were significantly decreased relative to the longevities of the controls. Furthermore, at the optimum dose, the animals exhibited a significantly decreased level of malondialdehyde (an end product of lipid peroxidation) and significantly increased levels of antioxidant enzymes (catalase and peroxidase) throughout their life span. The mechanism by which vitamin E increases enzyme activities is not clear. It is clear, though, that the beneficial effects of vitamin E are mediated through its effect on the antioxidant defense system. The banana fruit fly responds in a similar manner when its diet is supplemented with 100-mM methionine, an essential amino acid that can inhibit production of the superoxide radical (Sharma et al. 1995a). These two experiments suggest that higher levels of antioxidants are sufficient to increase longevity in the banana

388 Chapter 10 Stochastic Theories of Aging fruit fly. There is some debatable data suggesting that the optimal human dose may be about 400 units/day (E. R. Miller et al. 2004). Another approach to examining the effects of oxidation is the development of synthetic mimetics of antioxidant enzymes such as CuZnSOD or catalase (Giblin et al. 2003). Some of these manganese-based compounds, particularly EUK-134, are very effective in teminating free-radical reactions. When fed to normal-lived C. elegans, EUK-134 increased its mean and maximum life span by about 44%, such that the treated animals lived almost as long as did the long-lived age-1 mutants. The compound also rescued a short-lived mev-1 mutant and increased its life span by about 60% so that its survival curve resembled that of a normal animal. Since the mev-1 mutant is known to have elevated ROS production, the rescue shows that the mimetic drug probably exerts its effects by enhancing the animals’ antioxidant defense and bringing down the ROS to more normal levels. These experiments show that decreasing the animal’s antioxidant defenses leads to a short life, while increasing the animal’s antioxidant defenses leads to a longer life. It is difficult to measure free radicals directly, especially in vivo. It is much more feasible to measure the end products of free-radical formation. One easily measured end product is the group of intracellular pigments collectively known as lipofuscin, which I discuss in more detail in chapter 11. Lipofuscin granules accumulate slowly within the cell. They are believed to be the end product of a variety of processes, of which freeradical–induced lipid peroxidation is believed to be the major process. Many investigators measure malonaldehyde or other distinctive yet stable breakdown products indicative of ROS degradation of biomolecules and use these as a measure of the oxidative stress the organism is undergoing at that time. Working on the assumption that the formation of free radicals in poikilothermic houseflies is proportional to the amount of their physical activity, Sohal and Donato (1978) raised male houseflies under spatial conditions in which they could fly about freely (the high activity group) or could only walk but not fly (the low activity

group). The low-activity animals showed not only a significant increase in their mean and maximum life spans relative to the high-activity controls, but also a much lower rate of lipofuscin formation, suggesting that the rate of oxidative damage is modulated by the rate of physical activity and has something to do with the rate of aging. This suggestion was supported by Sohal and Donato’s observation of an inverse relationship between catalase activity and inorganic peroxide concentrations over the life span of the housefly. The catalase activity was high at young ages and then steadily decreased; the peroxide concentration was initially low and then increased (Sohal et al. 1983). In a related experiment, Sohal et al. (1986) found that they could identify at midlife, by means of their activity, certain potentially long-lived and short-lived individuals of the same housefly population. Biochemical assays on these individuals showed that the short-lived animals had lower levels of antioxidant defense activities (such as SOD and catalase activities or glutathione levels) and higher levels of inorganic peroxides and other biochemical indicators of free-radical activity. These data again lend themselves to the same interpretation and are consistent with the theoretical predictions of the oxidative-damage theory. Methionine is an essential amino acid and can inhibit production of the superoxide radical. Feeding this compound to a certain species of fruit fly led to a significant increase in mean and maximum life span, an increased level of catalase and peroxidase activity, and a decreased level of lipid or inorganic peroxides (Sharma et al. 1995a). The induction of antioxidant defenses by feeding will be of some interest if it turns out to be a general finding. These conclusions are not restricted to flies. A cross-sectional biochemical study done on gerbils showed that the level of molecular oxidative damage to proteins and to DNA increases with age and that the increased oxidative damage is due both to an elevation in the rates of oxidant generation and an increase in the susceptibility of the tissues to oxidative damage (Sohal et al. 1995). Thus the age-related increase in oxidative damage might be attributed to higher levels of

10.2 Stochastically Based Theories

free-radical generation, presumably by the mitochondria in aerobic muscles such as the heart, as well as to a decreased antioxidant protective ability on the part of the animals’ defense mechanisms. Similar data have been compiled for many different organisms (see, e.g., Halliwell and Gutteridge 1999). For example, the short-lived senescence-accelerated mice (SAM-P) have a higher rate of lipid peroxidation than their SAMresistant (non–short-lived) controls (Park et al. 1996). When the SAM-P mice were treated with an injectable free radical-trapping compound for 1 month, they showed a significant decrease in levels of oxidative damage (Butterfield et al. 1997). In addition to demonstrating the existence of intraspecific baseline differences in antioxidant defense activity in mice, these studies demonstrate that exogenous synthetic antioxidants can alter these baseline levels. When combined with the data regarding dietary guidelines and vitamin supplementation (Ames et al. 1993), this observation suggests that the presumably genetically set baseline levels of antioxidants can be altered by environmental modulation. This possibility offers hope of eventual practical interventions (see chapter 7). Macaque monkeys are being used to determine biomarkers of aging (see chapter 3). The factors being used to construct the biomarkers and to estimate the rate of biological aging derived from them include no measurements of antioxidants. Thus the observation that the levels of seven circulating antioxidants in the blood of the monkeys showed an inverse relationship with the calculated rate of biological aging is particularly persuasive, especially because it was also shown that antioxidant level influenced the animal’s susceptibility to disease (Short et al. 1997). Finally, note that high blood glutathione levels in humans are correlated with high self-rated health scores, while low blood glutathione levels are statistically associated with low self-rated health scores (Julius et al. 1994). Furthermore, a prospective longitudinal study involving 2900 men over an 18-year period revealed that significantly elevated blood levels of vitamins C and A are associated with significantly lower risks of death due to ischemic heart disease, whereas low

389

levels of vitamin C and beta-carotene are associated with a significantly higher risk of death due to cardiovascular disease (Stahelin et al. 1989). It is well known that humans suffer from an increased loss of physiological function as they age. It is likely more than just a mere coincidence that the ability of human plasma to protect against radioactive damage decreases linearly between the ages of 30 and 80 years (Lenton and Greenstock 1999). These epidemiological findings have been amplified in recent years by several other studies, as discussed in chapter 6. The data of figure 3.21 show that longer-lived human males have signficantly lower blood insulin levels. A similar finding was observed in monkeys on a caloric restriction regime. Long life, whether spontaneous or induced, is accompanied by low insulin and thus low blood glucose levels. The concepts presented in table 7.12 and figures 9.1 and 10.6 provide a conceptual mechanism connecting low insulin/glucose levels to high stress resistance levels to low ROS levels to extended longevity. The laboratory and clinical findings to date—namely, that intraspecific variation in antioxidant levels is strongly correlated with various indices of morbidity and/or mortality—are applicable to humans. These findings are consistent with the observations that high levels of lipofuscin are characteristically found in animals with vitamin E deficiencies and support the conclusion that lipid peroxidation is positively correlated with the polyunsaturated fatty acid (PUFA) concentrations in the membrane and negatively correlated with the concentrations of antioxidants such as vitamin E (Lippman 1983). Cells depend on high PUFA levels for normal membrane functioning and elasticity, yet the same PUFAs subject cells to peroxidative damage arising from oxygen metabolism. In the absence of sufficient antioxidants, the PUFAs of the cell membrane, which are part of the compartmentalization antioxidant defense system, are susceptible to oxidative damage. Different regions of the brain undergo complex age-related declines in various different antioxidant enzymes (Benzi et al. 1989). This fact may take on particular importance in view of the fact that the brain uses about one-fifth of the total

390 Chapter 10 Stochastic Theories of Aging oxygen demand of the body. The spatially distinct antioxidant enzyme patterns could provide the structural basis for the spatial and temporal pattern of cellular damage and/or decrement observed during the aging process. The best approach to testing the oxidativedamage theory is the genetic strategy, in which one can either (1) construct long-lived and shortlived animals and then assay them to determine if their antioxidant defense systems have been altered in the expected manner or (2) construct animals that have genetically altered levels of antioxidant defense systems and then test them for their effect on life span and rate of aging. Both types of these experiments have been done and were discussed in chapter 7. In general, the results of both types of experiments are consistent with the predictions of the oxidative-damage theory. The available data from each of the model systems indicate that life span is positively correlated with the levels of antioxidant defenses and that manipulating the levels of antioxidant defenses brings about a corresponding alteration in the life span.

10.2.8 Molecular Chaperones as Modulators of Stochastic Stresses Affecting Longevity There are other forms of stress besides oxidative stress, and the cell possesses other types of stressresistance mechanisms. One of the best studied of these other protective mechanisms is the heat shock response, which is found in all organisms from bacteria to humans. The term “heat shock” is a residue from their initial discovery in Drosophila as a protein family induced in response to hyperthermia (Ashburner and Bonner 1978). Increased awareness of their ubiquitous induction by varied stressors led to these proteins being more generally referred to as cellular stress proteins. The heat shock response is the most important of these other stress response systems, as noted in figure 9.5. When an organism is exposed to a nonlethal increase in temperature for a limited period of time, the cells vigorously up-regulate the expression of a family of conserved heat shock

proteins (HSPs) that help alleviate the stress by (1), protecting cellular proteins from being damaged; (2) allowing newly synthesized proteins to fold into a biologically functional state despite the stress; and (3) targeting proteins that have been damaged beyond repair for degradation. Note that the latter function means the cell has a builtin disposal mechanism for altered proteins. HSPs are present in normal animals even in the absence of any overt stress, for any cell must be able to fold correctly its newly synthesized proteins and tag for recycling its older or abnormal proteins (figure 10.8). But stressing the animals increases the level of abnormal protein folding and hence increases the demand for HSPs. The stress inducibility of the HSPs (at both the transcriptional and post-transcriptional levels) allows the number of HSP proteins to be rapidly adjusted to cope with the problems facing the cell. The ability to induce these stress-response genes declines with age. Young animals vigorously respond to a heat shock by rapidly mobilizing HSPs to such an extent that the animals’ survival is actually improved by the heat shock (Khazaeli et al. 1997). This surprising result comes about because induced HSPs not only protect the existing proteins from thermal damage but they also tag for disposal all proteins altered by preexisting unrepaired insults (see figure 10.9). Young animals exposed to heat shock exhibit a decreased age-specific mortality rate (qx) and/or an extended longevity. This widely observed phenomenon is called hormesis and occurs whenever animals are exposed to low-level environmental insults for which they possess an inducible repair system. The hormesis concept had its origins in homeopathic medicine but evolved over the years to refer to situations where an organism exhibits a maximum fitness as a consequence of exposure to some low level of chemical or environmental stress that would be harmful or even lethal at high doses (Arking and Giroux 2001; Parsons 2000; Salen 2000). I return to this topic in the last section of this chapter. Old organisms have a markedly diminished ability to respond effectively to heat shock, probably because their ability to up-regulate HSPs is significantly weakened (Keuther and Arking 1999;

10.2 Stochastically Based Theories

391

HSP70

mitochondria newly synthesised protein

other mitochondrial HSPs

other HSPs

correctly folded protein

correctly folded protein

Figure 10.8 A schematic representation of the role of heat shock proteins (HSPs) in facilitating the transport and folding of proteins in the unstressed cell. (After Broome et al. 2003.)

UNSTRESSED correctly folded protein moderate correctly folded and functioning proteins

stress

mild, nondamaging stress

protein aggregation and cell death

UNPROTECTED CELL

damaging stress correctly folded and functioning proteins increased synthesis of HSPs

PROTECTED CELL

Figure 10.9 The cytoprotective function of heat shock proteins (HSPs). Mild stresses induce synthesis of sufficient HSPs to enable the cell to survive subsequent damaging stresses that would normally prove lethal for the cell. (After Broome et al. 2003.)

392 Chapter 10 Stochastic Theories of Aging Niedzwiecki and Fleming 1990). It is now generally accepted that aging is associated with a reduced ability to express the HSPs in several tissues following stress. Heydari et al. (2000) have shown that the inability to up-regulate the synthesis of HSP70 in rat hepatocytes is traceable to conformational changes affecting the ability of the heat shock transcription factor (HSF) to bind to the HSP70 promoter. Perhaps this might serve as a general model for the indirect manner in which HSP expression is lost with age. I postulated in chapter 9 that it was the gradual overwhelming of the cell’s antioxidant defense system (ADS) enzymes and HSPs by abnormally aggregated cellular proteins that underlies the transition from the health span to the senescent span portions of the life span (figure 9.5). Such a situation might arise as a result of the accumulation of unrepaired proteome damage leading to the aggregation of abnormal proteins (Morley and Morimoto 2001) as well as the inability of HSF to induce HSP synthesis. Conversely, over-expression of certain HSPs in the mitochondria can significantly extend the longevity of normal-lived animals (Aikagi et al. 2003; Kurapati et al. 2000; Morrow et al. 2004a). There is indirect evidence that mitochondrial HSPs may increase that organelle’s efficiency, decrease its production of ROS, and thus play a role in bringing about the expression of the extended longevity phenotype (Arking et al. 2002b; Curtsinger and Khazaeli 2002; Ross 2000). Heat shock interests us because it can increase the survival and longevity of young animals, because overexpression of HSPs may be part of the pathways leading to extended longevity, because it underlies the stress response of specific tissues (e.g., Broome et al. 2003), and because the agerelated inability to fully activate HSPs (Heydari et al. 2000) can serve as a model of what happens to the gene expression patterns in all aging cells (see table 7.3). It is implicit in all studies of aging and the genetics of longevity that changes in the proteome bring about the observed loss of function. For these reasons, studies on protein alteration with age is now an active area of proteomics research. A genome-wide analysis was used to

identify the direct transcriptional targets of the yeast HSF (Hahn et al. 2004). Almost 3% of the entire genome (~180 genes) was found to constitute such targets. These genes encode proteins that have a broad range of biological functions that extend beyond the known HSP tasks of protein folding, degradation, and trafficking to include processes involving energy generation, cell signaling, small molecule transport, transcription, and so forth. Deciphering how the up-regulation of this set of interacting HSF target genes is brought about by the gene interaction network and how that expression maintains the gene interaction network in its optimal (i.e., healthy) state promises to offer insights into the more general problem of how to prevent stochastic insults from shifting the gene expression network from a healthy to a senescent state. Parsons (1995, 2000) suggested that resistance to stress is the fundamental trait through which long life is expressed. The data and discussion in chapter 7 identified the genetic mechanisms responsible for longevity determination. The ability of the insulinlike signaling pathway to switch gene expression patterns from one enhancing stress resistance to one enhancing reproduction indicates that Parson’s view is generally correct. The observation that the appearance of many signs and symptoms of aging are coordinated is not new; the idea that they may be linked at a deep biological level through the intervention of a few common mechanisms such as the insulinlike signaling pathway or the heat shock response is an interesting and challenging concept which has changed the way we view the life span (see figure 9.6). Taken as a whole, the data demonstrate the existence within organisms of a a few highly conserved genetically programmed responses to limit stochastic damage arising from oxidative and other forms of stress. Oxidative damage is certainly a major component underlying the age-related loss of function in many species. Equally important, however, is the organism’s response and its genetic modulation of the stresses it faces. The variable response in these several stress-resistance mechanisms may account for much of the individual heterogeneity in life histories.

10.2 Stochastically Based Theories

10.2.9 Epigenetic Aspects of Stress Responses and Individual Variability in Longevity Jazwinski and colleagues (1998) have postulated that the subdivision of a population into a normally aging component and a slowly aging component may be mathematically modeled by assuming only one essential property of this process: increasing change in the sense of deviation from the initial state of the organism. The change envisioned is not the temporary physiological responses associated with homeostatic maintenance of some baseline value, but rather a modification that leads to significant alterations in the baseline value itself. The changes that constitute aging are not programmed (see chapter 4) and thus are in principle reversible. Thus, a system may either show increased change (i.e., normal aging) or decreased change (i.e., reversed or slow aging). There should be some obvious interest in understanding just what sort of process is capable of reversing aging. As described by Jazwinski et al. (1998) and illustrated in figure 10.10, the probability of change in either direction can be estimated by the equation: P n+1 = Ae-Pn where P is a dimensionless variable that describes the state of the system and is related to the probability of change, A is a factor that accounts for the effect of aging on this probability, and n = 0, 1, 2, 3 . . . (consecutive states of P, A>0). A large value of P indicates rapid change or aging. The probability of aging is related to the previous state of the system and is not dependent on a particular value of P. The iterative solution to this equation results in changing values of P that converge on a stable fixed point before bifurcating into two different states: one characterized by a high P value (i.e., P converging on A) and the other characterized by a low P value (i.e., P converging on 0). Thus, the individual aging state is indeterminate and can be represented by one or the other of the two fixed points. Given a large number of individuals, there is a certain probability

393

of each of them converging on the high P value (normal aging) state, complemented by a certain probability of each of them converging on the low P value (reversed or delayed aging) state. If the probability of the former state is high and that of the latter state is low, this equation adequately describes, at least to a first-order approximation, the survival outcome and demographic longevity observed in the large populations studied by Carey et al. (1992; see also Vaupel et al. 1998). This epigenetic stratification equation indicates that the division of a population into normal-aging and reversed or delayed-aging subsets occurs theoretically in all populations. It also indicates that the main characteristic of the slowaging group is the ability to resist change. In the context of the experimental data reviewed above, we can interpret this property as indicating that mutational and oxidative damage to DNA and proteins probably constitutes the normal aging change processes taking place during adult life, and the extraordinary ability to resist this damage process characterizes the long-lived subset of organisms. Given these findings, it is a testable hypothesis that the hormetic induction of the ADS and/ or hsp genes and the subsequent prevention of oxidative damage and/or the accumulation of abnormally folded proteins provides a mechanism through which deleterious changes can be halted and epigenetic stratification can proceed. One might ask how this hormetic induction process differs from the more ordinary interindividual variation in enzyme levels often seen in a population. The answer probably lies in the fact that interindividual variation in enzyme level most often is the result of allelic differences in genes governing the expression of the ADS. Such allelic (or SNP) differences give rise to hereditary differences in the constitutive expression level of the ADS genes, such as take place in normal-lived Drosophila Ra and Rb strains (Arking et al. 2000). But hormetic induction does not yield an increased constitutive level of gene expression. It yields instead a spiked series of temporary gene up- and down-regulation such that each period of exposure to a low-level stressor yields sufficient expression of ADS and/or hsp genes and other

394 Chapter 10 Stochastic Theories of Aging

5 high Loss of homeostasis

3 P

Propensity to Age

4

2

Early stages

1

low

Maintenance of homeostasis

0 0

1

2

3

4

5

A Effect of Change on P Figure 10.10 Epigenetic stratification, in which a population of aging beings will move rather slowly from the initial state (A = 0, P = 0) of the system to some unstable point (3>A>2, P = 1). The population now bifurcates, most of them aging (i.e., losing function) rather rapidly and moving to (A = 5, P = 5), while some small subset do not age (i.e., gain function) and move to (A = 5, P = 0). One implication of this mathematical model is that not all changes in gene connectivity must necessarily lead to loss of function. (After Jazwinski et al. 1998.)

beneficial enzymes/proteins so as to counteract any and all of the cellular damage brought about by the stressor. This removal of cumulative damage in the cell amounts to a reversal of aging changes and thus to epigenetic stratification. This may be viewed as a possible explanation of the slowed and even decreased aging observed in older animals, including humans (see figure 2.24). We should not

forget that only one-third of human centernarians have a familial (i.e., hereditary) component to their extraordinary longevity (Perls and Terry 2003). The other two-thirds must have some nonhereditable factor underlying their long life, and this epigenetic induction of chaperones and other protective proteins may well play such a role in these people.

11

Systemic Theories of Senescence

11.1 If Aging Is Systemic, Why Is There So Much Variability? Systemic theories of biological function often invoke thoughts of an organized program designed to bring about the function in question. The word “program” means different things in different contexts, but it often is interpreted as meaning a set of coded instructions that a computer must follow to complete the task assigned to it. One implication of this concept is the idea that there is a very tight and precise linkage between the coded instructions and the operation of the computer; the computer can do nothing that is not accurately written down in the instructional sequence beforehand. Biologists use terms such as “genetic program” or “developmental program” as a convenient shorthand to describe the coordinated and sequential events that constitute organismic development. This adoption of a metaphor originally developed in cybernetics makes it easy to assume that the rigor and precision of the computer term is equally valid of the biological term. This is not correct. Many of the processes upon which normal plant and animal development depend are not directly specified by the nucleotide sequence in the genome (Wilkins 1986). These essential processes are emergent properties, implicit perhaps in our definition of what we expect normal ectodermal or mesodermal cells, for example, to do, but not explicitly set down (Edelman 1988). During embryonic development, different cells predictably migrate to certain regions of the body

and meet with another population of distinct cells originating elsewhere that have predictably migrated to the same area, and the two populations interact to form a new tissue type. To the best of our knowledge, no gene directs an ectodermal cell to migrate to a specific position and there interact in a certain way with a mesodermal cell type that bears complementary instructions. Because these are emergent properties that are not precisely spelled out in any single DNA strand, we should not be surprised if any two organisms show significant environmentally modulated variation (i.e., chance) in the timing and manner with which these higher-order interactions are expressed (see Kirkwood and Finch 2002). As a result of “noise in the system,” the developmental program of humans, for example, gives rise after about 266 days to a baby. But as any obstetrician and most mothers know, very few infants are born after exactly 266 days of development. Most (about 75%) are born sometime within 252–278 days, and a substantial minority are born at even more extreme times. These differences amount to more than a 10% variation in normal developmental rate. Our own developmental program yields not a precise result, but a normal probability distribution of results. Emergent properties are more prone than determinant ones to vary at the macroscopic level as a result of random microscopic variations. Now, if aging itself is an emergent property, why should we be surprised when human adults— subjected to considerably more environmental modulation than any fetus—show comparable variation in the times at which individuals display

395

396 Chapter 11 Systemic Theories of Senescence various biomarkers of aging? The presence of variation in aging is not an argument against systemic theories, nor is it an argument for stochastic theories. It simply exists. For the purposes of this discussion, it will be helpful to reiterate what we pointed out in chapter 9. The term “systemic” is being used here to denote an explanation based on the occurrence of a hierarchical cascade of interconnected events. I most emphatically do not mean an explanation based on the idea of determinative and sequential gene action designed to produce aging. That would constitute an adaptive theory of aging (that we age because there is a particular reason for us to age), and, as we saw in chapter 4, aging has evolved not because it is adaptive but because the force of natural selection declines with age (we age because there is no reason not to age). By “systemic,” I mean simply interlocking or cascade mechanisms that can give rise to a sequential series of common aging phenotypes. The two major systemic processes that might modulate aging and senescence are genetics and metabolism. I discussed the genetic pathways responsible for longevity determination and resistance to senescence in Chapter 7. In this chapter, I focus on several different aspects of metabolism that have significant effects on aging.

11.2 Metabolism Metabolism is a complex process. Its intricacies have convinced many that it is capable of explaining the complexities of aging. I focus attention here on different facets of metabolism, alterations in which have been considered to give rise to the processes of senescence. Because all enzymes are coded for by genes, metabolism is ultimately a genetically controlled process. But metabolism has its own levels of enzymatic and feedback controls built into its networks of interrelated processes, and so it is logical to examine the ordinary workings of metabolic control of aging somewhat apart from the ultimate gene-level controls. When we talk about metabolism, we are using one word to describe an incredibly com-

plex interplay of thousands of individual genecontrolled and environmentally modulated reactions with multiple feedback loops that impart robust homeostatic properties to the network. Evidence of strong genetic control is difficult to discern at this level. It has long been known that bigger animals have a longer life span than do their smaller relatives (see figure 4.11). It is also known that an organism’s basal metabolic rate (O2 consumption/unit of body weight) is inversely proportional to body weight. The early coupling of these facts led to the belief that longevity and metabolism are bound in a causal relationship: that high metabolic rates lead to, or are associated with, short life spans. However, I have pointed out that this relationship at the interspecific level is rather loose and may represent a correlative rather than a causal relationship (see figure 6.5). In chapter 6, I briefly reviewed some of the information suggesting that experimental alterations of metabolic rate are capable of producing, in some organisms, corresponding alterations in life span. In addition, I presented data showing that the metabolic rate appears to decline with advancing age (see figure 5.23). These and similar observations led to the widespread concept that longevity can be best understood as a function of metabolic decline. The two metabolically based theories of aging and senescence that I examine are the rate of living theory and the mitochondrial-oxygen radical theory. The two theories propose rather different mechanisms and so give rise to rather different predictions. In addition, I also discuss the metabolic changes that the organism must undergo if it is to support extended longevity.

11.2.1 The Rate-of-Living Theory The rate-of-living theory was first proposed by Buffon in 1749 and was popularized by Pearl in 1928. It states that longevity is inversely proportional to metabolic rate. The formal concept was based in part on the observations of Rubner (1908), who noted (erroneously, as it later turned out) that mammalian species with very different species-specific life spans nonetheless expend a

11.2 Metabolism

similar amount of metabolic energy per gram body weight per lifetime. This was equivalent to saying that all animal cells had the same fixed amount of calories available to them. Loeb and Northrop (1917) reported that poikilothermic animals, such as Drosophila, show an inverse relationship between adult life span and ambient temperature. They inferred (but did not prove) that this situation arises as a result of the temperature-dependent breakdown and/or accumulation of various metabolites critical to the organism’s health. Pearl (1928) then wove these observations, among others, into the rate-of-living theory. The original theory makes two predictions: (1) There is a predetermined amount of metabolic energy (a metabolic potential) available to the organism that can be expressed equally well in terms of oxygen consumption per life span or kilocalories expended per life span, and when this amount of energy is gone, the organism dies. (2) There is an inverse relationship between metabolic rate and aging. The theory sounded plausible and was widely accepted, despite the confused and circular logic that Pearl used in writing about it (Lints 1989). However, recent findings indicate that the theory is incorrect as commonly stated. The first prediction has been disproved by the comparative analyses of metabolic rates in different species of animals. For example, the data obtained from the analysis of 77 species of mammals (see figure 6.5) show that these different mammals have not just one predetermined metabolic potential but a spectrum of lifetime energy potentials with values ranging from 220 to 781 kilocalories per gram per lifetime. Experiments on invertebrates have shown that different species of insects also have different metabolic potentials. Thus, different species have different metabolic potentials, in contradiction to the theory’s predictions. Other studies suggested that the metabolic potential was uniform with each species. The demonstration by Sohal and Donato (1978) that low-activity flies live longer than high-activity flies (see chapter 10) was widely interpreted as showing that the total amount of calories a fly can expend (or of oxygen that it could consume, which is just another way of

397

measuring the same thing) is fixed and that the length of its life depends on whether it spends them quickly or slowly. This interpretation seemed to be substantiated by the many observations showing that poikilothermic animals, whose metabolic rate is proportional to the ambient temperature, live longer when kept at low temperatures than when kept at high temperatures. In fact, such results were thought to support the theory’s second prediction of an inverse relationship between metabolic rate and life span. But other data show that the metabolic potential not only differs among species, but it does not even stay at a constant value for different populations of the same species. In one study, individuals from a genetically selected long-lived (L) strain of Drosophila and individuals from their progenitor normal-lived (R) strain were raised at different temperatures, and their metabolic rates were measured (figure 11.1). The mean daily metabolic rates of the two strains were statistically equivalent to each other across the range of tested temperatures. The long-lived organisms spent about the same number of calories per day and in much the same manner as did the normal controls, yet they lived significantly longer. As a consequence, during their entire life span the long-lived strains expended about 40% more calories than did the normal-lived strains simply because they lived longer. The L strains have a significantly higher lifetime metabolic potential. The conclusion is inescapable that the long-lived Drosophila strains do not bring about their increased longevity simply by husbanding their calories (Arking et al. 1988). Another mechanism must be involved. Conclusive evidence that at least two different physiological functions are involved in the aging process of these Drosophila is shown in figure 6.4. Within each strain, temperature can alter life span. However, these environmental treatments are clearly separable from the genetically based interstrain differences, since the latter are not abolished by the former. Not only do the environmental and genetic treatments exert their particular effects on life span by affecting separate physiological compartments, but their effects are probably additive. Metabolic rate appears to

Metabolic rate (µI O2/mg/hour)

398 Chapter 11 Systemic Theories of Senescence

5 4 L

3 2

R

1 0

0

10

20

30

50 40 60 Adult age (days)

70

80

90

100

Figure 11.1 The measured metabolic rate at 22oC throughout the lifetimes of a normal-lived control strain (R) and a long-lived strain (L) of Drosophila. Strains with different longevities have similar metabolic rates. (After Arking et al. 1988.)

be only one of several factors that regulate the life span of an organism. The rate of living theory is plausible, but it oversimplifies aging by reducing it to the manipulations of a single dependent variable. The predictions made on the basis of this simplistic model have not been upheld by the data. A logical deduction from these data is that there may be a qualitative difference between normal and long-lived animals, and this deduction is upheld when examining the role of mitochondria in aging.

11.3 Mitochondrial Damage Theories Another approach that seeks to lay an updated metabolic framework under the aging process is exemplified by theories suggesting that the cause of the physiological senescence characteristic of aging cells resides in the cumulative effects of oxidative injury to the mitochondria. In this view, the sequence of events that lead to aging, as postulated by Miquel and Fleming (1984) and by Wallace and colleagues (1995), is as follows. First, as embryonic cells differentiate they usually lose the ability to divide. Low division rates are characteristic of many tissues in the aging adult (see chapter 12). Second, nondividing cells have a much lower requirement for mitochondrial replication. Even though the mitochondria in irreversibly differentiated, nondividing or very slowly

dividing cells can divide, they do so at a very reduced turnover time (about 1 month) compared to nondifferentiated and rapidly replicating cells. Thus the mitochondria of nondividing, differentiated cells must last much longer than the mitochondria of rapidly dividing cells. The third and key point is that almost all of the oxygen consumption by the cell takes place on the inner membrane of the mitochondria. This is the site where the oxygen-induced free radicals form as a side effect of mitochondrial respiration. This is also the site where they might be most likely to cause oxidative damage to the structural and functional components of the mitochondrial membrane. The molecules involved in the final steps of energy production are located within and are part of this mitochondrial membrane. Therefore, the accumulation of oxidative damage to the mitochondrial membrane would be expected to impair the mitochondrion’s ability to produce energy. This impairment alone might account for many of the metabolic and physiological declines, such as the decreased metabolic rates (see figure 11.1) that are almost always observed in aging cells and organisms. It is well known that damaged mitochondria are often (but not inevitably) found in aged cells (see figure 5.6). Finally, the free radicals may also cause irreparable damage to the DNA of the mitochondrial genome (which is separate from the nuclear genome)—damage that could well result in an accelerated loss of the organelle’s energy produc-

11.3 Mitochondrial Damage Theories

tion and that might lead, in turn, to further decreases in the cell’s replicative and repair abilities. It is speculative but plausible to assume that such decreases lead to the pleiotropic onset of at least some common age-related dysfunctions; hard evidence of this connection is still needed. This “oxygen radical–mitochondrial injury” theory has the advantage over the older rate-ofliving concept in that it postulates that aging is due to certain specific and measurable reactions, taking place in defined structures, that bring about particular types of molecular damage and physiological senescence. In this sense, the new theory is certainly easier to test precisely than were the older concepts. The new theory is also attractive because it ties all of the many observations on decreased metabolic functioning together with the high plausibility of the free radical–oxidative damage theory. But attractiveness is one thing, and proof is another. What does the evidence say?

11.3.1 Mitochondrial Structure and Function Mitochondria are self-replicating organelles found in the cytoplasm of eukaryotic cells. They are inherited only via the maternal line. They are likely derived from oxidative prokaryotic cells that entered into a symbiotic relationship with a glycolytic proto-eukaryotic cell some 1.5 billion years ago. Their genetic code is slightly different than that used by the nuclear genome. As a consequence, they retain their own DNA (mtDNA), which is more similar to prokaryotic DNA in many characteristics than it is to eukaryotic nuclear DNA. In humans, the mtDNA is a closed circle of 16,569 nucleotides that encode the genes for 13 proteins of the respiratory chain, as well as 2 ribosomal RNAs and 22 tRNAs. The mitochondrion contains about 1000 or so different types of protein molecules plus some as-yet undetermined number of different types of lipid molecules (Jensen et al. 2004). A finding of great interest is that the protein composition of mitochondria is a function of their cell type. It seems as if only half of all mitochondrial proteins are

399

found in every cell type, while the other half can vary substantially between liver, heart, kidney, and brain. This finding, when replicated, will give rise to two questions: Just how different are the mitochondria from different tissues, and is there any correlation between their tissue-specific differences and the tissue’s performance in the aging organism? Obviously the small mtDNA genome cannot encode all the molecules necessary to the organelle. The hundreds of other genes have been captured by the nucleus over the ages, as evidenced by the fact that such genes display a standard Mendelian inheritance. That type of inheritance is not possible for a maternally inherited genome. So most mitochondrial gene products are coded in the nucleus, translated in the cytosol, and imported by special mechanisms into the mitochondria. It turns out that the 13 proteins coded for by the mtDNA are essential to the operation of the organelle. One can view the relationship between the two genomes of our cell as being of a competitive nature in which the nucleus has captured almost all the organelle’s genes since the endosymbiotic event occured, but the organelle has retained enough so as to have a “veto power” and thus retain its own “independent” existence. This evolutionary history explains many of the aspects of mitochondrial aging. Each cell can contain hundreds or even thousands of mitochondria, depending on the cell type and size. Metabolically active liver cells might have as much as 20% of their cytosolic volume taken up by the mitochondria. The extremes are bounded by red blood cells, which have none, and by ova,which have about 100,000 mitochondria. New live-cell imaging techniques using fluorescent probes have revealed that mitochondria exist in the cell not only as individual and discrete organelles but also as a continuous, interlinked “mitochondrial reticulum” which weaves its way about the cell (Rutter and Rizzuto 2000). The transition from the reticular to the individual state (sometimes called the “thread–grain” transition) seems to be under the control of certain GTPases and must reflect some functional organization important to the cell. Skulachev (2001) suggests that the reticulum may represent an intracellular

400 Chapter 11 Systemic Theories of Senescence power-transmitting cable within the cell, allowing energy to be transported from the site where it is synthesized to the site where it is needed. The imaging techniques also reveal that the reticulum is not homogeneous but rather exhibits localized patches of activity or the transient pinching off of a single mitochondrion. This new view of mitochondrial anatomy may affect our understanding of how altered mitochondrial function brings about senescence. Mitochondria have multiple functions. They are the sites of some aspects of intermediary metabolism (citric acid cycle, beta-oxidation of fats), the urea cycle, heme biosynthesis, calcium storage, and, most important, the production of ATP. This fungible energy molecule ATP is generated by oxidative phosphorylation in the inner membrane of the mitochordrion. This process is carried out by a set of at least 87 proteins arranged into 5 major complexes integrated into the membrane (figure 11.2). At least 74 of these proteins are coded for by the nucleus (table 11.1). The inner membrane of the mitochondrion has multiple copies of this respiratory chain scattered throughout it. The basic point of the operation is that the oxidation of the organic molecules we use as fuel results in the reduction of certain important coenzymes (i.e., NADH in figure 11.2), which are then oxidized (i.e., NAD+) by complex I, and the protons generated in the reaction are transported to the mitochondrial intermem-

2H+

4H+

Complex III

Complex IV

4H+

Complex I

NADH

UQ Complex II

½ O2

NAD Succinate

brane space. Their electrons are then transferred to coenzyme Q10. Electrons from other reduced compounds generated by the TCA cycle can enter at complex II, which is succinate dehydrogenase. The electrons are shuttled from complexes I and II via ubiquinone to complex III. Oxygen is the final electron acceptor and is reduced to water at complex IV. The high proton gradient across the inner membrane provides the power that enables complex V to phosphorylate ADP, thus creating a new molecule of ATP and water. Protons may escape from the intermembrane space back to the matrix in at least four ways. First, they can be translocated into the outer compartment by complex V and their chemiosmotic energy captured to drive the synthesis of ATP as described. However, the mitochondrion also contains at least five different types of uncoupling proteins (UCP) that have unique tissue-specific distributions and serve somewhat different functions (Goglia and Skulchev 2003; Krauss et al. 2002). What follows is our current (and somewhat tentative) understanding of the roles of these UCP in modulating mitochondrial function. UCP1 catalyzes the net translocation of protons from the intermembrane space of mitochondria into the matrix, but without passing through complex V. Instead, the chemiosmotic energy is released as heat. This happens in all cells but is particularly pronounced in the brown fat cells of the body, and it represents the second escape path

4H+

Complex V

H2O

Fumarate ADP + Pi

ATP

Figure 11.2 The electron transport system of a typical mitochondrion. The protons (H+) are pumped across the inner membrane by the coupled redox reactions in each complex. The protons are at a high concentration in the inner matrix. They can be transported down the concentration gradient in a controlled manner in complex V, and the released energy is used to generate ATP.

11.3 Mitochondrial Damage Theories

401

Table 11.1 Two Genomes Contribute to the Respiratory Chain Complex I: II: III: IV: V:

NADH ubiquinone oxidoreductase Succinate ubiquinone oxidoreductase Ubiquinol cytochrome c oxidoreductase Cytochrome c oxidase ATP synthetase Totals

for the protons. The third path is represented by the remaining four UCPs. These proteins regulate a number of important cell functions by their ability to uncouple oxidative phosphorylation by means of mitochondrial proton leak in response to specific metabolic signals. These functions include the regulation of substrate oxidation, ATP production and turnover, as well as regulating the buildup of reactive oxygen species (ROS). For example, some of the UCPs are thought to protect the inner membrane by translocating reactive fatty acid peroxides, which might otherwise cause oxidative damage to the organelle, out of the inner space (Goglia and Skulchev 2003). UCP2 and UCP3 are induced to uncouple oxidative phosphorylation when confronted with high levels of negatively charged molecules such as superoxide. Under these circumstances, the UCPs transport negatively charged superoxide from the inner space to the intermembrane space where they might be inactivated (Echtay et al. 2002). These are clearly antioxidant defense mechanisms. The fourth manner in which protons pass out of the inner space is by inadvertent leaks and slips (G. C. Brown, 1992). Leaks occur when there is a passive leak of protons across the membranes, perhaps due to details of membrane structure, and which likely play a role in senescence, as I discuss below. Redox slips occur when an electron goes through a complex that pumps protons (i.e., I, III, or IV) but inexplicably does not pump a proton. There is empirical evidence suggesting that these leaks and slips may preferentially occur at complex I and at the boundary of complexes II and III. The latter supposedly comes about because the ubiquinone picks up two single

Nuclear encoded

Mitochondrial encoded

Total

38+ 4 10 10 12 74+

7 0 1 3 2 13

45+ 4 11 13 14 87+

electrons for transport to complex III. But this two-step process increases the possibility of redox slippage of the one-electron states and the subsequent production of reactive oxygen species (ROS). Thus the perfect system shown in figure 11.2 is not a realistic picture of the mitochondria. There is a constant flux of protons and electrons flowing in both directions, mostly under some sort of regulatory control, but occasionally not. Uncoupling allows heat production while also allowing for the reduction of ROS production because the complexes that generate ROS do so more readily when the proton gradient is high. Almost all of the ROS formed in the cell is produced by these means in the mitochondrion. I discussed in chapter 10 the deleterious effects of these ROS. Consider the situation: The process by which we generate the energy needed to maintain cells is the same process that generates the molecules that will damage cells. The generation of these ROS at this site and the pattern of damage that this brings about lies at the heart of this chapter. It is worth pointing out some of the changes that occur in mitochondrial function as the cell’s energy demand varies. One might think that putting high ATP demands on the cell, such as during physical exercise, would increase ROS production. Some of my friends use this reasoning as an excuse not to exercise. They are wrong (table 11.2). ROS production plummets when ATP demand is high. The slight decrease in membrane potential associated with high energy demand likely stems from the rapid use of the protons for ATP synthesis; they have no chance to accumulate. This decreased proton concentration or membrane potential brings about a

402 Chapter 11 Systemic Theories of Senescence Table 11.2 Mitochondrial Function Varies with Cellular Energy Demand Cellular energy demand

ATP synthesis

Membrane potential

ROS production

Respiration

Proton leak

ATP turnover

Substrate oxidation

Low

Very low

Very high

Very high

Very low

Very high

Very low

Very low

High

Very high

Somewhat lower

Much lower

Very high

Very low

Very high

Somewhat higher

Source: compiled from data presented in Nicholls (2002). Note: ROS, reactive oxygen species.

nonlinear and striking decrease in the level of inadvertent electron leaks, and this likely accounts for the ROS decrease. It has been suggested that artificially lowering the membrane potential slightly to just below the threshold needed for superoxide production might be beneficial (Skulachev 1996). It should be noted that mitochondrial activity is believed to be regulated by process other than proton concentration. These other processes (ATP/ ADP ratio, phosphorylation of cytochrome c oxidase, stress signals) will also alter ROS production ,but in response to cell signals other than the ones discussed here (Kadenbach et al. 1999, 2000).

11.3.2 Mitochondrial DNA and Senescence Each mitochondrion can contain numerous mtDNA molecules, with estimates of the total number ranging between 1000 and 5000 molecules per cell (Wallace 1992). The precise mechanisms regulating mtDNA copy number are still unclear, but all the major factors known to be involved are coded for by the nuclear genome. If a human cell in culture is rendered devoid of mitochondria and then repopulated with equal portions of two mitochondrial populations, the population with the full-length mtDNA repopulates cells less efficiently than does a population of partially deleted mtDNA (Moraes 2001). This finding implies that newly mutated mtDNAs may have a replicative advantage over their normal counterparts in the ordinary somatic cell. Are the mtDNA molecules under oxidative attack? The circular DNA molecules are not

wrapped about specific nucleoproteins as in the nucleus but exist naked and open to the mitochondrial environment, which is, as just discussed, replete with ROS. This environment led Harman (1972) to propose that mtDNA was a probable target for attack by ROS. It has been estimated that the amount of ROS generated by the mitochondria under normal conditions may be equivalent to only 0.1% of the oxygen they consume, although this can rise to as high as 4% under experimental conditions in which antibiotics are used to inhibit the respiratory chain. The steady-state concentration of superoxide in the mitochondrial matrix is probably about 5- to 10-fold higher than in the cytosolic and nuclear spaces (Cadenas and Davies 2000). Probably because of the proximity to such high levels of reactive radicals, oxidative damage to mtDNA has been estimated to be substantially higher than the damage to nuclear DNA (Shigenaga et al. 1994). (It is likely that technical problems with DNA isolation led to artificially high estimates of oxidative damage to mtDNA in the earlier literature; better methods have yielded lower absolute values, which are still elevated relative to nuclear DNA [Van Remmen and Richardson 2001].) This finding of mtDNA damage was confirmed by a study that measured the extent of H2O2-induced damage in specific regions of the cell’s mtDNA and nuclear DNA in vitro (Yakes and Van Houten 1997). The results showed that oxidative stress induces multiple defects in the cell’s mitochondria. The expected and most obvious result was that all but the lowest level of stress resulted in mtDNA from oxidatively

11.3 Mitochondrial Damage Theories

stressed cells having two to three times as many DNA lesions as did nuclear DNA from the same cells. The damage occurred rapidly. DNA lesions induced by short-term exposures (15 minutes) to peroxide were efficiently repaired in both DNA compartments. However, longer exposures (60 minutes) resulted in damage to the mtDNA that was not repaired, even though the (lesser) damage done to the nuclear DNA during the same exposure was repaired within 1.5 hours. These long-term exposures also resulted in a dosedependent loss of mitochondrial redox function in oxidatively stressed cells and permanent growth arrest, eventually leading to apoptosis of some cells. A diagram of the differential damage cascade induced by oxidative stress is shown in figure 11.3. The extraordinary sensitivity of mitochondria to oxidative stress is not something peculiar to just this one study or to just cultured cells. Soma-

H2O2 etc.

403

tic mtDNA mutations in humans—consisting of deletions, duplications, and base substitutions— are preferentially found in certain regions of the brain (basal ganglia, cerebral cortex), skeletal muscle, and heart. All three of these postmitotic tissues have a high rate of oxidative metabolism, and there are correlative data suggesting that cells generating high levels of oxidative phosphorylation also generate high levels of ROS. The temporal association of mutant mtDNA with certain types of damage by-products (such as 8-oxoG) generated only by oxidation is reasonable evidence that the observed mutations are the result of oxidative damage (Wallace et al. 1995). Deletions of the mtDNA are also correlated with increased oxidative stress (Wei et al. 1996). Treatment of cells or organisms with antioxidants seems to relieve oxidative stress. There have been many other experiments done with various species both in vivo and in vitro, using a variety of

ROS Nuclear DNA

mtDNA low nuclear DNA damage high mtDNA damage

low mtDNA repair

mitochondrial function

high nuclear DNA repair

Loss of cell function

Figure 11.3 The different events that result from the fact that DNA repair activity levels are much higher and more complete in the nuclear DNA compartment than they are in the mitochondrial DNA compartment. Damage in the latter compartment accumulates much more quickly than in the former, and so critical thresholds below which function is lost are reached much more quickly in the mitochondria. (Redrawn from data in Yakes and Van Houten 1997.)

404 Chapter 11 Systemic Theories of Senescence experimental approaches, which agree with these essential results (Barja 2002; Cadenas and Davies 2000; Van Remmen and Richardson 2001; Wallace 1999). Other types of evidence are presented later in this chapter. Taken together, they form robust empirical support that mtDNA of healthy cells is relatively more sensitive to oxidative stress than is the nuclear DNA.

11.3.3 Mitochondria and Oxidative Stress There is also evidence that the sensitivity of mitochondria to oxidative stress increases with age. It seems likely that both the initial sensitivity and the age-related sensitivity of mitochondria are brought about by similar or related mechanisms. There are several possible alternative mechanisms. For example, the mutant mtDNAs are preferentially replicated and so they increase in old cells; the damaged mitochondria have a decreased degradation rate and so they accumulate in old cells; or the functional activity of the electron transport chain may be inhibited by oxidative stress and so the repair functions are inhibited by lack of available energy, and so forth. I discuss the evidence supporting these concepts below. The concept that oxidative stress to the mitochondria increases with age is an attractive hypothesis, provided it is supported by empirical data, because it offers the possibility of directly accounting for the increased vulnerability of aged cells and organisms to oxidative insult that has been documented in other chapters of this text. This relative difference between young and aged cells leads to the belief that the mitochondria– oxygen radical theory offers a potentially powerful explanatory insight into the onset and progress of senescence (de Grey 1999; Harman 1972; Miquel et al. 1980). The mitochondria have several antioxidant defense systems (ADS) that help protect them against the effects of ROS. If oxidative stress plays a major role in mitochondrial function, then manipulating these ADS may enable us to deduce the role of ADS. One of the major enzymatic defenses is MnSOD, and its effects have been studied in flies and mice (see chapter 7). Flies homozygous for a

null mutant of MnSOD, and thus having no enzyme activity, show a higher relative level of developmental deaths and have a variable but very short adult life span (Duttaroy et al. 1997). Homozygous MnSOD mutants are exquisitely sensitive to exogenous oxidative stresses such as paraquat or oxygen. Flies heterozygous for the null MnSOD gene, and thus presumably possessing 50% of the normal activity, live a more normal life span but are very sensitive to exogenous oxidative stress. Overexpression of the MnSOD gene in a normal animal, presumably leading to 150% of the normal level, allows the animal to live substantially longer than do controls (Parkes et al. 1999; Sun & Tower 1999). Overexpression of the MnSOD gene in also seen in the selected long-lived La strain of Drosophila (Arking et al. 2002a). Flies thus show a direct correlation between MnSOD activity and longevity, suggesting that the scavenging of ROS plays an important role in protecting the mitochondria. Manipulation of the MnSOD gene in mice leads to a somewhat different conclusion (Van Remmen et al. 2003) Knockout mice heterozygous for the MnSOD (i.e., Sod2) gene have only about 30–80% of the normal levels of MnSOD activity. There is no compensatory up-regulation of other ADS enzymes, so the mice have a compromised defense against oxidative stress. These Sod2+/- mice have reduced respiration, increased sensitivity to both endogenous and exogenous oxidative stress, and increased levels of oxidative damage to the mtDNA over their life span. These mice show the same survival pattern as do wildtype animals over the first 25 months of their life, although they appear to enter senescence a few months earlier than controls, and may have a shorter senescent span (Van Remmen et al. 2003). Overexpression of various ADS enzymes has generally shown no real benefit in terms of an increased life span in Sod2 heterozygous mice relative to wild type or nontransgenic mice, although they do express an increased resistance to exogenous oxidative stress. Thus, the MnSOD-based defenses of mice seemingly play a different, less direct role in defending the mitochondria against oxidative stress than is seen in flies.

11.3 Mitochondrial Damage Theories

Is there a greater redundancy of antioxidant defense in the mitochondria of mice relative to shorter lived animals such as the fly or worm? One suggestion that this might be the case is afforded by the methionine sulfoxide reductase A (MsrA) gene. Mice homozygous for the knockout mutation of MsrA had a significantly shortened life span and a greater sensitivity to exogenous oxidative stress (hyperoxia) than did heterozygous or wild-type mice (Moskovitz et al. 2001). Perhaps mammals have other types of specific repair enzymes such as this that can pick up the slack induced by the knockout of a general type of ADS enzyme as MnSOD. If so, then perhaps mammals depend on a more layered defense than do the shorter lived worms and flies.

11.3.4 Mitochondria and DNA Repair Mitochondria have several DNA repair systems, of which the base excision repair (BER) system is the best characterized (Mandavilli et al. 2002). Only a few genes from this system have been identified in short-lived laboratory models (worm and fly), but this may be a technical matter and not a robust explanation of short life. Oxidative stress can produce more than 100 different types of base modifications, one of the most common of which in both nuclear and mtDNA is 7,8dihydro-8-oxoguanine (8-oxoG), which is repaired via the BER system. Recent studies show that mammalian mitochondria efficiently remove and repair 8-oxoG from their genomes, but with the remarkable twist that the repair efficiency of this one system increases with age (Stevnsner et al. 2002). Yet 8-oxoG accumulates during aging in the mitochondria to concentrations about 15 times that found in nuclear DNA. The two observations can be reconciled by hypothesizing that the rate of oxidative damage in the mitochondrion must be increasing with age such that it overwhelms the mitochondrion’s defenses. Mitochondria appear to have the same types of DNA repair systems as the nuclei but with much less redundancy in the number of components within each system. This lack of redundancy may play a role in the greater susceptibility

405

of the mitochondrion relative to the nucleus. Recall, for example, that loss of apparently redundant nuclear DNA repair enzymes in yeast can lead to 10- to 1000-fold increases in mutation rate (chapter 7). DNA polymerase gamma is the only polymerase known in mitochondria, and it plays roles in both repair and replication, which are done by different enzymes in the nucleus. Any damage to this enzyme or to the factors controlling it will lead to widespread failures in the repair and replication of mtDNA. Flies carrying an engineered mutation in this one molecule have a signficantly shorter life span and decreased reproductive capacity (Kaguni, cited in Mandavilli et al. 2002), effects that would likely be significantly modulated in a redundant system. A similar situation applies with the DNA ligase enzyme. Some forms of oxidative damage to mtDNA are repaired by an AP endonuclease, which, although present, is much more difficult to activate in older animals than in younger ones (Mandavilli et al. 2002). Thus, it is not only the presence in the mitochondria of specific DNA repair enzymes that are important but also the systems controlling their activation. The electron transport system will not operate efficiently if any of its 87 or so proteins are missing or in short supply. The transcription of mtDNA is very sensitive to certain species of ROS, and ROS will significantly inhibit the transcription of the 13 mitochondrial genes both in vitro and in vivo (Kristal and Yu 2000). It is likely, but not proven (see table 7.12), that there is a signal mechanism regulating nuclear and mitochondrial transcription so that the 87 or more required proteins are made in the required amounts. The slowdown of mtDNA transcription might have the effect of decreasing the amount of ATP necessary for various synthetic and repair functions as well as increasing the endogenous level of ROS production (see table 11.2). This would result in the doubled insult of increased oxidative stress and decreased preventive and repair processes. Correlative data consistent with this concept is found in the fact that human skin fibroblasts (donors of which ranged in age from 5 months gestation-age to 103 years old) in culture display a complex mitochondrial-based aging

406 Chapter 11 Systemic Theories of Senescence phenotype characterized by an age-related decline in mitochondrial protein synthesis, an agerelated decline in respiration rates possibly due to a loss of respiratory control, and a marked agedependent decline in the coupling efficiency of respiration and phosphorylation (Greco et al. 2003). It is reasonable to conclude from the above discussion that oxidative stress has multiple effects on different aspects of the mitochondrion and initiates a cascade of detrimental losses of function, all of which decrease the mitochondrion’s efficiency. It is for this reason that figure 11.3 was drawn with the positive feedback loop from the initial decrease of mitochondrial function back to the production of more ROS, more mtDNA damage, and so forth. Positive feedback cycles are inherently unstable, and the feedback cycle in mitochondria is no exception. This “fast-forward” feature inevitably results in a significant loss of cell function. The study of mitochondrial diseases reveals that this positive feedback loop may be too simplistic (Wallace 1999). The first mitochondrial diseases to be molecularly analyzed were Leber’s hereditary optic neuropathy (LHON) and dystonia, both of which result from the same mtDNA missense mutation. The expression of the mutation varies considerably. LHON presents as a sudden onset blindness in mid-life caused by death of the optic nerve. Patients with dystonia present early in life with a generalized movement disorder, impaired speech, and impaired mental function, and short stature, all of which are frequently accompanied by degeneration of certain brain ganglia. Dystonia is thought to be correlated with the presence of a high level of mutant mtDNA, whereas LHON is thought to be associated with a lower level. The mutation causes substantial reduction in complex I activity, yet its phenotypic manifestation depends critically on the frequency of the particular mtDNA mutation in the cell’s total population of mitochondria. This apparent dependence of the phenotype on the quantitative distribution of the mtDNA mutation is seen in other mitochondrial diseases. The generation of mtDNA molecules that have large deletions of their coding sequences has

been much studied in an attempt to correlate the levels of deletions with the loss of function and/ or the presence of pathologies. In patients with particular syndromes of muscle degradation (e.g., “ragged red fiber disease,” which is similar to dystonia), particular deletions can be found at very high levels (~73–98%) of the mtDNA genomes present in muscle fiber (Shoffner et al. 1990). In other cases, the level of mutated mtDNA may approach 50% of the patient’s mtDNA, yet the patients remain reasonably functional (Wallace et al. 1995). Some normal aging people have the same mtDNA mutation found in people with a mitochondrial genetic disease, and this deletion accumulates with age (Arnheim and Cortopassi 1992; Zhang et al. 1992). These findings imply that the mtDNA mutations are just one of several processes that decrease the reserve capacity of the organism. The delayed onset and variable symptoms of some mitochondrial diseases must be due to processes that affect the frequency of the mutant in the cell. We need a mechanism that can explain the existence of quantitative thresholds. Mice whose mitochondria have been genetically engineered so as to have a proofreading-deficient version of Polga, the nucleus-encoded catalytic subunit of the mtDNA polymerase, develop a mutator phenotype with significant increases in the numbers of point mutations and amounts of deleted mtDNA (Trifunovic et al. 2004). Such mice exhibit a premature aging phenotype as evidenced by premature loss of function in many tissues of the body. Such mice provide us with another presumptive causative link between the rate of mammalian aging and the level of mtDNA mutations. It is thought that the damaged mtDNA molecules are unevenly distributed among the cells in a tissue, and the function and survival of cells that have high numbers of damaged mitochondria may be reduced. This may be related to the tissuespecific protein composition of miotochondria (Jensen et al. 2004). Such mosaicism has been shown in the adult human brain (Corral-Debrinski et al. 1992). More recent evidence in muscle from older monkeys also points to a mosaic of normal and abnormal myofibrils.

11.3 Mitochondrial Damage Theories

11.3.5 The mtDNA–Oxidative Stress Connection Oxidative stress can induce the formation of various types of base substitutions and/or deletions in mtDNA. However, measurements of the frequency of such mutants in the normal cell indicate that they do not affect enough of the cell’s mitochondria to bring about any significant aging phenotype. During aging, however, the mtDNA damage appears to accumulate exponentially, and this rapid rise late in life suggests that the mutant mtDNAs are preferentially replicated. There are both theoretical (de Grey 1999) and empirical (Moraes 2001) reasons to believe that differential replication of different-sized mtDNA molecules takes place, and this supports the idea that mutant mtDNAs accumulate with age. An alternative mechanism leading to the same outcome was proposed by de Grey (1997, 1999). His hypothesis is based on the assumption that mitochondria with reduced respiratory function caused by a deletion or mutation will consequently inflict less damage on their own membranes than will normal mitochondria. As a result, the mutated mitochondria will suffer less frequent lysosomal degradation than the normal mitochondria, and hence the defective mitochondria will preferentially survive and replicate. Eventually, de Grey proposes, once a deletion or mutation occurs in a mitochondrion of a nondividing cell, it will rapidly populate that cell and thereby destroy the cell’s respiratory capability. De Grey’s (1998) hypothesis, called the “survival of the slowest,” is still being tested. However, as discussed by de Grey (1998), its implications are consistent with observed effects of other oxidative-damage models (see, e.g., the MARS model discussion in chapter 13). Cells with destroyed respiratory capability must rely on glycolysis for ATP production and can only stabilize their NAD+/NADH ratio by expelling electrons from the cell. If the rate of electron efflux from the cell exceeds the electron-accepting capacity of the electron acceptors in the plasm, then reactive free-radical species are likely to form and to give rise to increased peroxidation of serum lipids such as the low-density lipoproteins (LDL).

407

When peroxidized lipid is imported into a healthy cell, it is destroyed by the cell’s antioxidant defenses. But this task is thought to overload the defective cell’s ability to degrade the pro-oxidants that it is generating as a result of aerobic metabolism. This increases the oxidative stress on the cell and sets off an unstable positive feedback system in which the increase in the number of damaged mitochondria causes the oxidative stress to increase even more, further increasing the number of damaged mitochondria. The “reductive hotspot” hypothesis of mitochondrial damage is plausible and consistent with much known data, including the muscle fiber data discussed below (de Grey 2002), but it is not yet completely accepted (see Lightowers et al. 1999). A special issue of Aging Cell dealing with the overall topic of mitochondria and aging was published in February 2004, and the interested reader is referred there for more details (Nicholls, 2004).

11.3.6 Sarcopenia as a Mitochondrial Disorder Sarcopenia is a loss of muscle mass and function that occurs in older individuals, and it has clinical implications for mobility, energy intake, respiration, and independence. In humans, there is about a 40% loss of muscle mass between the ages of 20 and 80 years. The precise nature of the mechanism(s) responsible for this age-related loss of muscle mass is not yet known, but it appears to involve both a general decline in skeletal muscle protein synthesis and an actual loss of muscle myofibrils. Physical exercise alleviates but does not completely overcome this process. In this section, I focus on the loss of fibrils and the role mtDNA deletion mutants play in bringing about this phenotype (McKenzie et al. 2002; Pak et al. 2003). Muscle wasting has been described in rodents and in nonhuman primates. The diagnostic agerelated criteria are a significant reduction in muscle cross-sectional area, a loss of muscle mass, and a decrease in fiber number. The first two criteria might be brought about by either a decline in skeletal muscle protein synthesis or an

408 Chapter 11 Systemic Theories of Senescence actual loss of fibers; decrease in fiber number can be due only to actual loss. The magnitude of changes that take place in rat muscles as a result of sarcopenia are illustrated in table 11.3. Between mid- and late-life, the affected muscle undergoes a 45% reduction in weight, a 30% reduction in muscle cross-sectional area, and a 27% reduction in the number of muscle fibers. These significant differences imply major changes in mobility and body metabolism. The mitochondrial diseases discussed earlier (Wallace 1999) suggested an association between muscle fiber malfunction and the level of mutated mtDNA. Researchers wondered if a similar association held in the case of sarcopenia. It was known that a specific mtDNA deletion mutant (mtDNA4977) commonly accumulated with age in humans and other primates. The initial investigations into the etiology of sarcopenia showed that the frequency of this deletion mutant was less than 0.1% of the total mtDNA present in the cell homogenates, too low a number to indicate that the mutation played anything other than a minor role. However, in situ hybridization studies of aged muscle tissue showed that the distribution of the mtDNA4977 deletion mutant was not random in the muscle but was found at high concentrations in some cells while being very low in most others. This focal distribution implied that the deleterious effects of the mutant mtDNA might be restricted to certain cells only, restoring the possibility of a mechanistic link between the mutation and sarcopenia. (McKenzie et al. 2002; Pan et al. 2003).

Mutations in the mtDNA will likely affect the functioning of the electron transport system (ETS). Use of various histochemical stains allowed the investigators to assay the metabolic status of serially sectioned complete myofibrils. Thus one could stain one section of the cell and determine whether it had normal amounts of cytochrome c oxidase activity (complex IV, COX+), indicative of a functioning ETS, or not (COX-). Using another stain on an adjacent section of the same cell, one could determine whether it had normal levels of succinate dehydrogenase activity (complex II, SDH+), or whether it was hyperactive (SDH++). The logic of these choices is that cells with a normal ETS should be COX+/ SDH+. Cells with an abnormal ETS should be COX-/SDH++ because the decreased respiratory activity of the mutant mitochondria would induce a compensatory upregulation of the citric acid cycle and thus result in SDH hyperactivity. The results were instructive. The percentage of abnormal COX-/SDH++ cells increased about 5 to 10-fold in late life in both rats and monkeys (Pak et al. 2003). Laser microdissection and polymerase chain reaction/hybridization analysis of adjacent sections allowed the determination of the mtDNA genotype in the COX-/ SDH++ cells. The mtDNA deletion was observed only in fibers that were ETS abnormal. Normal fibers did not contain the mtDNA deletion mutant. (The recycling of NAD necessary for the damaged fibril to continue functioning might be accomplished by other metabolic cycles/organelles linked to the mitochondrion [e.g., the glyoxalate cycle of yeast; figure 7–6].)

Table 11.3 Effects of Sarcopenia on Rat Skeletal Muscle Age

Muscle weight (g) Cross-sectional area (mm2) No. of fibers Estimated no. of RR fibers

5 months

18 months

36–38 months

1.4 59 10,426 0

1.7 65 10,075 4

0.75 45 7,606 567

Source: from data on the rectus femoris muscle of FBN rats presented in McKenzie et al. (2002). Note: RR, ragged red muscle

11.3 Mitochondrial Damage Theories

The sequential study of the long (~1000 mm) ETS abnormal muscle fibers showed that many fibers display a gradual atrophy as indicated by a decline in cross-sectional area (McKenzie et al. 2002; Pak et al. 2003). Some fibers atrophied to the point that they were broken at one or several spots, leaving noncontinuous remnants of the original fiber. These segmental abnormalities were often accompanied by a reduction in the number of nuclei, suggesting the possible presence of cellular apoptosis. These observations, taken together, give rise to a model of how mtDNA mutations may lead to sarcopenia (see figure 11.4). Presumably, oxidative damage gives rise to an mtDNA deletion that might have a replicative advantage over larger wild-type genomes. As the number of mutant mtDNA molecules increase, the regions of the muscle fibers in which they are located develop energy deficiencies. These localized deficiencies lead to increased ROS production and thus increased oxidative stress. The cell’s efforts to obtain energy by upregulation of the nuclear subunits of the ETS will fail to alleviate the situation. The fibril atrophies and eventually disappears. Note that during the entire process, the abnormal mtDNA molecules and abnormal ETS mitochondria would be focally located and would not constitute a large percentage of the total number of mitochondria present in the muscle. The sensitivity of the mitochondrial genome to oxidative damage leads to the senescence of the skeletal muscle.

11.3.7 Mitochondria of Long-lived Animals The studies discussed so far are consistent with the view that loss of mitochondrial function is somehow associated with oxidative stress. In chapter 4 I discussed the case of birds, in which a high rate of oxygen consumption is combined with a high maximum life span, and pointed out that an investigation of their physiology might offer insight into previously unsuspected protective mechanisms. Studies have been done on rats and pigeons, since the two animals are of similar body size but differ significantly in their maxi-

409

mum life spans (4 years for rats, 35 years for pigeons). Pigeons convert significantly less (~74% less) oxygen into free radicals in the mitochondria than do rats (Herrero and Barja 1997b). This low rate of radical production is a result of a very low proton leak from complex I in birds relative to rats. Examination of mammals with life spans ranging from 3.5 to 46 years shows that life spans are tightly and inversely correlated with the rate of free-radical leak and not with other variables such as ADS enzyme levels or antioxidant levels (figure 11.5) (Barja 2000; 2002). The higher longevity of long-living animals must be due in part to the capacity of their mitochondria to decrease free-radical leakage at the respiratory chain. This translates, of course, into a lower rate of oxidative damage in the mtDNA. Another characteristic of long-lived vertebrates is a low degree of fatty acid unsaturation in cellular membranes, particularly the inner mitochondrial membrane (Pamplona and Barja 2003). This low saturation is advantageous because it decreases membrane sensitivity to lipid peroxidation and indirectly reduces the risk that other tissues will undergo secondary lipoxidationderived damage. The lipid composition of a membrane is cumbersome to discuss, but it can be summarized into a useful number, the double bond index (DBI), based on the relative molar quantities and characteristics of the fatty acids involved. Rats fed a diet with a low DBI nonetheless had significantly higher DBI values, relative to pigeons, in their heart and liver mitochondria as well as in their skeletal muscle. The same phenomenon is found when comparing mice with canaries or parakeets. It is also found when comparing long-lived to short-lived mammals. Thus, a low degree of unsaturation of cellular and mitochondrial membranes seems to be a characteristic of longevous vertebrates in general and probably evolved via selective effects on the activities of the various fatty acid synthesis enzymes involved in regulating the degree of unsaturation of the polyunsaturated fatty acids (PUFAs). The physiological significance of a low DBI may lie in the fact that PUFAs have a high number of double bonds and are very sensitive to

410 Chapter 11 Systemic Theories of Senescence

A

B

C

D

E

Figure 11.4 A model of the sequence of events by which oxidative damage to mitochondrial DNA (mtDNA) gives rise to functional inactivation of the electron transport system (ETS) and eventually to muscle fiber loss. (A) A group of normal muscle fibers with a cytochrome oxidase c+/succinic dehydrogenase+ (COX+/SDH+) phenotype. An mtDNA deletion giving rise to an ETS abnormal portion of the muscle fiber occurs in the darker area of B, giving rise to a COX-/SDH+ phenotype. Preferential replication of the mutant mtDNA genome relative to wild type gives rise to an increased area of abnormal COX-/SDH++ cells, as in C–E. These abnormal cells undergo atrophy of both fibrils and nuclei (D) and finally break (E). The process presumably continues spreading through the muscle fiber. (After McKenzie et al. 2002.)

oxidation. Oxidation of PUFAs leads to the formation of hydroperoxides and endoperoxides, which then break down to yield a variety of reactive intermediates. To test this hypothesis, rats were fed a diet rich in saturated fat (coconut oil) or in unsaturated fat (menhaden oil). The DBI

of the rat heart mitochondria was lower in the saturated than in the unsaturated group. These saturated mitochondria were significantly less sensitive to oxidative damage than were those of rats fed the unsaturated diet (Pamplona et al. 2002). Certain enzymes involved in fatty acid

11.3 Mitochondrial Damage Theories

411

13 12 11 10 9

Mouse

5 8-oxodG/10 dG in mtDNA

8 7 Rat

Guinea pig

6 Rabbit

5

Sheep

Pig

4 3

Cow

Horse

2 1 0

10

20

30

40

50

MLSP (years)

Figure 11.5 There is an inverse correlation between oxidative damage in the heart mitochondria and the maximum life span (MLSP) in eight mammalian species (r = –0.92, P < .001) A similar correlation is noted for brain mitochondria, but no significant correlation is seen between MLSP and oxidative damage in heart or brain nuclear DNA (compare to figure 6.6). (After Barja and Herrero 2000.)

biosynthesis appear to have their activities adjusted by natural selection so that longevous mammals and birds naturally synthesize fewer PUFAs with a low DBI without going on a special diet. This strategy does not alter the overall level of PUFAs or the average chain length, as these traits are important for other aspects of normal membrane function (fluidity, viscosity, etc.). It should be noted that caloric restriction shifts the body’s synthesis of PUFAs away from the highly unsaturated arachidonic acid and toward the less unsaturated oleic and linoleic acids (Laganiere and Yu 1993). The mitochondria of longevous vertebrate animals are characterized by a low rate of endogenous ROS production, particularly from com-

plex I, and a characteristically low DBI value for their PUFAs. Together, these two adaptations lead to a low mitochondrial lipid peroxidation rate, a low mtDNA damage rate, and a slow rate of aging. There are other nonmitochondrial adaptations that indirectly reduce the generation of mitochondrial ROS and thus support extended longevity. Table 11.4 illustrates some of these key functional differences between mice and parakeets. The two animals are approximately the same size, but the parakeet has a much larger heart with respect to its body size than the mouse. Yet the bird’s heart mitochondria produce substantially less (~63%) ROS than the mouse. Some of this lower production of ROS is undoubtedly

412 Chapter 11 Systemic Theories of Senescence Table 11.4 Nonmitochondrial Adaptations That Contribute to the Long Life of Birds Trait Body size (g) Heart size (mg) Heart/body size (%) Heart mtROS production (mM H2O2 min•mg protein)

Mouse

Parakeet

Mouse/Parakeet

36 ± 2.8 200 ± 13 0.6 1.6

32 ± 0.9 467 ± 42 1.5 0.6

1.125 0.428 0.4 2.67

Source: from data of Herrero and Barja (1998). Note: mtROS, mitochondrial reactive oxygen species.

due to the lower proton leak and fatty acid differences discussed above. But some of it is likely due to the fact that the parakeet’s heart does less work in delivering the same volume of blood as does the mouse’s heart. The proton concentration in the bird heart mitochondria is thus probably not at a maximum level, and this gives rise to a large nonlinear reduction in the ROS produced by the mitochondria.

11.3.8 Invertebrate Mitochondria and Senescence Similar but perhaps not identical patterns of mitochondrial damage with aging are seen in invertebrate models. In the nematode, Melov and colleagues (1995) have shown that the frequency of mtDNA deletions also increases with age. The deletions increased more slowly in the age-1 longlived mutant, which has higher levels of antioxidant resistance; this finding certainly supports the oxidative-damage origin of mtDNA mutations. However, a different pattern of damage is seen in a standard laboratory strain of Drosophila (Calleja et al. 1993). No age-related increase in mtDNA frequency is observed; the level of deletions is low (about 1%) and constant throughout life. However, there was a significant decrease in the steady-state levels of several important mitochondrial RNA transcripts in Drosophila, including those for both rRNA and certain oxidative enzymes. Loss of flight ability in older flies was correlated only with low levels of mtDNAencoded cytochrome oxidase subunits, as there was no difference between the two groups re-

garding nuclear-encoded subunit prevalence (Schwarze et al. 1998a,b). This finding may suggest that there is not a tight coordination between mitochondrial and nuclear activity in Drosophila. The loss of mitochondrial function is positively correlated with the levels of both ATP and ROS production by the fly’s mitochondria. The mitochondria of old Drosophila are incapable of producing the high levels of energy characteristic of young flies, but this decreased efficiency of energy production comes about via a different mechanism from that observed in mammals and nematodes. There is more than one way to make some organelle inefficient, and we should be careful not to overgeneralize, especially across phyla. In normal-lived flies, the mitochondria produce significant amounts of superoxide on the matrix side of the inner membrane from complex I, as well as significant amounts on the intermembrance space from complex III (Miwa et al. 2003). Situations bringing about a mild reduction in the proton electromotive force greatly lowered the superoxide production from complex I. The mitochondria of long-lived flies are different from those of normal-lived ones. The long-lived La strain of flies have an enhanced level of antioxidant defense enzymes (chapter 7); however, their mitochondria can impart longer life to otherwise normal-lived animals (see table 7.3). Both normal- and long-lived flies have an agerelated increase in the endogenous ROS produced by their mitochondria, but Ross (2000) has shown that the age-related increase in the La mitochondria is significantly lower (~20–40%) than the normal-lived flies (table 11.5). This decreased ROS production, coupled with their higher ADS

11.3 Mitochondrial Damage Theories

enzyme levels, likely results in a low level of ROS actually available to damage the cell (see table 7.5 for details). As the mitochondion requires the incorporation of at least 1000 nuclear proteins, one method of increasing mitochondrial efficiency would be selecting genetic variants of nuclear-encoded mitochondrial proteins. Such a process is reported to underlie the greater longevity of a particular strain of Drosophila simulans (a close relative of the more common D. melanogaster species used by many geneticists) relative to the appropriate controls (Melvin et al. 2005). This particular strain has five coding changes in the genes coding for certain cytochrome c subunits, and these changes result in various amino acid replacements in the protein subunits. The differences in cytochrome c oxidase activity among the experimental and control strains are correlated with these amino acid changes in the complex IV loci. The whole-organism oxygen consumption and the mitochondrial density are lowest in this strain and are consistent with the conclusion that this strain’s mitochondria are more efficient than the controls. Presumably, the more efficient mitochondria lose fewer electrons by slip or leakage, and so they can produce lower amounts of ROS but the same amount of ATP while having a lower oxygen consumption. There may well be other amino acid replacements in the nuclear-encoded genes that could achieve this same goal via affecting the efficiencies of other complexes. In humans, specific mitochondrial haplotypes are

associated with extended longevity, and perhaps similar mechanisms are being used in those cases (de Benedictis et al. 1999; G. Rose et al. 2001).

11.3.9 Other Interventions That Affect Mitochondrial Function 11.3.9.1 Vitamin E

There are interventions other than those discussed above that can significantly alter mitochondrial function. Vitamin E is a lipophilic antioxidant and might exert some effect on ROS production in the mitochondrial inner membrane. Mice fed a vitamin E-deficient diet for 15 weeks showed a fivefold increase in skeletal muscle mitochondrial H2O2 production (Chow et al. 1999). Conversely, rats fed the same diet supplemented with vitamin E showed a dose-dependent decrease in both skeletal muscle and liver mitochondrial H2O2 production. The highest dose (2000 IU/day) yielded mitochondrial ROS production levels about 25% of those suffered by rats on the vitamin E-deficient diet (0 IU). In effect, high doses of vitamin E shift the rodent mitochondrion to behave much like that of a bird with respect to mtROS production (see table 11.4). 11.3.9.2 Carnitine and Lipoic Acid

Another intervention affecting mitochondrial function is feeding lipoic acid and/or acetyl-Lcarnitine to rats (Liu et al. 2002). Lipoic acid is a

Table 11.5 Long-lived (La) Flies Have Lower Mitochondrial Reactive Oxygen Species (mtROS) Production Than Normal (Ra) Flies

Age (days) 15 29 42 58 70

mtROS production (nM H2O2 min•• mg protein) Ra

La

% Difference La relative to Ra

1.9 2.1 3.1 4.1 —

1.5 1.7 2.0 2.5 2.8

–21% –20% –36% –40% —

Source: data from figure 3 of Ross (2000).

413

% Increase in mtROS production as a function of age Ra

La

— 11 48 32 —

— 13 18 25 12

414 Chapter 11 Systemic Theories of Senescence coenzyme required in carbohydrate metabolism. It is reduced in the mitochondrion to dihydrolipoic acid, a potent antioxidant. L-carnitine is required for the transport of long-chain fatty acids into the mitochondrion for beta-oxidation, ATP production, and the removal of excess short- and medium-chain fatty acids. Studies have shown that each of these compounds appears to improve long-term memory and other cognitive functions in aged rodents. Liu et al. (2002) showed that the improvement of memory depends on these compounds altering the functioning of mitochondria in brain structures that play a major role in memory formation, as shown in figure 11.6. The usually high levels of oxidative damage to the hippocampus mtDNA are significantly attenuated by carnitine and lipoic acid, especially when given together. In several brain regions (CA3, CA4, DG), the combined treatment lowered the oxidative damage levels in the brain structures to a level statistically identical to (or even lower than) that of the young control animal. Although the effect was most marked in the hippocampal regions, the treatment had beneficial effects on the entire brain (CX, WM) as well, implying the possible upgrading of other unspecified mental functions. The loss of memory with age may be

caused in part by oxidative damage in the mitochondria. Long-term administration of these compounds resulted in a reduced level of mtDNA oxidative damage and in improved performance on memory tasks. The effects of these compounds are not limited to the brain, for supplementation of these compounds to aged rats significantly improves the glutathione redox system in both skeletal muscle and heart (Kumaran et al. 2004). The components of the glutathione system are usually widely distributed in all tissues and play an important role in scavenging ROS as well as in providing reducing equivalents to other enzymes involved in DNA and protein synthesis. A dietary supplement (Juvenon) containing these compounds is available. 11.3.9.3 Insulin

In rats, caloric restriction (CR) lowers the rate of ROS production, which is susceptible to control by plasma insulin. CR results in a small (~10%) drop in the proton electromotive force, which yields a significant (~2–2.5-fold) decrease of the ROS production rate (Lambert and Merry 2003). This comes about because of a decrease in the substrate oxidation activity (leading to a slower

Immunoreactivity of oxo8G/oxo8dG

Young Old Old + ALCAR Old + LA Old + ALCAR + LA

CA1

CA3

vs. Young vs. Old

CA4

DG

CX

WM

Figure 11.6 Extent of oxidatively damaged DNA as assayed by immunostaining in neurons of specific brain regions of rats of different ages and supplement regimes. CA1, CA3, CA4, and dentate gyrus (DG) are part of the hippocampus; CX indicates cerebral cortex; and WM indicates white matter of the rat brain. Rats were supplemented with a-L-carnitine (ALCAR) and/or lipoic acid (LA) or no treatment (old). Note that rats treated with both supplements had damage levels lower than or comparable to those of young animals. (After Liu et al. 2002.)

11.3 Mitochondrial Damage Theories

proton pumping rate) and an increase in proton leak activity (leading to a faster proton return rate). These two responses cancel each other out, accounting for the fact that this beneficial effect of CR is brought about without affecting the mitochondrial respiration rate (Lambert et al. 2003). Continuous administration of insulin to the CR rats led to an increased substrate oxidation rate and a decreased proton leak activity, increasing the proton electromotive force and the ROS production rate, but without affecting the respiration rate. The internal control properties of the mitochondria appear to be very stable and not directly altered even by toxic substances. Insulin did result in a significant shift in the control levels. Plasma insulin levels have significant effects on individual cells (de Cabo et al. 2003) and on gene activity patterns via regulation of the insulinlike signaling pathway (chapter 7). The direct effect of insulin on the mitochondria seems to be limited to this modulation of ROS production. The conclusion of Lambert et al. (2003) is consistent with other studies on rat mitochondria (e.g., Lopez-Torres et al. 2002). Healthy, lean, older humans have a significantly increased level of insulin resistance than do healthy, lean, younger individuals (Petersen et al. 2003). The elderly also have an approximate 40% lower rate of skeletal muscle mitochondrial oxidative phosphorylation activity, suggesting that their mitochondria are not functioning in a fully effective manner. Gene expression patterns obtained from microarrays of biopsies of healthy humans revealed that several of the genes most highly overexpressed in the older individuals are those induced in response to DNA damage (Welle et al. 2003, 2004). Significantly underexpressed genes include those involved in energy metabolism and mitochondrial protein synthesis. There were a number of other differences noted in the Welle et al. (2003, 2004) study. It is plausible that the increased insulin resistance of skeletal muscle in healthy, elderly humans stems from the lifetime effects of normal (i.e., higher than optimal) levels of plasma insulin on the mitochondria, resulting in an increased level of ROS production and of oxidative damage to the mtDNA and other mitochondrial structures. In

415

this scenario, insulin may be viewed as a negative modulator of mitochondrial function. 11.3.9.4 Heat Shock Proteins

The heat shock proteins (HSPs) protect the cell against a variety of stresses (see chapter 7). When mitochondria of plants or animals are subjected to respiratory stress, they up-regulate the synthesis of their small HSPs (Kuzmin et al. 2004). This response does not induce the synthesis of the more well-studied HSPs, suggesting the conserved existence of a specific respiratory stress response. This may not be a minor aspect of the cell’s response to stress inasmuch as a substantial proportion of the cell’s total number of HSPs, as much as one-third in Drosophila, localize to the mitochondria. Mobilization of some of these mitochondrial HSPs (mtHSPs) has a beneficial effect on longevity. In Drosophila, HSP22 and HSP68 localize to the mitochondria. The selected long-lived O strain flies (Rose 1984) have higher levels of mtHSP22 RNA present as young adults than do their normal-lived B strain progenitors (Kurapati et al. 2000). The targeted up-regulation of HSP68 significantly delays the onset of senescence and extends the longevity of the adult fly in a manner comparable to that obtained with selection, CR, or insulinlike signaling pathway down-regulation (Morrow et al. 2004a). The mechanisms by which the HSPs bring about this longevity extension is not yet known, although it would be reasonable to suspect that it has something to do with reducing the effects of ROS production or damage (see chapter 9). 11.3.9.5 Targeted Gene Delivery

to the Mitochondria Nuclear-encoded proteins can enter the mitochondrion only because they contain signaling sequences that will carry them through the translocases of the inner and/or outer membrane. The existence of such sequences opens the door to using them as potential intervention mechanisms, most of which have not yet been tried but would involve the delivery of proteins or RNAs or DNAs into affected mitochondria as a means of correcting

416 Chapter 11 Systemic Theories of Senescence some age-related defect (de Grey 1999, 2000). The targeted delivery of DNA has been successfully demonstrated in mouse cells in vitro (Fliert et al. 2003), thus setting up a system for testing this idea. The ultimate intervention, reviewed by de Grey (1999), is to replace all 13 of the mitochondrially encoded proteins in the electron transport system with nuclear-encoded homologues. This would ensure that oxidative damage to the mtDNA would not affect the operation of the ETS because a continuous supply of new and undamaged proteins could be imported from the nucleus. This amounts to abolition of the mitochondrial genome and its incorporation into the nuclear genome. There are practical problems that need be solved before this approach can be tried out in laboratory animals and ethical considerations to be faced before it can be considered for human intervention.

11.4 Metabolic Changes Necessary to Support Extended Longevity If normal-lived organisms exist at an optimal mix of energy allocations to reproduction and somatic maintenance (see figure 4.5), then their overall metabolism must be optimized accordingly. If a normal-lived strain is then coaxed to live long, whether by selection or transgenes or CR or some other intervention, then it is reasonable to expect that their metabolism must undergo some sort of corresponding shift. Buck and Arking (2001) examined this question in selected strains and found that there are specific changes in various enzyme activities that seem to be characteristic of different longevity strains. The long-lived strains have a higher flux through the pentose shunt. This generates increased levels of NADPH, which is used for the detoxification of peroxides formed after the dismutation of superoxide by CuZnSOD. The elevated ADS levels in these strains (see table 7.4) require correspondingly elevated NADPH levels if the enzymes are to function effectively. These required cofactors can be produced by the pentose shunt. Classic insect biochemistry studies showed that insects usually have a deficient NADPH reoxidizing system. The activity of the

pentose shunt depends in turn on the rate of reoxidation of NADPH, which in turn depends on the animal having a low level of the NADPH diaphorase enzyme; it cannot be a coincidence that all of long-lived strains significantly downregulate this enzyme. This shift implies that other enzymes of intermediary metabolism should also be altered to accommodate this changed flux through the pentose shunt; such alterations are observed (Buck and Arking 2001; Arking et al. 2002a,b). The metabolic changes associated with extended longevity are reversed when long-lived animals are reverse selected for a normal life span, indicating that the metabolic changes bear a causal relationship to the longevity phenotype. These metabolic changes may be regarded as a prerequisite for the expression of extended longevity by an organism. Animals that cannot bring about these necessary metabolic alterations will probably not fully respond to many longevity-extending interventions. That failure may be used to question the validity of the intervention, as was done with certain experiments in Drosophila (Mockett and Orr 2000) and in mice (Van Remmen and Richardson 2001). But quantitative trait loci (QTL) mapping of Drosophila has revealed the existence of at least 16 loci that regulate the variation within energy metabolism and respiration (Montooth et al. 2003). These loci map away from the actual enzymes involved, suggesting that trans-acting regulatory regions of the genome are important sources of metabolic variation. So-called weak normal-lived strains may well be those in which the metabolic regulatory loci do not permit much variation. Even a supposedly simple intervention such as the insertion of one extra copy of a gene requires the animal to adjust to bring about the existence of a supportive metabolism for the fullest expression of the transgene. A nonrobust response indicts the metabolic flexibility of the normal strain more than it does the intervention. An integrative account of just how these genetic, mitochondrial, and metabolic interventions mesh in shifting the organism from a high ROS/low-defense status to the low ROS/highdefense status characteristic of longevous animals is presented in figure 11.7. Many organisms, including yeast and worm model organisms, have the ability to undergo fac-

11.4 Metabolic Changes Necessary to Support Extended Longevity

417

Restriction of Energy Intake (Yeast, Nematode, Fly, Mammal)

Mutant Genes

Down-regulate ISP

Mitochondrial Membrane Changes insulin

Pentose shunt

NAD(P)+ Rate of Proton Leakage ROS Formation

ADS Gene Expression

Mutant Genes

Increased Metabolic Efficiency Higher ROS Scavenger Levels

Oxidative Damage to Proteins, Lipids, DNA

Cell-cell signaling

Genetic Stability

Maintenance of Proper Cellular State

Maintenance of Optimal Global Gene Expression Pattern

Delayed Onset of Senescence & Extended Longevity Phenotype Figure 11.7 Mutations in certain genes, caloric restriction, down-regulation of the insulinlike signaling pathway (ISP), and/or changes in the mitochondrial membranes give rise to similar outcomes because they alter different aspects of a tightly integrated set of biochemical, metabolic, and genetic processes which alter the cell biology such that the organism expresses an extended longevity and a delayed onset of senescence.

418 Chapter 11 Systemic Theories of Senescence ultative anaerobic metabolism. Mammalian cells likely have little if any capacity to do so. There are now more than 100 genes of C. elegans known to lead to life extension when mutated. This number approaches 1% of the genome and suggests that processes other than those discussed in chapter 7 may be at play. Rea and Johnson (2003) have suggested that some of these longevity genes may work independently of the insulinlike signaling system by exploiting the organism’s ability to obtain energy via a fermentative metabolism. This capacity may have been selected for in an organism that is subjected to periodic anaerobic situations (e.g., rainfall-soaked earth). Some data indicate that the C. elegans mitochondria appear to contain a modified complex II that can operate in reverse as a fumarate reductase (see figure 11.2), which permits fumarate to act as a terminal electron acceptor and form succinate. The succinate can be converted to propanoate and secreted out of the cell. But this capability allows the continued production of flavin adenine dinucleonde (FAD+), which can act as a terminal electron acceptor in the absence of oxygen. The metabolic processes of the cell can keep going, albeit at a reduced rate, as long as a terminal electron acceptor is present. A major effect of this switch to a specialized form of anaerobic metabolism is the reduction in the levels of mitochondrial ROS and the concomittant reduction in the levels of oxidative damage to the mitochondria and the cell. Organisms that possess a facultative anaerobic metabolic shift may exploit that capability as a specialized way to shift into a somatic maintenance mode when ecological circumstances demand it. Parsons (1995, 1996, 2002) suggested that selection for resistance to stress in general and oxidative stress in particular is a main factor regulating longevity. It is plausible that high stress resistance in general is linked to the efficient use of one’s metabolic resources, as documented above for the long-lived mammals or birds or flies or worms. Given the progressively accumulating energy costs during aging, one might predict that animals that are stress resistant are better able to deal effectively with these high energy costs, since their initial efficiency allows the allocation of more energy to somatic maintenance. The stress-

resistant animals should live longer than non– stress-resistant animals. A reevaluation of the energy uses of selected La animals led to the conclusion that their increased efficiency allowed them to maintain both a high fecundity and a long life (Arking et al. 2002b). The discussion of the genetic mechanisms underlying longevity determination in chapter 7 placed some emphasis on the insulinlike signaling pathway and its ability to differentially regulate somatic maintenance and reproduction. We should not be surprised that there is considerable selective pressure to shift metabolic fluxes through various pathways to provide the supporting framework within which resistance to oxidative stress may be efficiently expressed. A comparison of the QTLs responsible for metabolic variation in an aerobic organism such as Drosophila with the QTLs responsible for metabolic variation in a (presumptive) facultative anaerobe such as C. elegans might provide a wide variety of unexpected mechanisms that could lead to adaptive metabolic variations. It has proven possible to construct an in silico version of the yeast cell’s metabolic network, based on the genome scale reconstruction of the underlying biochemical reaction networks (Familli et al. 2003). The model has yielded predictions in good agreement with the known behavior of yeast cells. Future addition of the QTL information might make this model particularly useful for investigating the allowable metabolic variability consistent with enhanced longevity. The data presented in this chapter have shown that optimizing mitochondrial efficiency seems to be a fundamental control point for life span optimization. The mechanisms by which ROS production and oxidative damage are minimized are as varied as the species we have examined. Changes to fatty acids or to the molecular details of complex I or to reduced ADS levels or to the reversibility of complex II or other manipulations yet to be discovered have been brought about by natural selection. Their diversity should encourage us to consider metabolic fluxes other than the canonical ones in the biochemistry texts. We are now at the point whereby some of the metabolic adjustments necessary to increase our own mitochondrial efficiency may be brought about in humans by cultural means (see chapter 15).

12

Senescence as a Breakdown of Intracellular Regulatory Processes

12.1 Basic Assumptions Aging is in many ways a cell-level phenomenon, and it was once thought that aging was autonomous within the cell. Certainly, much of the body’s intercellular communication involves responding to environmental signals by adjusting the insulin and insulin-like growth factor-1 (IGF-1) levels to inform each cell to shift its insulinlike signaling pathway (ISP) from a growth mode to a somatic maintenance mode, or vice versa. Changes in these signals can lead to changes in the cell’s aging (see chapter 7). But aging cannot be fully autonomous within the cell if the cells continually depend on extracellular sources for key signals regulating the expression patterns of their genome. There is clearly an interaction effect. Senescence involves both stochastic and systemic mechanisms that affect both intercellular and intracellular components. Do we age mostly because our individual cells age, or do we age mostly because of extracellular events? Given the interactive nature of the stochastic and systemic senescent mechanisms, we would not be surprised if the data lead us to answer yes to both questions. And so we would also want to understand whether the events seen at the cellular level play a key role in bringing about the senescent traits characteristic of the organism as a whole.

12.2 Aging in Dividing Cells: Cellular Life Spans and Organismic Longevity 12.2.1 Historical Background From the point of view of gerontology, the history of cell culture may be divided into two phases. The first phase began in 1910 with the pioneering work of Ross Harrison, who was the first to demonstrate that vertebrate tissue can be kept alive outside the body and that the cells will continue growing and differentiating when bathed with body fluids in vitro. An early idea was that aging and death are the price of cellular differentiation. In other words, the organism is mortal, but if its component cells could be freed from the constraints of the body and from the need to differentiate into specialized parts, then they might be able to live forever. Carrel (1912) and Ebeling (1913), using Harrison’s techniques, reported that they were able to keep fibroblast cells obtained from chicken embryonic heart tissue alive and in a state of continuous proliferation for at least 34 years by culturing them in laboratory glassware on plasma clots for support and nutrition. Since the maximum life span of the chicken is about 30 years, this seemed to be a good operational demonstration of cellular immortality. Further support for the concept of cellular immortality came from the discovery of several different cell lines, given

419

420 Chapter 12 Senescence as a Breakdown of Intracellular Regulatory Processes names such as L cells (Earle 1943) or HeLa cells (Gey et al. 1952), that could be grown continuously in culture without showing any decline in proliferative vigor. Only later was it realized that these cell lines are derived from transformed (i.e., cancerous) cells, and thus the lessons drawn from them are not directly applicable to theories of normal cellular aging. The second phase of cytogerontology may be conveniently dated from the classic paper of Hayflick and Moorhead (1961) showing that cultured human fibroblast cells divide a finite number of times and then stop dividing. This key observation led to new studies and to new interpretations of older studies. For example, Hayflick (1982) pointed out that not only had no one confirmed Carrel’s and Ebeling’s original observation of immortal cells, but that the procedures used by Carrel for the preparation of his chick embryo extract probably permitted the contribution of new and viable cells to the chick heart strain at each feeding. This suspicion was borne out by the fact that when Carrel’s experiments were repeated using more modern and more stringent procedures, it was not possible to keep the chick cultures alive for more than 44 months (Gey et al. 1974). It now seems likely that Carrel’s culture was not composed of 34-year-old cells and their descendants but was probably instead a 34-year continuous culture in which fresh embryonic cells were added every week or so. These new insights, encouraged by the Hayflick and Moorhead report, led to the concept that the visible and functional manifestations of aging in organisms have an intracellular basis. This is the concept that I explore in this chapter in some depth. An excellent description of these original experiments and their current consequences is found in Hayflick (2003).

12.2.2 The Cell Cycle

Our cells reproduce by mitosis, and the presence (or absence) of this event is what organizes the manner in which we view the cell. The cell cycle is defined as the interval between the completion of mitosis in the parent cell and the onset of mi-

tosis in one or both daughter cells (figure 12.1). The average length of the cell cycle and of its component phases varies widely from cell type to cell type and can be altered by a wide array of various factors (Baserga 1985). There are several types of nondividing cells. Cultured cells can be kept quiescent and viable in the G0 state for long periods of time if factors required for cell division are simply removed from the medium. In vivo, adult liver cells are metabolically very active but mitotically almost totally inactive. Yet, if part of the liver is surgically removed, these quiescent cells will leave the G0 state and enter S phase after a lag of about only 18 hours or so. The cells are thus capable of reproducing if they receive the appropriate signal. The terminally differentiated cells are quite different from all other cells. They have ceased dividing and cannot be recalled into the cell cycle.

S G2

M

G1

Terminal differentiation

G0

Figure 12.1 A diagrammatic representation of the cell cycle. After mitosis (M), the cell enters the first gap period (G1), during which it prepares for genome replication. This replication takes place during the synthesis (S) period. The cell then enters a second gap period (G2) preparatory to mitosis. Cells can leave the cycle via two mechanisms. Cells that are temporarily arrested in the their growth (e.g., by removal of exogenous growth factors from the medium) are called quiescent cells, because they are capable of reentering the cycle if provided with the proper conditions. Such cells are considered to be in the G0 state. Terminally differentiated cells cannot be recalled into the cell cycle and will eventually die.

12.2 Aging in Dividing Cells: Cellular Life Spans and Organismic Longevity

Many of our tissues are composed of such terminally differentiated cells. Their life span is variable, ranging from a few days to 2 weeks for keratinocytes and intestinal epithelial cells, to 3 months for red blood cells, to a lifetime for most neurons or skeletal muscle cells. Although all normal cells have a finite life span, they also have the capability, under extraordinary circumstances, of surviving substantially longer than normal. For example, transplantation experiments have shown that the cells of mouse skin (Krohn 1966) and of mouse mammary tissue (Daniel 1977) can survive in vivo for periods far exceeding the maximum life span of the original donor. The mouse mammary tissue has been kept alive longer than 6 years when transplanted to new hosts at yearly intervals. The tissue has a much shorter life span, about 2 years, when it is transplanted at quarterly intervals. In both cases, the growth tends to slow down as the number of transplant generations increases. Thus the number of transplant generations is a better indication of cellular aging in vivo than is the mere passage of time, and there appears to be a 15% loss of proliferative ability per transplant generation regardless of the time interval (Daniel 1977). This loss of growth potential does not seem to be closely tied to a corresponding loss of the cell’s physiological competence, for even very old and nonproliferating mammary gland cells will produce abundant milk if maintained in lactating hosts. Thus, although the loss of proliferative ability is one useful sign of cellular aging, it is probably not correct to equate it with a deteriorative event. The cell seems to be constructed such that a common precondition to its becoming functionally and structurally specialized is that it must first leave the cell cycle (see figure 12.1; also seee Walsh and Perlman 1997). Differentiated tissues characterized by continuous cell replacement often rely on stem cells as the progenitor of the new cells. I explore this topic in greater detail in the next section. The only way a cell can stay in the cell cycle indefinitely is to be transformed. Human fibroblast cells are quite resistant (but not totally so) to both spontaneous and chemical or virus-

421

induced transformation; rodent cells can be spontaneously or chemically transformed with relative ease. (This differential resistance to transformation may be regarded as the manifestation of the human’s better mechanisms affecting genome stability and longevity determination.) Regardless of their origin, transformed cells exhibit loss or inactivation of the control mechanisms that regulate cell movement and cell division. Such cells show no loss of their proliferative ability with each successive passage through the cell cycle. Of course, they usually have also lost the ability to form a normal differentiated cell type. Many transformed cells have the ability to induce tumors in appropriate host animals. Thus, the acquisition of an indefinite growth potential is the first of several steps necessary to transform a normal cell into a neoplastic cell. When normal cells are assayed in situ in normal host animals under normal conditions, many cell types exhibit changes analogous to those seen in vitro. In general, these changes include (1) the length of the cell cycle increases with age, particularly in the G1 and S phases, (2) the proportion of cells within a tissue that are able to divide decreases with age, (3) there is much more heterogeneity in the transit time among cell populations taken from old animals than from young animals, and (4) the cells get larger as they age (Lesher et al. 1961). It is reasonable to wonder if the loss of function associated with senescence and observed in most tissues have anything to do with the decreased proliferative ability of its component cells, as observed both in vivo and in vitro. This question has been addressed in several reviews (Campisi 1996; Cristofalo and Pignolo 1995) and we will discuss it below.

12.2.3 Cellular Senescence in Vitro The typical history of a cell culture is shown in figure 12.2. After the initial explantation of the cells from their donor to an in vitro environment, there is a variable period of time (Phase I) during which the cells grow and divide rather slowly. One convenient but potentially inaccurate method of measuring the rate of cell division is

422 Chapter 12 Senescence as a Breakdown of Intracellular Regulatory Processes but this common measurement of cell senescence may be obscuring that information. In Phase I, the first several PDs may take several weeks. This phase probably reflects the time needed for the cells to adapt to the somewhat different environment of test tube and culture flask. Phase I is followed by a period (Phase II) characterized by a very rapid rate of cell proliferation, which maintains high numbers of cells in the culture. The PD time is relatively constant, usually measured in days. The cultures are usually quite healthy. This plateau period is followed by a time of declining proliferative capacity (Phase III), in which the amount of time required for the population to double appears to increase exponentially. Presumably, this phase is established as

to determine the amount of time it takes for the cell population to double in number. If the petri dish is initially half-filled with cells, then one population doubling (PD) occurs when the petri dish is completely filled with cells. However, this metric measures the behavior of the whole population and not of the individual cells. For example, the population doubles when every cell present in the beginning of the period divides once, or when half of the cells divide twice consecutively, and so on. Thus any given PD time could be the outcome of the differential divisions of some subset of the population. And this less than satisfactory situation is likely aggravated by the fact that cells increase in size as they age. We should keep in mind that our main question in this chapter has to do with the behavior of individual cells,

Subcultivations 10

30

20

Growth rate of culture

0

50

40

Transformed cell line

Phase II (exponent) Phase I (primary cell culture)

0

1

2

3

4

Phase III

5

6

7

8

9

10

11

12

Time (months) Figure 12.2 The behavior typical of cells in culture. During phase I (the primary culture), the cells initially grow slowly, but the growth rate soon increases. This phase ends with the formation of the first confluent sheet. Phase II is characterized by rapid and luxuriant growth. Normal cells in this phase are called cell strains. The genome can be altered at any time during this phase such that the cell will be transformed and converted into a cell line that has a potentially infinite life span. Normal cells enter phase III after a characteristic number of population doublings, their growth rate decreases, and the strains can no longer be maintained by normal culturing techniques.

12.2 Aging in Dividing Cells: Cellular Life Spans and Organismic Longevity

more and more cells become committed to terminal differentiation and slip out of the cell cycle. When the proportion of proliferating cells in the population declines to the point at which less than one PD occurs per culture period, then the culture was thought to be senescent. From this point on, there will be relatively fewer dividing cells than at the same point in the previous generation, so the population inevitably will show a decrease in growth rate. Many experiments end at this point, but I discuss below what happens to these nondividing cells. Certain cell cultures are immortal; they never enter Phase III but appear to be permanently locked into Phase II (see figure 12.2). These transformed cells will continue to proliferate as long as the appropriate nutrients and other factors are supplied. In addition, many such immortal cell lines are composed of cells that have an abnormal chromosome constitution. Such mutants or variants probably have suffered some damage to

423

the genetic mechanisms that regulate cell division, and they may be regarded as precancerous, as described earlier. That there might be a direct connection between cellular and organismic aging was first suggested by Hayflick’s report in 1965 that fibroblasts derived from human embryos underwent about 48 PDs in vitro (with a range of 35–63), whereas fibroblasts derived from adults underwent only about 20 PDs (with a range of 14–29). Note that the ranges for the PD of fetal and adult tissues do not overlap, despite the obvious large variance. In addition, there is no tight correlation between donor age and PD number: The cells from an 87-year-old yielded 29 PDs, while those from a 26-year-old yielded 20 PDs. A more extensive cross-sectional analysis showed a decrease in the mean number of PDs with increasing age (figure 12.3). On closer inspection of these data, however, this decrease seems to be driven almost entirely by the apparently real difference in PDs

70 Normal subjects Werner's syndrome subjects

Number of cell doublings

60

50

40

30

20

10

Fetal

0–10

10–20 20–30 30–40 40–50 50–60 60–70 70–80 80–90 Age (years)

Figure 12.3 The cumulative number of cell doublings achieved by human skin fibroblast cultures plotted as a function of the age of the donor. The calculated linear regression line (black line) for the control group is drawn between the first and ninth decades and has a regression coefficient of 0.20 ± 0.005 (SD) cell doublings/year, with a correlation coefficient of –0.50. The gray line is the lower 95% confidence limit for the regression line. Note that cells from people with Werner’s syndrome undergo significantly fewer doublings than do age-matched controls. (After G. M. Martin et al. 1970.)

424 Chapter 12 Senescence as a Breakdown of Intracellular Regulatory Processes between the very young (individuals 10 years old and younger) and the very old (individuals 70 years and older). There is no significant correlation between donor age and PDs for individuals between the ages of 11 and 69 years, when the age-specific mortality rate (qx) increases by 100-fold (see figure 2.14). If there is an aging clock, it seems not to keep very good time. A reexamination of this phenomenon revealed a large variability but no statistically significant age association with replicative life span for fibroblasts taken from adult humans (Cristofalo et al., 1998; see figure 12.4). The two sets of data yield two different interpretations. The data of figure 12.3 would lead one to hypothesize a significant correlation between organismal senescence (as encoded by donor age) and replicative senescence (as encoded by the number of PDs). The inability of a cell to divide would seem to be an important part of its senescence. The data of figure 12.4 would lead to the hypothesis that there is no correlation at all between donor age and replicative senescence and that replication may not be an important part

of the senescence of cells either in vitro or in vivo. The disagreement still exists today, based in part on technical considerations and in part on conceptual differences (see Cristofalo et al. 2003; Rubin 2002; and J. R. Smith 2003). The real question is not whether the senescence of a cell is tightly correlated with its mitotic history, which is perhaps the more popular conception of the problem. Neither is the real question whether the pathway to cell senescence in vitro is identical to the one in vivo, for it has been pointed out that fetal cells kept in culture do not eventually acquire the characteristics of adult-derived cells kept in culture (Cristofalo et al. 1998; Wharton 1988). The important scientific question is whether the processes involved in the replicative senescence of the cell in vitro are closely related to the processes responsible for senescence of cells in situ. This chapter presents evidence that leads us to an affirmative but nuanced answer to this question. It was once thought that limitations on the cellular life span, however regulated, are intrin-

70 r = -0.018 p = 0.08479 n = 116

Maximum PDL

60 50 40 30 20 10 0 0

-10

0

10

20

30

40

50

60

70

80

90

100

Donor Age Figure 12.4 A reevaluation of the relationship between donor age in years and the proliferative capacity in vitro of human skin fibroblast cells obtained from people enrolled in the Baltimore Longitudinal Study on Aging. Males are represented by diamonds and females by circles (N = 116). The regression line has r = –0.18, P = .85. Note that there is no obvious decline in population doubling values with age, although there seems to be substantial individual variation. (After Cristofalo et al. 1998.)

12.2 Aging in Dividing Cells: Cellular Life Spans and Organismic Longevity

sic to the cells and do not appear to reflect other variables, such as the amount of time spent in culture, the nutritional state of the medium, and so forth. This autonomy is essential to any cellbased theory of aging. This assumption was challenged by the findings of Balin et al. (2002) that the cell population emerging in vitro from a biopsy is not representative of the cell population in situ and that the various culture methods employed by different labs can significantly affect the replicative life span determinations. More specifically, some types of cell senescence occur in response to various external stressors such as oxidative stress (Ben-Porath and Weinberg 2004). The number of PDs observed in culture are not intrinsic to the cell but are related to external variables. For example, the number of PDs until replicative senescence is characteristic of each cell strain, yet it may be significantly increased or decreased simply by altering the oxygen concentration in the culture chamber (Balin et al. 2002; Parrinnello et al. 2003). Thus, the underlying assumption that replicative senescence occurs only as an aging response to the number of cell divisions is not correct. The current thinking is that cell replicative senescence occurs both as a response to exogenous stress and as a response to cell division, and I discuss this in more detail later in the chapter. Perhaps the senescent phenotype in vitro is merely the final common pathway for cell deterioration from many causes. Although there is a clear difference in the proliferative ability of cells derived from very old or very young donors, there does not appear to be a clear difference in the proliferative ability of cells derived from individuals of intermediate ages. This does not mean there is no relationship between cellular aging and organismic longevity; on the contrary, it means simply that if such a relationship exists, it is neither linear nor very tight. Hayflick and Moorhead (1961) originally interpreted their findings as indicating that the limited proliferative ability of somatic cells is programmed and can be viewed as a repeatable cellular expression of senescence. This interpretation is commonly known as the “Hayflick hypothesis” or “Hayflick limit.” We now know that this limit is not completely intrinsic. Despite the

425

temptation to do so, this hypothesis does not require us to extrapolate beyond the data and suggest that senescence of the organism is caused by the loss of its cells’ proliferative ability. Whatever the nature of the intrinsic events that bring about the normal aging of the individual, they probably have only an indirect relationship to the events that take place at the cellular level. If long-lived somatic cells undergo changes that alter both their ability to undergo mitosis and their ability to consistently maintain their normal level of function in situ, then one could imagine how such a loose linkage could be empirically observed. Data similar to the results reported by Hayflick and Moorhead (1961), all showing that cells obtained from adult donors are reproducibly different from cells obtained from very young and/or embryonic donors, have been reported by several investigators using a variety of human tissues, including T-lymphocytes, glial cells, keratinocytes, vascular smooth-muscle cells, lens cells, and endothelial cells (for references see Cristofalo and Pignolo 1995). The term “Hayflick limit” refers to the repeatable number of population doublings that different cell types can typically achieve under standard culture conditions. Its widespread use reflects the general acceptance of the original observations and, to a lesser extent, of the Hayflick hypothesis. That the replicative life spans are genetically determined was suggested by a study in which skin fibroblasts from three pairs of monozygotic twins showed no significant difference within each twin pair, but did show differences between pairs (Ryan et al. 1981). In addition, the observation that fibroblasts obtained from patients suffering from Werner’s syndrome (see chapter 8) display a shortened replicative life span and a variable but poor growth capacity also suggests the existence of a genetically modulated relationship between organismic aging and cell proliferation ability. Perhaps the most interesting data come from interspecific comparisons of PD number and species-specific maximum life span (table 12.1). These data are open to all the criticisms voiced by Promislow (1993); nonetheless, they illustrate some possible connections. The two variables, PD

426 Chapter 12 Senescence as a Breakdown of Intracellular Regulatory Processes Table 12.1 Relationship between Species-specific Cell Doublings and Species-specific Maximum Life spans Range of population doublings for cultured normal embryo fibroblasts

Species Mouse Mink Chicken Cow Man Tortoise

Young donor Old donor

14–28 30–34 15–35 40–60 40–60 112–130 90–102

Mean maximum life span (years) 3.5 10 30 30 115 175

Source: from data summarized in Hayflick (1977).

number and maximum life span, have an imperfect but clear relationship: Compared to longlived species, short-lived species tend to lose their replicative ability after fewer PDs. One assumption implicit in the Hayflick hypothesis is that a cell strain with a high number of PDs is composed of aging cells; that is, the Phase III cells are mostly senescent, abnormal, or dying cells. This is a crucial assumption of a cellbased theory and one that should be carefully examined. Two main points contradict this assumption. First, any two individual cells, including sister cells, isolated from the same generation of the same cell strain can show a large variation in their ability to undergo future PDs, the number ranging from 0 to 33 (J. R. Smith and Whitney 1980). This variability suggests that in normal cultures a large fraction of the cells in every generation are characterized by a low proliferation potential. Such cells will be present in the population in lower numbers than rapidly proliferating cells; thus they have a lower probability of surviving through the change of culture dishes associated with the beginning of each new cell generation. Nondividing cells appear to be recruited anew in each cell generation, possibly in a stochastic manner, as earlier suggested by Holliday and colleagues (1977). As a result, the cells that make up a cell strain are not in lockstep with each other, as is implied by the early data, and thus there can be no rigid or precise intracellular counting mechanism used by all the cells to determine their age synchronously. The ab-

sence of such a mechanism accounts for the failure to obtain a significant correlation between PDs and donor age from 10 to 69 in the data of figure 12.3. The second point is that all of the data presented here that were used to support the concept of autonomous cell aging can be equally used to support the opposing idea: that the departure of cells from the mitotic cycle may be a sign not of aging but of differentiation (Bell et al. 1978) or of stress (Ben-Porath and Weinberg 2004). Cell growth and cell differentiation have long been considered mutually exclusive or alternative possibilities for cells. In fact, differentiation leads in many cases to cells that cannot divide, such as nerve cells or enucleated red blood corpuscles. Bayreuther and colleagues (1988, 1992a,b; Bayreuther and Gogol 1993) showed that the nondividing fibroblasts are not moribund but will, if allowed to live in vitro for a long period of time, eventually differentiate into various sorts of normal-looking connective-tissue cells. It is instructive to consider these experiments in some detail.

12.2.4 Differentiation and Senescence in Vitro Human skin fibroblasts were cultured and underwent about 53 PDs in about 300 days, at which time they had exhausted their growth potential. These results are consistent with the data obtained from many laboratories since the 1970s.

12.2 Aging in Dividing Cells: Cellular Life Spans and Organismic Longevity

Using appropriate culture techniques, however, these nonmitotic cells were kept alive for about another 300 days, during which time the cells completed their differentiation. Bayreuther et al. (1992a,b) reported that these differentiated cells eventually either died via apoptosis (which I discuss later in the chapter) or were transformed. During this long period of time in culture, the cells do not stay the same but progress through a seven-stage sequence of cell differentiation (table 12.2). The various stages are distinguished from one another by whether they are mitotic or postmitotic and by their morphology and their biochemical activities, including different patterns of gene activity. By these criteria, cell types I–III are considered to be mitotic cells that differ in their shape and their mitotic potential, while cell types IV–VI are postmitotic and differ in their size. Cell type VII represents the final stage of cell differentiation and comprises the cells that are degenerating via apoptosis or have been transformed and are now immortalized. Note how the dominant cell type shifts over time, from a preponderance of type II in the PD 17 culture to a preponderance of type VI in the PM (post-mitotic; see table 12.2) 40 culture. The cells in the culture at the end of the year are very different from the cells that were used to start the culture.

These mitotic and morphological changes are accompanied by striking quantitative changes in the cell’s synthetic abilities (see table 12.2). The PM 40 cells’ ability to synthesize DNA has disappeared, but their ability to synthesize other macromolecules has increased greatly. In addition to these quantitative changes, each cell type is qualitatively distinguished by its own particular polypeptide pattern, indicating the synthesis of cell-type–specific proteins. The quantitative and qualitative measures both suggest that reproducible changes in gene expression patterns are taking place. Certain proto-oncogenes (e.g., c-jun) are down-regulated in the postmitotic cells, leading to speculation as to their role in inhibiting DNA synthesis and blocking the cells from entering mitosis (Brenneisen et al. 1994). Finally, cells identified as being in particular stages of differentiation are found in vivo at particular and different locations in the skin; the mitotic types I–III are stratified in the reticular layer; the postmitotic types IV and V are localized in the papillary layer, and type VI is localized under the basal lamina (Bayreuther et al. 1992b). Thus, the starting population of the fibroblasts usually used in these cultures is in a predifferentiated state and consists of different types of cells destined to form heterogeneous tissues within one particular organ.

Table 12.2 Progressive Differentiation of HH8 Fibroblasts and Alterations in Their Mitotic and Synthetic Capabilities Macromolecular synthetic activity

Distribution of cell types (%) Mitotic?

Culture

Yes Yes Yes No No No No No No

PD17 PD30 PD55 PM 1 week PM 3 week PM 7 week PM 11 week PM 24 week PM 40 week

427

I

II

III

IV

V

VI

VII

RNA

Protein

Collagen

18 3 0 0 0 0 0 0 0

70 78 1 0 0 0 0 0 0

10 16 79 30 3 0 0 0 0

1 2 14 18 31 16 4 0 0

0.5 0.5 3 40 43 32 5 2 0

0.5 0.5 3 12 23 51 87 89 88

0 0 0 0 0 1 4 9 12

1.00 1.33 3.36 6.51 9.93 11.5 9.5 10.8 11.6

1.00 5.6 7.9 9.5 11.2 12.6 12.4 12.9 11.8

1.00 1.4 4.3 5.7 6.2 6.3 9.6 9.7 10.2

Source: data from table 1 of Bayreuther et al. (1992). Note: PD, population doubling generation number; PM, postmitotic weeks in culture; cell types I–VII are described in the text.

428 Chapter 12 Senescence as a Breakdown of Intracellular Regulatory Processes 1993) suggest that what is encoded in the cell’s genome is not an aging clock per se, but rather a mechanism that enables healthy cells to slip out of an active mitotic mode into a terminal predifferentiation mode in which cell division is repressed. This phenotype is different from that of a transformed cell line that undergoes continued proliferation. Transformed cells have somehow disabled their mitotic repression system, and investigations into those cells may allow the identification and characterization of that system. The existence in vitro of replicatively senescent cells suggests the existence of a mechanism that enables nonhealthy cells to stop dividing. The replicative senescence may be either reversible or irreversible (Beausejour et al. 2003). Thus, there are four different cell replicative senescent phenotypes, and this finding has clarified the otherwise confusing data. Each of these four different cellular phenotypes are probably a result of different types of alterations in a complex of separate but linked cell signaling pathways, each of which respond to some particular category of stressor(s). Before inquiring about the nature of these pathways, I first discuss the role of telomeres in replicative senescence.

In addition to this spatial heterogeneity, the distribution of these six cell types in humans depends on age (table 12.3). Note that, as might be predicted from the data in figure 12.3, the younger the individual, the greater the proportion of mitotic fibroblast cell types in primary cultures of the skin. This description of progressive cell differentiation suggests that the growth curve of figure 12.2 is too simple because it reflects only the changes in growth rate and not the changes in the cells. Presumably, cell types I–III would occupy the Phase I and part of the Phase II portions of that curve, while cell types IV–VI would occupy the rest of the Phase II portion. Cell type VII would probably occupy the Phase III section of the curve.

12.2.5 Four Cell Replicative Phenotypes Late-passage, nondividing fibroblasts appear to be able to survive a long time in culture and show signs of differentiation in vitro. There is still some controversy over the complexities of the differentiated cells observed in culture by different laboratories and controversy over the eventual fate of these cells. Not every lab has duplicated Bayreuther’s (1992a,b; 1993) observations in detail. Nonetheless, there appears to be agreement on the two main points: that the postreplicative cells are stable and can survive for a long time, and that they resemble terminally differentiated cells (Campisi 1996). Bayreuther’s results (1992a,b;

12.2.6 Telomeres and Cellular Senescence Regardless of the nature of the relationship between organismic aging and the Hayflick limit, the question remains of how the cell counts off

Table 12.3 Age-dependent Distribution of Fibroblast Cell Types in the Skin of Human Donors Fibroblast cell type (%) Age (years) 10 30 50 70 90

MFI

MFII

MFIII

Sum MF

PMFIV

PMFV

PMFVI

SUM PMF

15 7 0 0 0

59 54 38 6 2

13 12 30 39 12

87 73 68 45 14

3 11 6 14 21

1 3 7 5 8

9 13 19 36 57

13 27 32 55 86

Source: data from table 1 of Bayreuther et al. (1992). Note: MF, mitotic fibroblast stage I, II, III; PMF, post-mitotic fibroblast stage IV, V, VI.

12.2 Aging in Dividing Cells: Cellular Life Spans and Organismic Longevity

each cell division. How does it remember the number of its past mitotic events, and how does this memory eventually impede the cell from transiting the cell cycle again? Barbara McClintock (1941) and H. J. Muller (1938) were the first to recognize that chromosome ends are different from the rest of the chromosome. Once the structure of DNA was deduced and the enzymatic basis of its replication worked out, it became apparent that the replication mechanism was incomplete (Olovnikov 1971, 1996). Because of the synthetic characteristics of conventional DNA polymerases, which can replicate DNA only in the 5' to 3' direction and cannot initiate synthesis of a DNA chain without the assistance of a primer template, the ends of the chromosome cannot be replicated completely. One DNA strand of a linear chromosome will be replicated to the very end, but the other strand will have a short 8- to 12-base gap at the 5' end. In principle, each chromosome in a cell that divides repeatedly will shorten progressively from both ends until an essential sequence becomes eliminated or inactivated (Olovnikov 1973). The loss of these presumably vital gene functions was assumed to initiate the aging process. The scenario is logical. The problem is determining whether this loss actually occurs in an aging cell and whether it has anything to do with the aging of the organism. Telomeres are the structures that cap the ends of chromosomes. They consist of short terminal DNA repeats (figure 12.5a). They serve the triple functions of protecting the chromosome against damage, maintaining the normal length of each chromosome, and possibly maintaining nuclear organization via their association with the nuclear membrane and other cell structures.Telomeric DNA sequences of a variety of species show surprising patterns of similarities and differences among species. For example, TTAGGG/CCCTAA is the sequence of all telomeric DNA in all vertebrates, as well as some protozoans and fungi, but other fungi and other invertebrates have very different sequences, and Drosophila has no such sequence and uses an entirely different telomeric mechanism (Mason and Biessmann 1995; Zakian 1995). The telomeric DNA sequences are replicated not by the normal DNA polymerase alone,

429

but by DNA polymerase along with an unusual enzyme called telomerase, which contains short RNA sequences complementary to the DNA repeats. As long as telomerase is active, the short underreplicated gaps in the DNA are filled in by hexameric repeats of 5'-TTAGGG-3' DNA coded for by the RNA portion of the enzyme, so the chromosome will not shorten. But if the enzyme is inactivated, each chromosome will lose some telomeric DNA at every round of replication (figure 12.5c,d). The telomere hypothesis of cell aging postulated that potentially immortal cells (e.g., germ cells and cancer cells) maintain their telomerase activity and can divide indefinitely because their chromosomes will not shorten (figure 12.5b,e). Cells with a limited replicative life span, on the other hand, should have no telomerase activity, so the progressive shortening of the telomeres during cell division may serve as a mitotic clock for replicative senescence. It was hoped that the hypothesis would explain the mechanism underlying the Hayflick limit. Harley et al. (1990) demonstrated that telomeres progressively shorten during replicative growth of human cells in culture, the mean telomere length decreasing by about 50 base pairs per doubling. Replicative capacity appeared to be proportional to mean telomere length (Allsopp and Harley 1995). When human fibroblasts are subjected to high oxygen partial pressure, the resulting oxidative stress irreversibly blocks proliferation. Under these conditions, the rate of telomere shortening increases fivefold, and the cells cease proliferating when the telomere length reaches about 4 kilobases (von Zglinicki et al. 1995). This last observation implies that a critical telomere length provides the signal for cell cycle exit in replicative senescence. It also shows that the rate of telomere shortening is critically dependent on environmental factors. Differentiation of cells in culture is normally accompanied by loss of telomerase activity and shortening of the telomeres (Kruk et al. 1996; Sharma et al. 1995). Finally, Peirera-Smith and Ning (1992) showed that hybrids between immortal cells (with telomerase activity) and normal cells (without such activity) have a limited life span (and

430 Chapter 12 Senescence as a Breakdown of Intracellular Regulatory Processes

(b) Germ (reproductive) cell (a) DNA prior to cell division Telomerase + DNA polymerase

Telomeres fully replicated (d) Cell no longer divides

Telomeres (TTAGGG)n (c) DNA after divisions Many divisions

DNA polymerase alone Telomeres shorten

Telomeric ends lost

(e) Cancer cell (continues to divide)

Telomerase reactivated Telomeres fully replicated

Figure 12.5 Outline of the status of the telomeric ends of chromosomes during cell division in normal and cancerous cells. (a) The full telomeric DNA sequence in a chromosome prior to cell division. (b) The chromosomes of germ cells that will reproduce to create the next generation contain both DNA polymerase and telomerase. In these cells, replicated chromosomes maintain their full telomeric length. (c) In the presence of DNA polymerase but the absence of telomerase, the telomeric ends shorten with each replication. (d) Once a cell “runs out” of telomeres, it no longer enters mitosis and cell division stops. (e) Should the telomerase become active again, the cell can continue to divide indefinitely, as cancer cells do.

lack telomerase). When an immortal cell was artificially stimulated to produce longer telomeres than normal and then fused with a normal cell, the life span of the resulting hybrid was longer than expected (Wright et al. 1996), suggesting that proliferative capacity is determined by telomere length. In vitro, the mean telomere length decreases by about 15 base pairs per year in cells of primary cultures obtained from donors of different ages (Harley et al. 1990). In immortal tumor cells, however, stabilization of the telomeres was correlated with the appearance of telomerase activity. However, not all immortal cells express telomerase (Bryan et al. 1995), nor is telomerase

activity found only in immortal cells (Broccoli et al. 1995). Many somatic cells (e.g., skin, hair, intestinal, and blood cells) normally proliferate extensively, and these cells also express high levels of telomerase, if only during their periods of rapid division (Ramirez et al. 1997). High telomerase activity is also found in germline cells, which also show no shortening of the chromosome ends. Together these data provide the conceptual basis of the telomere hypothesis of aging and immortality. According to this hypothesis, germline cells maintain the length of their chromosomes by maintaining their telomerase activity (figure 12.5b). Sometime during embryogenesis

12.2 Aging in Dividing Cells: Cellular Life Spans and Organismic Longevity

431

hTRT– control cells displayed the telomere shortening characteristic of ordinary cell cultures, as described above. This observation demonstrates that the transfected telomerase is functionally active at the chromosome level. The most interesting observations come from the comparisons of cellular life spans in these genetically engineered cells: The non–telomerase-expressing hTRT– cells stopped dividing at a number of cell population doublings characteristic of the cell type, whereas the telomerase-expressing hTRT+ cells continued to divide for at least 40 doublings beyond the mean values of controls (figure 12.6). The results of this gain-of-function experiment show that expression of exogenous telomerase can significantly increase the replicative life span of at least these two human cell types by maintaining the youthful length of telomeres. Some organisms (e.g., mammals such as rabbits and hares (Forsyth et al. 2005) as well as invertebrates such as Drosophila, Podospora, Saccharomyces) do not exhibit telomere shortening, even though they age. Conversely, fibroblast

(after 16–20 weeks in humans), telomerase becomes repressed, and, in somatic cells, the telomeres shorten until the cells reach a critical point (the Hayflick limit), where they stop dividing for reasons already explained (figure 12.5c,d). Transformation enables cells to bypass the Hayflick limit by reactivating their telomerase and maintaining chromosomal integrity (figure 12.5e). There may also be a telomerase-independent mechanism that enables some immortal cells to maintain their telomere length. Persuasive data showing that telomere shortening is in fact related to the onset of replicative senescence was obtained by transforming either normal diploid human retinal epithelium cells or foreskin fibroblast cells with vectors containing the human telomerase reverse transcriptase (hTRT) subunit (Bodnar et al. 1998; figure 12.6). Some of the clones Bodnar et al. obtained expressed significantly elevated levels of telomerase activity, which was due to the transfected hTRT cDNA and not to their endogenous gene. These hTRT+ cells showed increased telomere size, but

Retinal pigment cell clones Mean PD value of hTRT+ clones

Mean PD value of controls

40

50

60

70

80

90

100

Fibroblast cell clones

Mean PD value of controls

50

60

70

80 90 Population doublings

Mean PD value of hTRT+ clones

100

110

Figure 12.6 The effect of telomerase expression on the life spans of two cell types. In both cases, clones with hTRT- inserts (controls) are represented by circles and clones with hTRT+ inserts (exogenous telomerase expression) are indicated by triangles. Cell life spans are represented as population doubling (PD) values for each clone. In both cell types, the clones expressing exogenous telomerase have mean PD values significantly higher (at least 20–40 PDs) than the control clones. Note that almost all the control clones are senescent (filled symbols) or nearing senescence (half-filled symbols), while there are no senescent hTRT+ clones (open symbols). (After Bodnar et al. 1998.)

432 Chapter 12 Senescence as a Breakdown of Intracellular Regulatory Processes clones from patients with Hutchinson-Gilford progeria (see chapter 8) can senesce despite the presence of active telomerase (Wallis et al. 2004). These exceptions demonstrate that the telomereshortening hypothesis is not universal, at least in detail. In addition, the hypothesis says nothing about the aging of nondividing cells such as neurons and muscle; thus, it cannot explain the aging processes in all the cells of any single organism. Adult mice seem to have a tissue-specific regulation of telomere length, not an age-dependent regulation (Prowse and Greider 1995)—an observation that complicates the extrapolation of the telomere hypothesis from the cell level to the organism level. A reasonably conservative interpretation of the data is that some type of mechanistic connection exists between telomere length and proliferative ability for the cells of many but not all organisms, but identifying the role of telomeres does not make the connection between cell proliferation and organismal aging any clearer. What does seem to be clear is the strong separation between normal and transformed, or immortalized, cells. Telomeres are lengthened or stabilized in almost every type of immortal or cancer cell line examined, as a result of either the reactivation of telomerase or another novel mechanism (Bryan et al. 1995). This observation has opened up a fruitful line of inquiry for cancer therapies and aging interventions based on genetic engineering methods (Fossel 1998; Kim et al. 1994; Shay 2005). One finding that cast some doubt on the telomere as a highly accurate “replicometer” (Hayflick, 2003) is that the molecular biology of DNA replication suggests that the telomere should lose at least 8–12 nucleotides at each PD, but fibroblasts under standard cell culture conditions lose up to 200 nucleotides per PD. The telomere loss per PD decreases if the cells are cultured under a lower oxygen concentration. Something else must be eroding the telomeres. Activation of certain oncogenes or of genes that regulate the cell’s progress through the G1 phase of the cell cycle can induce a state of replicative senesence that is operationally indistinguishable from that observed in the classic senescent fibroblast. If the same phenotype can be induced by different

stimuli, this casts doubt on the uniqueness of any specific stimulus. Boukamp (2003, p. 113) asked “whether telomere shortening is the causal event (the clock work) for aging or just a marker (the hand of the clock) of an as yet unidentified mechanism.” A computer model of telomere loss led to the conclusion that the main determinants of the Hayflick limit are parameters of cell population kinetics rather than any sort of mitotic clock (Golubev et al. 2003). A review of the empirical data by von Zglincki (2002, 2003) led to the similar conclusion that telomere loss is partly due to the “end-replication problem” as discussed above but is mostly due to the effects of oxidative stress. DNA damage is a major consequence of the reaction of reactive oxygen species with biological macromolecules (see chapter 10). In G0-arrested fibroblasts under oxidative stress, the telomeres accumulate about 10-fold more single-strand breaks than elsewhere in the genome, largely because of the lower efficiency of DNA repair in the telomere (Petersen et al. 1998). These singlestrand breaks in the telomere are translated into telomere loss shortly after the cells are allowed to resume dividing. Cells with low levels of antioxidant defense have relatively high levels of telomere loss per PD, whereas cells with high levels of antioxidant defense have low levels of telomere loss per PD. In fact, these latter cells have PD losses of only 5–20 nucleotides per PD, which is at the expected level due to the end replication problem only. Thus, most of the telomere loss seen in vitro is due to oxidative stress. This conclusion is supported by data obtained from studies of human chrondocytes raised in vitro under low or high oxygen concentrations: Cells under high oxygen displayed higher levels of oxidative damage, lower PD levels, and premature senescence relative to the cells under low oxygen (J. A. Martin et al. 2004). Overexpression of telomerase extended the cells’ PD values only in the cultures raised under the low-oxygen conditions. Although telomere lengths are equivalent in newborn boys and girls, they are longer in adult women than in men (Aviv et al., 2005). This observation raised the question as to whether the longer life span of women is related to telomere

12.2 Aging in Dividing Cells: Cellular Life Spans and Organismic Longevity

length. The relationship is most likely correlative rather than causal, reflecting the female’s greater resistance to oxidative stress. Telomeres are not replicometers but are “rather sentinels for genomic damage and mutation risk” (von Zglincki 2003, p. 125). We do not need to describe a replicometer type of mechanism within the cell, but rather one that allows the cell to bring about a protective response of replicative senescence in response to oxidative DNA and other damage. Is there an empirical relationship between telomere length and cell function in intact animals as well as in cultured cells? Women under high psychological stress have higher levels of oxidative damage and significantly shorter telomeres than do nonstressed controls (Epel et al. 2004; Sapolsky 2004). Other human studies suggest that individuals diagnosed with premature myocardial infarction (MI) have significantly shorter telomeres (~300 base pair difference) in their leukocyte chromosomes than do comparable age-matched controls (Brouilette et al. 2003; Samani et al. 2001). The individual’s risk of dying in the near future was significantly and inversely related to the difference in telomere length, but this risk may be abolished by the prophylactic use of statin drugs. It is tempting to interpret this correlation as a causal relationship and relate the telomere differences to differences in endothelial cell response to injury. However, the beneficial effect of the statin drugs suggests that some other mechanism may be involved. In the rat, at least, telomeres shorten with age in a tissue- and gender-specific manner (Cherif et al. 2003). Thus, the observed correlations may arise from multiple causes, and we need to be cautious rather than jumping to a conclusion. In any event, these model systems promise to be even more informative as they are investigated further.

12.2.7 Cell Signaling Pathways Leading to Replicative Senescence I begin our investigation of the pathways responsible for replicative senescence with an examination of the transformed cell lines. A somatic-cell genetics approach, taken by Pereira-Smith and

433

Smith (1983), assumed that the control of cell proliferation is genetic and that senescence was the normal phenotype of the cell. Given these two assumptions, a nonsenesescing cell line must have a mutated gene in the cell proliferation regulatory pathway. Nonsenescing cell lines that complement each other (i.e., that generate senescing hybrids) would have mutations in different genes. If the two cells had lesions in the same gene, complementation would not be possible, and their hybrid cell would display an immortal phenotype. To test these hypotheses, the investigators fused different immortal cell lines and determined which cell pairs formed immortal hybrids. The results, based on the outcome of many cell fusion experiments, showed that replicative senescence is dominant to replicative immortality (see J. R. Smith and Pereira-Smith 1996). These studies identified four complementation groups, suggesting that there are at least four genes or gene pathways controlling cell proliferative senescence. Note that this is a minimum, not a maximum, number. In some cases, it was possible to assign specific chromosomes to specific complementation groups, such as chromosome 4 to complementation group B (Ning and Pereira-Smith 1991). No immortal cell lines could be assigned to more than one complementation group, which meant that a small number of highly specific genes are involved in promoting cell senescence (Pereira-Smith and Ning 1992). Finally, cells of diverse tissue origins may belong to the same complementation group, suggesting that the genes controlling replicative senescence do not act in a tissue-specific manner. One of these genes in complementation group B on chromosome 4 has been tentatively identified as coding for a transcription factor (Ehrenstein 1998). The gene, now called MORF4, is up-regulated in senescent and quiescent cells, but down-regulated in actively dividing cells. These investigations showed that transformed cells had an up-regulated or active genetic system regulating their proliferative ability. The existence of an active genetic mechanism in nontransformed cells was first supported by the findings of Lumpkin and collaborators (1986), who isolated poly-A+ RNA from senescent human

434 Chapter 12 Senescence as a Breakdown of Intracellular Regulatory Processes diploid fibroblasts and injected it into cells that were capable of proliferating. The mRNAs significantly inhibited the host cells from entering mitosis relative to controls. The inhibitory ability could be abolished if the mRNA fraction was treated with RNase before being injected. Lumpkin et al. noted that the inhibitory signal is not present in young, or proliferation-competent, cells. The implication of these data is that the inhibition of cell proliferation is an active process that requires the transcription of some gene(s) and that the presence of the corresponding mRNA in the cell is necessary if the cell is to be inhibited from entering mitosis. Quiescent cells have high levels of the SD1-1 gene. This gene is induced by the p53 gene and has been identified as an inhibitor of multiple complexes of cyclin G1 and cyclin-dependent kinases (Smith 1992). Presumably, repressing the SD1-1 gene would release the inhibition and allow cell division to resume. Because transformed cells activate genes so as to enable them to divide, and senescent cells activate genes that inhibit them from dividing, it follows that the regulatory system controlling whether a cell divides must have genes encoding both activating and repressing proteins. Whether a given cell is enabled or inhibited to divide likely depends on the specific genes activated in the cell. In addition, the existence of quiescent cells indicates that some mechanisms of mitotic inhibition are reversible. However, there is an extensive layer of post-translational control superimposed on transcriptional regulation. For example, the retinoblastoma protein (RB) is a tumor suppressor, and in its nonphosphorylated form it causes the cell to arrest in the G1 phase of the cell cycle. If the RB protein is post-translationally phosphorylated, it loses this inhibitory activity, and the cell will proceed through the cell cycle. RB is phosphorylated by a protein kinase called p45, but that p45 may be sequestered by a second protein called statin, or p57. Removal of the p45 protein prevents it from activating the RB protein and relieving the G1 arrest (Jazwinski et al. 1995). The phosphorylation of other cell-signaling proteins depends not only on the activation of a kinase but also on the regulated presence of a

phosphatase to remove the phosphate group from the protein. For example, the Ras/ERK/MAP kinase pathway responds to various stressors and activates a wide variety of genes. It is controlled by the MAP kinase activities, which are normally controlled via a negative feedback regulation of the MAP kinase phosphatases. In senescent fibroblasts, MAP kinase activities increase, as do the downstream Jun kinase activities, but only because the cell’s protein degradation mechanisms (proteasome) become increasingly inactive, probably as a result of oxidative stress (Hutter et al. 2002; Rivett et al. 2002; Torres et al. 2003). Consequently, the level of many proteins increases, including the phosphotases in this particular case, and this brings about manifold and varied effects on different cell processes depending on the specific signaling molecules affected. Post-translational control allows the cell to respond quickly to metabolic changes. But when the metabolic changes are those arising as a result of oxidative stress, then the rapid amplification of these deleterious effects by the posttranslational control system can skew the cell into successive states of decreasing function. The above discussion suggests that the cell’s transition through the cell cycle can be blocked by permanently repressing one or more genes responsible for the synthesis of key growth regulatory proteins. The validity of this assumption is illustrated in table 12.4, which summarizes the results of many studies into the ability of G1 regulatory genes to be induced in quiescent or senescent fibroblasts. The fact that very few early but many late genes are inactivated suggests the existence of gene cascades in which one early gene might activate multiple late genes. The same phenomenon occurs in the cell’s ability to mobilize its antioxidant defenses (table 7.12). A feedback control loop may also be operative here, in that one function of some late genes is to downregulate one of the early genes (e.g., p21). The p21 protein is an inhibitor of the various CDK genes, whose activity is essential for cell division. The down-regulation of p21 can take place in quiescent but not in senescent cells. The failure to down-regulate maintains the inhibition and prevents the cell from dividing. Figure 12.7 sug-

12.2 Aging in Dividing Cells: Cellular Life Spans and Organismic Longevity

Table 12.4 Differential Inducibility of Mitogenic Genes in Quiescent and Senescent Fibroblasts No. of G1 regulatory genes capable of being induced in Stage Early G1 Mid-G1 Late G1

Quiescent cells

Senescent cells

8 9 9

6 7 2

Source: data from Campisi (1996).

gests the actual pathways involved in that it shows some of the potential regulatory circuits within the cell connecting the inputs of an assortment of stressors to the outputs consistent with one of the four replicative phenotypes discussed above. There is a paradox here. The inhibition of cell proliferation is controlled by an active and positive genetic process, implying that the regulation has an adaptive function. But I have already reviewed the evidence suggesting that aging is not adaptive. If this reasoning is correct, the inhibition of cell proliferation should not be viewed as an aging phenotype. Yet proliferation-incompetent cells are much more prevalent in older organisms, so this inhibition seems to be part of the aging phenotype. One solution to the paradox was provided by Dykhuisen (1974) and Holliday (1990), who suggested that the limit to cell growth is a barrier against uncontrolled proliferation of potential or real tumor cells. Another solution to this paradox is that the loss of proliferation potential of somatic cells is simply another manifestation of the disposable-soma theory (Kirkwood 1987), which I discussed in chapter 4. The implication of figure 12.7, also stated by Ben-Porath and Weinberg (2004), is that the cell’s replicative senescence is not adaptive in any way but is simply a manifestation of the particular cell’s exposure to oxidative and other sorts of stressors, since some forms of oxidative damage inhibit the cell from dividing, while other forms of damage transform it into a proliferating cancer cell. Stochastic damage brings about outcomes that might appear at first to be programmed but are really just the result of sto-

435

chastic damage on different parts of the cell’s regulatory apparatus. Given these facts, we would expect few cells from an embryonic donor to undergo replicative senescence because most of them would be in the predifferentiation state. But with each cell divison, there is a finite chance that a given cell will be recruited into the nondividing or predifferentiation population and have its gene activity patterns altered, signal transduction patterns changed, DNA synthesis-inhibiting proteins synthesized, and future cell divisions inhibited. With time, one would expect more and more cells to cease dividing and to begin differentiating. Growth would thus be more often associated with cells from young donors. This explanation could indirectly account for the difference in the remaining PDs observed between cells from embryonic and adult donors (see table 12.1) and may also account for the difference in the ease with which endothelial cells from adult donors of different ages are able to establish primary cultures—that is, to enter Phase I (Ogborn and Martin 1985). Together, these findings suggest that it is unlikely that the ultimate life span of the individual organism is the simple result of a failure in the proliferative capacity of its component cells. However, changes in the regulation of the cell cycle may be representative of the mechanisms that regulate other important physiological processes that play an important role in organismic senescence. It is a common clinical observation that the wounds of the healthy elderly heal more slowly, although they do eventually heal as well as do those of younger subjects (Gosain and DiPietro 2004). This agerelated change may arise from changes in the cell’s microenvironment (e.g., growth factors, inflammation) as well as from alterations in the elderly’s proliferative ability (Ashcroft et al. 2002). It is too simplistic to suggest that this lower rate of healing has its biological basis only in the described behavior of fibroblast cell in culture. It would be equally rash to completely ignore these changes in a cell’s proliferative ability when trying to understand how tissues and organs age in situ. Other cell types, such as the T-lymphocyte (Effros 1996; Perillo et al. 1989), are exhibit a

436 Chapter 12 Senescence as a Breakdown of Intracellular Regulatory Processes Various Stressors (Oxidative Stress, UV, Chemicals, etc.)

Appropriate Stress Reactions ( Telomore, DNA Damage, Oncogene Expression, Loss of Cell Contact, etc.)

Loss of Telomere Proteins

RAS

p14

ATM/ATR RAS

RAF

Loss of Structure

MEK

DNA Damage Response

ERK p16

p16 p21 RB

p53

Irreversible Senescent Arrest

Reversible Senescent Arrest Cyclin D1 (Variable Expression) oncoRAS

p53

Apoptosis Loss of

Extensive Growth

Figure 12.7 The different pathways that lead to cell senescence, with a suggestion of the cell signaling transduction pathways involved. Note the difference between the reversible and nonreversible paths, and how many different paths converge on just a few key proteins.

characteristically limited in vitro life span, and these changes may be similarly related to the systematic decrements in function involving those cell types. Finally, human diploid cells are more susceptible to oxidative stress at late passage than at early passage, apparently because of decreased glutathione concentration and catalase activity (Yuan et al.1996), a finding with certain parallels to the situation in aging organisms (see chapters 7, 8, and 10). In this sense, replicative senescence in cells in vitro is a highly accessible model, the study of

which will enable us to better understand the changes with age of our own cells in vivo. Certainly cancer and atherosclerosis are the two most prevalent serious age-associated pathologies, and the successful treatment of cancer, at least, will depend in part on the knowledge of cell division regulation that we harvest from our studies on replicative senescence (Hartwell and Kastan 1994). The work initiated by Hayflick and Moorhead (1961) has enhanced our appreciation of the importance of basic cellular mechanisms in the aging process and has focused attention on understanding how

12.3 Senescence in Nondividing Cells

these mechanisms enhance longevity at both the cellular and the organismic levels.

437

understand the age-dependent changes that take place at the cellular level. In the following sections I examine several such systems of aging in normal tissues.

12.3 Senescence in Nondividing Cells 12.3.2 Fibroblasts and Skin Aging 12.3.1 Senescence and Differentiation in Vivo An extensive review of the role of cell senescence in the age-related changes observed in human aging has been written by Fossell (2004); a more limited overview has been done by Bird et al. (2003). Central to the cell senescence theory of aging is the idea that senescent cells, because of their altered phenotype, act as the proximal causal agents of age-related loss of tissue-specific function in vivo. This may happen because the senescent cell has altered concentrations of various proteins, which adversely affect the cell’s normal function. This change in the proteome may be the result of some specific change in the genome. For example, it has been suggested that Werner’s syndrome demonstrates the ability of cells with altered function to bring about a series of symptoms that closely resemble the usual aging process (Bird et al. 2003). However, the altered proteome may have arisen not from a specific gene change but rather from some general failure such as alteration in the levels of protein synthesis or protein degradation. Fossel (2004) illustrated this general failure hypothesis by discussing the consequences of changing protein turnover rates. In the simplest case, a daily damage rate of 1% per day to the specific proteins coupled with a 50% turnover rate would yield at equilibrium a mix of 98% undamaged and 2% damaged molecules on any given day. However, if the turnover rate were to drop to only 2%, then at equilibrium there would be a 50% damage level. The cell might be able to cope with a 2% damage level, but it is unlikely that normal function can be maintained in the face of a 50% damage level. Undoubtedly, we need to examine the cell biology of the differentiated and nondividing normal cell both in vivo and in vitro to better

Bayreuther’s work (table 12.3) showed that over the course of adult life, there is a shift in the types of differentiated fibroblasts found in the skin such that this organ becomes increasingly composed of postmitotic cells similar to the replicative senescent cells seen in culture. Senescent skin fibroblast cells were initially reported to produce higher amounts of a stainable form of the beta-galactosidase enzyme, which is found only in senescent cells (Dimri et al. 1995). This senescence-associated beta-galactosidase (SA-b-gal) staining technique initially showed that only the skin of older people contained such cells, although even in these individuals most skin cells do not fall into this category. The presence of this enzyme was the first evidence showing accumulation of senescent cells in vivo, and SA-b-gal seemed to confirm the indirect evidence that such cells accumulate with age in vivo and in vitro. Further studies showed that conditions other than age increased SA-b-gal staining in cells, and so the validity of the technique is not clear (Bird et al. 2003; Severino et al. 2000). Senescent and proliferative cells are different from one another at several different levels of organization, and this includes difference in gene expression. Certain changes in gene expression are important and relevant to both cellular and organismic physiology. Senescent cells may constitute a small minority of the cells in old skin (Campisi et al. 1996), yet their effect on the skin may be disproportional to their number. All or even most cells in an integrated tissue do not have to be senescent for the tissue to lose function. Human skin shows an age-related decrease in its ability to repair itself. This loss is associated with increased fragility of the skin, reduction in the amounts of types I and III collagen, and loss of the collagen’s regular organization. Dermal cells,

438 Chapter 12 Senescence as a Breakdown of Intracellular Regulatory Processes for example, switch from matrix-producing to matrix-degrading when they become replicatively senescent; it is reasonable to wonder whether this change plays a role in the appearance of degraded collagen in the extracellular matrix, arising as a consequence of the overproduction of active collegenase. The degradation of the matrix undoubtedly has collateral effects on neighboring cells because alteration of the extracellular matrix alters the gene expression patterns of the cells normally adherent to that matrix. Skin can be dissociated into its component parts. If the parts are subsequently mixed together, they will reconstitute morphologically normal skin. If one uses late-passage human fibroblast (i.e., senescent) cells, the reconstituted tissue shows an increased fragility and intermittent splitting of the dermal–epidermal junction. A major cause of the formation of wrinkles in situ has to do with the decreased strength of the contacts between these two tissue layers. However, if one uses fibroblasts immortalized by ectopic expression of telomerase (e.g., hTERT) and uses these in reconstituting aged skin, the resulting reconstituted skin does not show these age-related changes (Funk et al. 2000).

12.3.3 Endothelial Cells and Cardiovascular Senescence Cardiovascular abnormalities constitute one of the most important classes of age-related diseases. Major improvements in cardiovascular medicine during the past 50 years have led to significant improvements in the life expectancy of middle-aged and older individuals. Many of these treatments relied on surgical treatment and/or replacement of diseased vessels to alleviate the disabling symptoms of atherosclerosis. We now understand that the underlying causes of the disease stem from functional changes characteristic of senescent endothelial cells and that the surgical interventions must be accompanied by significant changes in lifestyle if the basic causes are to be alleviated. How can a small population of senescent cells bring about such extensive changes in the vascular system? Recent work has begun to answer this

question (for details and references, see reviews by Foreman and Tang 2003; Minamino et al. 2004). The endothelial cells line the inside of the entire vascular system. They are in direct contact with the blood flow and are exposed both to its hemodynamic stresses and to any molecules capable of inducing oxidative or other stresses. A summary of the functional cascades involved is shown in figure 12.8. This figure should be viewed as an extension of figure 12.7. Under normal circumstances, endothelial cells in vivo are quiescent, rarely divide, and have a turnover time of about 3 years. They have a limited proliferation ability and eventually enter an irreversible state of growth arrest. Such senescent cells are viable and metabolically active but are morphologically and functionally different from nonsenescent endothelial cells. They are identified by their ability to express the SA-b-gal stain, which is found in most replicatively senescent endothelial cells in vitro. The intracellular actin network of nonsenescent cells is oriented along the long axis of the cytoplasm, but in senescent cells it is organized in a circular pattern around the periphery of the cytoplasm. Integrins and other cytoskeletal proteins are similarly reoriented, and this relocation does not involve any noticable alteration in protein levels. Such qualitative changes in cytoskeletal organization would be expected to affect the cell’s migratory ability, for example. It is known that senescent cells fail to migrate in response to the signals that stimulate young endothelial cells. One consequence of the loss of migratory ability is the increasing inability of old cells to travel to a nearby wound site and begin the healing process. The patterns of senescence-induced patterns of altered gene expression are different for each cell type (Shelton et al. 1999). These specific alterations, when combined with changes in the cell’s ability to maintain protein turnover, lead to modulations in the level of important enzymes and other proteins. An important example is endothelial nitric oxide synthetase (eNOS), which is essential for the production of the cellsignaling molecule nitric oxide (NO). eNOS is expressed constituitively in young endothelial cells but is found in significantly lower levels in

12.3 Senescence in Nondividing Cells

439

Cell type specific changes in Gene Expression

Changes in Protein Turnover

Altered Protein Levels

Alteration in Endothelial Cell Function

eNOS

NO production

ICAM-1

Telomerase evels Leucocyte binding Inflammatory Response

Redistribution of cytoskeletal proteins

Figure 12.8 The role that the inflammatory response plays in damaging the endothelial cells and altering them so they are liable to give rise to plaques after an inflammatory response.

senescent cells. Because higher levels of NO inhibit binding of inflammatory cells to the endothelium, the decrease in eNOS allows leukocytes to adhere to the cell. This results in elevated levels of cytokine expression. In addition, senescent cells increase their levels of the intercelluler adhesion molecule-1 (ICAM-1). This molecule serves as the binding ligand for several membrane-bound molecules characteristic of inflammatory cells such as macrophages and neutrophils. The combined effects of eNOS and ICAM-1 act as a false alarm of infection and induce immune cells to bind to the healthy but senescent endothelial cell. This aberrant response triggers the body’s defenses, such as the initiation of the nuclear factor kappa B (NFkB) signal transduction pathway and the start of an inflammation response, which only succeeds in the immune cells causing further harm to otherwise healthy cells. The details of this inflammatory response are presented in chapter 13. Mild chronic oxidative stress is a major trigger for the shortening of telomeres and the premature onset of cell senescence in endothelial cells (Kurz

et al. 2004). Atherosclerosis is believed to be initiated by an inflammatory reaction of some sort. It is interesting that the lowered eNOS levels not only are dramatically reduced in those areas of the vasculature exposed to the highest levels of hemodynamic stress, but that the pattern of lowered eNOS and NO levels appears to mimic the normal distribution of atherosclerosis. The inflammatory response brought about by replicative senescence may be an important (albeit not the only) factor inducing the development of atherosclerosis in aging humans.

12.3.4 Cardiac Myocytes and Cardiovascular Senescence The heart has traditionally been thought of as being a postmitotic organ composed of a fixed population of myocytes whose number is determined shortly after birth and which had to last until the death of the organism. Under this paradigm, cardiovascular aging was envisioned as

440 Chapter 12 Senescence as a Breakdown of Intracellular Regulatory Processes arising from the interaction of the existing cells with various stressors that generated different pathological conditions in a positive feedback system, eventually leading to a significant decrease in cardiac function. Later observations cast doubt on this paradigm—particularly the recognition that heart transplant patients who received hearts from donors of the opposite sex eventually had significant numbers of host-type myocytes in their heart (Hucht-Zeisberg et al 2004). This study implied that there must be some source of new cardiac myocytes in the adult human. The existence of a cardiac stem cell niche would enlarge our perception of cardiac aging by now involving, in addition to the factors mentioned above, the population kinetics of the cells. Several groups have recently published studies that strongly support this new paradigm. Torella et al. (2004) did a valuable study in which they examined the effect of aging on the population kinetics of cardiac myocytes in the hearts of both wild-type mice and transgenic mice with a presumed overexpression of insulinlike growth factor-1 (IGF-1). In chapter 7, I presented the work of Holzenberger et al. (2003), which demonstrated that female mice heterozygous for a knockout mutation of the IGF-1 gene, with a presumed 50% underexpression of the IGF-1 protein, had a significantly delayed onset of senescence and increased longevity. The work of Torella et al. (2004) may be regarded as a partial counterexample because their transgenic mice overexpressed IGF-1, but only in the myocytes, due to a tissue-specific promoter. The resolution of these two studies and their apparently contradictory results needs to be addressed, and one possibility is presented below. Table 12.5 summarizes the data on cardiac myocytes presented by Torella et al. (2004). The myocytes of the wild-type mice undergo a large increase in oxidative damage that is inversely proportional to their telomerase activity and telomere length (remember the prior discussion of telomere sensitivity to oxidative stress). Telomerase is thought to be regulated in proportion to the levels of phosphorylated Akt in the nucleus; these levels are low in the wild-type mouse (remember that Akt [also known as protein kinase B PKB] is an integral component of the insulinlike signaling pathway; see figure 7.32). The presumed con-

sequence of these low phospho-Akt levels is that, although the number of stem cells detected in the heart increases with age, a large proportion of them are inhibited from dividing and so do not contribute to myocyte replacement. Molecular markers of cell senescence increase, the incidence of age-related cell death increases markedly, and thus more cells die than are replaced. The inevitable outcome is that the number of myocytes in the heart decrease significantly with age. Those surviving senescent myocytes are larger, their sarcomeres show lower levels of contractile activity, and the old wild-type heart as a whole shows signs of congestive heart failure (i.e., increased left vetricle end-diastolic pressure and decreased left ventricle developed pressure). The death of the organism can likely be traced to these alterations in population kinetics and function. Table 12.5 shows that this progression is not inevitable. Transgenic mice that overexpress IGF-1 show an increase in phospho-Akt levels over their life span, and this presumably leads to their increased telomerase activity and higher proportion of effective cardiac stem cells in the heart. The altered replacement cell kinetics leads to a more or less constant replacement rate of new myocytes to take the place of those senescent cells that undergo cell death. The transgenic young animals not only have a higher myocyte number than do their wild-type counterparts, but they maintain this higher cell count throughout their lifespan in contrast to the normal mice. These cells lose contractile function much more slowly than is the case with the wild-type, and the heart better retains its contractile properties. In fact, the functional indices of the hearts of old transgenic mice were comparable to those of middle-aged wild-type mice (Torella et al. 2004). Restoring the ability of the stem cell population to divide allows for the significant delay in the onset of the aging cardiomyopathy phenotype. The induction and regression of cardiac hypertrophy involves changes in gene expression patterns relative to the normal animal. The genes involved are being characterized (Friddle et al. 2000). Those involved in induction of the phenotype are not identical with those involved in its regression. The process by which the heart

12.3 Senescence in Nondividing Cells

441

Table 12.5 Insulinlike Growth Factor-1 (IGF-1) Retards Cardiac Aging in Mice by Altering Cell Senescence Progression and Stem Cell Kinetics Wild type, age (months) Marker 8-OH-dGa Telomere lengtha Apoptosisa Necrosisa Total stem cells (ckit+)a Blocked stem cells (ckit+, p16+) Dying/forming cells Myocyte no.a p16INK4a + p53 +

IGF-1-treated, age (months)

4

10/12

20/22

4

10/12

20/22

1.0 1.0 1 2 1 12% ~1.1 ~1 25% 5%

— — 4 8 — — ~3 — 42% 9%

1.64 0.65 12 60 1.9 70% ~16 ~0.67 82% 15%

0.61 1.32 0.89 0.93 1 10% ~0.7 ~1.5 9% 3%

— — 0.32 0.29 — — ~0.7 — 15% 5%

0.84 1.21 0.22 0.07 2.3 16% ~1.0 ~1.4 35% 9%

Source: data from Torella et al. (2004). aData

expressed relative to value of 4-month-old wild-type mice. Note that IGF-1 data are generally better across the board.

remodels itself is not just the reverse of the process by which hypertrophy was initially induced but involves two distinct gene pathways. Alterations in the population structure of the cardiac myocytes leads to these and perhaps other specific alterations in gene activity patterns. These findings do not apply only to mice. A similar study compared human hearts from healthy aged people to hearts of aged individuals suffering from cardiac myopathy (Chimenti et al. 2003). Although the human data are open to more questions than the mice data (people are, after all, somewhat less controlled than lab mice), the overall conclusion was similar. Cardiac cellular senescence led to an accumulation of old cardiac myocytes with impaired function, the effects of which could not be compensated for by replication of new myoctes. The ensuing loss of function led eventually to overt heart failure due to aging myopathy. How do we explain the apparent contradiction between the premature aging phenotypes induced by normal levels of IGF-1 in laboratory animals (see chapter 7) and the delayed aging phenotype induced by an overexpression of IGF-1 in the mice used by Torella et al. (2004)? Three possibilities come to mind. First, worms and flies are shorter lived organisms in which somatic

cell replacement does not play an important role. Although they do have germline and blood cell stem cells, there is no evidence indicating an important role of stem cells in the other somatic tissues. In addition, flies do not possess the standard telomeres, and so a lack of telomerase stimulation resulting from low IGF-1 levels would have no detectable effects on their life span. Stem cells appear to play a more important role in mammals. The 129 mouse strain used in the study by Holzenberger et al. (2003) is characterized as being a weak or nonrobust wild-type strain (Masoro 2004). If that is the case, then a reasonable speculation is that their stem cell component is less active than that of other more robust strains of mice. The up regulation in their systemic stress resistance mechanisms in other critical organs may simply have outweighed the detrimental effects of a decreased phosphoinositol-3-kmas (PI3K)-Akt- telomerase signaling pathway on the heart. Third, there are multiple isoforms of the PI3K molecule in flies and mammals. PI3K is an integral portion of many different signaling pathways regulating diverse pathways. The existence of diverse PI3K isoforms suggests that the molecular specificity of the different signaling pathways depends in part on using different isoforms. The PI3K isoform involved in up-regulating stress resistance may

442 Chapter 12 Senescence as a Breakdown of Intracellular Regulatory Processes not be the same isoform involved in regulating telomerase activity. In this case, then, there may not be any contradiction between the two studies if the processes involved use different signaling pathways. These hypotheses remain to be tested.

12.3.5 Blood Cells and Cardiovascular Senescence People whose blood cells have a longer telomere length (i.e., were in the upper half of the population) live significantly longer than people of the same age whose blood cell telomere length was in the lower half (Cawthon et al. 2003). This decreased longevity of the shorter telomere group was characterized by a 1.86-fold higher mortality from all causes, with heart disease (3.2-fold increase) and infectious disease (8.5-fold increase) being major components. This correlative data are consistent with the information discussed above. Overall cardiovascular mortality may depend on altered population kinetics in the endothelial, myocyte, and white blood cell compartments of the circulatory system. There may be a genetic component for this difference in telomere length in blood cells. De Haan and Van Zant (1999) searched recombinant inbred mice for quantitative trait loci (QTL) associated with the ability of stem cells to undergo cell cycling and division. Such QTLs were located on chromosomes 4, 7, 9, and 11. A study by Gelman et al. (1988) surveyed these same strains for QTLs involved in regulating life span and found four loci located on chromosomes 2, 4, 7, and 11. A comparison of the two data sets showed that life span and cell cycling ability were regulated by identical loci on chromosomes 7 and 11 (the QTLs on the other chromosomes differed). Disappearance of mature blood cells from the circulation roughly corresponded to the maximal life span of the strain. Thus, high levels of cell cycling are inversely correlated with life span. This conclusion suggests that lower levels of stemcell activity over a longer period of time are more likely to extend life span than are higher levels of stem cell activity over a shorter period of time

and that the control of stem cell activity is under polygenic control. It is important to point out that many aspects of cardiac aging are not directly explained by these studies on cell senescence. Lakatta (2003) has written a series of review articles in which these other functional changes are described in some detail, and the interested reader is referred to them. But certainly cell senescence plays an important causal and/or contributory role in the aging of the cardiovascular system.

12.3.6 Stem Cells and Senescence in Other Tissues The preceding sections presented evidence that cell senescence underlies much of the age-related loss of function characteristic of skin and cardiovascular aging. Are these isolated cases or are they proof-of-concept examples of a more general situation? I believe the latter to be the case. Fossel (2004) has described the possible role of cell senescence in all the different tissues of the body. He states that “cell senescence doesn’t cause aging so much as it is aging” (Fossel 2004).Other recent reviews (de Haan and Van Zant, 2002; Schlessinger and Van Zant, 2001) reinforce this view and bring home the point that stem cells exist in many (perhaps all) tissues, that they are not immortal but can age, and that an aging stem cell population can adversely affect tissue and organ homeostasis in a manner similar to those described earlier. Newly formed replacement cells play an important role in maintaining tissue and organ function. This is true even of such complex structures as the memory circuits of the hippocampus (Jessberger and Kempermann, 2003). The radial glial cells in the ventricular wall of the brain give rise to the adult neural stem cells (Merkle et al. 2004). Now that their source is known, it should be possible to study stem cell lineage and potency in greater detail. Stem cells are susceptible to environmental insults such as oxidative stress, and those effects can limit their replicative potential. Some of the beneficial effects on longevity of the enhanced stress resistance discussed in chapter 7 may arise because

12.4 The Death of Terminally Differentiated Cells

their effects are channeled through processes related to somatic cell replacement and maintenance of functional homeostasis.

12.3.7 Lipofuscin and the Waste Accumulation Theory In its simplest form, the waste product theory of aging proposes that cellular aging is caused by the accumulation of intracellular waste products that cannot be destroyed or eliminated except through the process of cell division. It is well established that postmitotic cells, such as neurons and cardiacmuscle cells, accumulate deposits of irregularly shaped, lipid-rich, yellowish brown cytoplasmic pigment granules during the aging process. The varied forms of pigment granules that are encompassed by the term “lipofuscin” likely are a diverse and heterogeneous group of molecules that may have little in common with one another save the fact that they all wound up in the same granule. Lipofuscin is now generally believed to arise as a result of free radical-induced auto-oxidation of the molecular components of the cell, particularly the membranous structures that contain unsaturated lipids (see chapter 10). Some studies have shown that lipofuscin contains chemical components similar to those generated by in vitro auto-oxidation of subcellular structures. Studies such as the one involving the lowactivity and high-activity flies that we discussed in chapter 10 were thought to have manipulated the in vivo auto-oxidation rate and to have affected the rate of lipofuscin accumulation. In another study, crayfish were raised at five different temperatures ranging from 13 to 33°C, and their lipofuscin content was measured at different ages throughout their life (Sheehy et al. 1995). At any given age, the animals raised at the higher temperatures appeared to age more rapidly and had higher lipofuscin levels than did the animals raised at lower temperatures, suggesting a proportionality among temperature, rate of living, and free-radical production. But no evidence was put forth that lipofuscin was itself harmful or served in any way other than as an index of the organism’s past metabolic history.

443

The auto-oxidation rate has also been manipulated through diet. Vitamin E is one of the better known naturally occurring antioxidants (see chapter 10). Animals deficient in vitamin E have elevated levels of lipid auto-oxidation products in their tissues, and such animals appear to have a higher rate of lipofuscin accumulation in their tissues (Nandy 1984). Taken together, this evidence leads to the conclusion that the lipofuscin arises via the auto-oxidation of subcellular lipid components. Does lipofuscin have any deleterious effect itself? Lipofuscin appears to be chemically unreactive and thus probably doesn’t interfere at the chemical level with the proper functioning of the cell; however, in vitro measurements have shown a good relationship between the rate of autooxidation (which generates lipofuscin) and the species-specific maximum life span. This relationship suggests that the level of oxidative stress is the important factor in determining cell senescence. According to this point of view, the lipofuscin granules serve simply as visible biomarkers of reactive oxygen species-induced cellular damage. The age pigments arise when the cells’ antioxidant defense systems begin to decline and can no longer cope with the flux of reactive oxygen species. This conclusion not only points to the oxidative stress theory of aging (see chapter 10) but also suggests that our understanding of the aging process in nondividing differentiated cells depends on being able to characterize the mechanisms that limit the amount of oxidative stress suffered by the organism.

12.4 The Death of Terminally Differentiated Cells 12.4.1 Cell Death as a Stochastic Process: The Red Blood Cell Model Most differentiated cells survive uneventfully for a long time. But some, such as keratinocytes, plasma cells, and erythrocytes, have a limited life span in situ. They senesce and die. In many cases, such cells are destroyed or phagocytized by specialized cells.

444 Chapter 12 Senescence as a Breakdown of Intracellular Regulatory Processes What markers distinguish a dying cell from a healthy cell, and what do they have to do with the aging process? The fact that cells communicate with one another via their cell membranes suggests that a senescent cell has undergone important alterations at this structural level for it to be recognized as a target for phagocytosis. Moreover, the macrophage cells of the immune system are known to phagocytize the old, but not the young, red blood cells (RBCs) in humans and other mammals. Investigations into the molecular changes occurring during aging revealed that a terminal-differentiation antigen, the so-called senescent-cell antigen, appears on the surface of cells as they age (for review see Kay 1985; Kay et al. 1994). The presence of this senescent-cell antigen was initially revealed by the fact that certain types of immunoglobulin G (IgG) autoantibodies bind in high amounts to the membranes of senescent human RBCs, but only in very low amounts to the membranes of young RBCs. Middle-aged RBCs bind the IgG molecules at an intermediate level. Macrophages appear to recognize and target for destruction only those RBCs that have high numbers of IgG molecules on their membranes. These data suggest that the senescent-cell antigen appears gradually and that its appearance leads to increasing binding levels of IgG. The molecular aging of the RBC membrane is a cumulative process. The real question is the identity not of the IgG molecule, but of the senescent-cell antigen to which it binds. What is this molecule? Does it have any function in the nonsenescent cell? And by what process does it gradually appear in the membrane? Finally, does this antigen have any general applicability to the problem of aging and senescence? As mature human RBCs have a limited ability to synthesize new proteins, the senescent-cell antigen must have arisen as a result of the modification of a preexisting protein. Immunological investigations by Kay (1985) revealed that this antigen is one of the breakdown products of a normal membrane protein that was given the trivial name “band 3 protein,” otherwise known as the “anion exchanger” family of proteins. At

least four isoforms have been identified, and they have been found in all membranes examined (Kay et al. 1995). Their presence in the aging RBC makes them a convenient model, but the specific processes involved in this case may be more easily generalized as that of using oxidative damage in a membrane protein to signal the functional state of the cell. A similar general function was postulated for telomeres (Zglincki 2002, 2003). This transmembrane protein has a complex structure, is present in large amounts in the plasma membranes of normal RBCs, and appears to be responsible for a multitude of important functions such as anion exchange and glucose transport, serving as the binding site for hemoglobin, as well as for various enzymes involved in glucose metabolism. The membrane protein apears to link the internal surface of the cell membrane with the internal filamentous cytoskeleton. This is obviously an important molecule—one that is probably vital to the continued physiological functioning of the RBC. As the RBC ages, all of these normal functions decrease, and the number of IgG molecules bound to the surface of the cell membrane increases. The IgG molecules are binding to a particular segment of the band 3 protein that has been snipped off and translocated so that it now lies on the external surface of the membrane and can serve as a recognition and/or binding site for the autoantibodies. This altered molecule is the senescent-cell antigen and appears to arise in part as the result of oxidative damage (Kay et al. 1994). Mutations affecting the band 3 protein also affect aging of the cells in situ. One mutant exhibits accelerated aging, another exhibits decelerated aging, and a third is associated with neurological defects (Kay et al. 1994). In fact, brain cells from people with Alzheimer’s disease undergo posttranslational alterations in their band 3 proteins similar to those seen in the anion transport regions of the band 3 protein of aging RBCs (Kay and Goodman 1997). The band 3 protein is conserved across all vertebrates, suggesting that it has an important function (Kay et al. 1995). Molecules other than the band 3 protein may also be involved in this process of membrane aging (Aminoff 1988). If we assume that the RBC

12.4 The Death of Terminally Differentiated Cells

membrane is similar to that of the other cell types in mammals (and there is every reason to believe so), then it seems reasonable to conclude that the mechanism of membrane aging and breakdown is an important molecular process intimately involved in the regulation of the life span of some specialized somatic cells.

12.4.2 Cell Death as a Controllable Process: Programmed Cell Death Organisms appear to lose cells as they age. Some cells are killed by external injuries or trauma via necrotic cell death. Necrosis generally results from cell injury and is characterized by swelling of organelles, clumping of chromatin, and increased permeability of membranes. DNA breaks down randomly. The entire process is believed to be a chaotic breakdown of numerous cellular systems. But some cells kill themselves deliberately, using a mechanism encoded in their own genome. Different cell types from different organisms, dying as the result of a variety of chronic (as opposed to traumatic) causes, nonetheless undergo a strikingly similar series of cytological and biochemical alterations that are now generally referred to as “physiological (programmed) cell death” or, more commonly, “apoptosis.” Apoptosis often represents terminal differentiation and is the evolved mechanism by which multicellular organisms rid themselves of unneeded or dangerous cells. This topic has been reviewed by Lockshin and Zakeri-Milovanovic (1984), Kerr et al. (1987), White et al. (1994), Zakeri and Lockshin (1994), Driscoll (1995), Steller (1995), Vaux and Strasser (1996), Adams (2004), Orrenius (2004), and Green and Kroemer (2004). The following discussion is based largely on these sources. All cells of multicellular organisms carry the information and the mechanisms necessary to bring about their own destruction, but that this terminal pathway can be invoked only by specific developmental and physiological signals. Such signals may be either autonomous or nonautonomous. The process is characterized by shrinkage and fragmentation of cytoplasm and compaction of chromatin, accompanied by nonrandom DNA cleavage into

445

180 base pair fragments. Apoptosis is both a common and an evolutionarily conserved process and is under strict genetic control because its inappropriate occurrence could be detrimental for the organism. Nonetheless, a variety of different signals that may originate from either inside or outside the cell can influence the cell’s decision between life and death. This multiplicity of signals suggests that the genetic control mechanisms are complex. Examples of apoptosis during embryonic life include the loss of excess interdigital tissues, the death of unneeded neurons, and the destruction of the “wrong” set of sexual primordia. Apoptosis occurs during metamorphosis of all animals, as witnessed by the tadpole’s resorption of its tail en route to becoming an adult frog. The continuous elimination of cells has been observed in a variety of adult tissues—both in slowly proliferating tissues such as liver epithelium, adrenal cortex, and prostate gland, and in rapidly proliferating tissues such as the intestinal epithelium. The cell death within the adrenal cortex has been estimated to be just sufficient to balance the cell gain by mitosis. Programmed cell death occurs in endocrine-dependent tissues upon hormone withdrawal (e.g., the human premenstrual endometrium) or, in some cases, upon hormone addition (e.g., the tadpole tail, the insect intersegmental muscle, Müllerian ducts). Other signals involve the lack of particular growth factors. Despite the heterogeneity of the signals, they all seem to focus in on one target—namely, the gene pathway that will activate the apoptotic program. Shortly before the onset of apoptosis, the mitochondrial membrane loses its characteristic polarization. Without the existence of a membrane potential, it is no longer possible for the mitochondrion to produce ATP via oxidative phosphorylation as described in chapter 11. Cytochrome c and other proteins are released from the mitochondrion and translocate to the cytoplasm. There the cytochrome c cooperates with another expelled mitochondrial protein (Apaf-1) to bind an inactive cysteine protease called procaspase-9. These complexed proteins constitute the apoptosome. They cleave the procaspase-9 molecule into its active form, and this assemblage then touches off

446 Chapter 12 Senescence as a Breakdown of Intracellular Regulatory Processes the cascade of reactions leading to chromatin and organelle disintegration and eventual cell death. Under normal circumstances, cytochrome c is prevented from being released by the inhibitory effects of at least two other mitochondrial membrane proteins (Bcl-2 and Bak), all of which are members of a protein family that either positively or negatively regulate apoptosis. A related protein, Bax, is normally present in the cytoplasm and can respond to pro-apoptotic signals that arrive via other pathways (e.g., p53 induction of apoptosis). The caspases normally present in the cytoplasm are inhibited by members of a different protein family, known as the inhibitor of apoptosis proteins (IAPs). Finally, a third group of proteins that activate apoptosis (BH3 proteins) are normally inhibited by the Bcl-2 protein and its homologs. All these signals converge on one mechanism: the cytoplasmic Bax protein translocates into the mitochondrion, where it activates the process and causes the release of cytochrome c. The net result is that the system is prevented from activation by the compartmentalized dispersion of several proteins needed to start the process and by the maintenance of several of these proteins in an inactive state. Thus there is an elaborate triple-tiered regulatory system inhibiting the accidental activation of the apoptotic pathway. Yet it does not take much to set it off. For example, an inward surge of Ca2+ ions (perhaps arising from a stroke or from the subtler insults discussed in figure 13.4) can depolarize the mitochondrion and initiate apoptosis. In addition, once it starts, apoptosis is thought to be unstoppable. Yet the lens cells of the eye appear to practice a controlled sort of apoptosis by which they destroy all their organelles (for otherwise they would not be transparent) but do not destroy themselves nor their ordered layers of crystallin proteins (Dahm 2004). Such controlled partial destruction is termed autophagy and is discussed below. The apoptotic process is still being characterized, and so the foregoing discussion should be regarded as a tentative summary description which will likely be changed by future data. What is the relationship of apoptosis to aging? There is no obvious direct relationship. Wildtype nematodes usually lose 131 specific cells to

apoptosis during development. Animals carrying gain-of-function cell death (ced) mutations, such as ced-9, develop and live normally, despite the presence of the extra cells. Loss-of-function ced-9 mutations, however, are lethal and cause many extra cell deaths in embryos. The conclusion that aging and cell death proceed independently of each other is supported by the observation that loss-of-function ced-3 and/or ced-4 mutants appear to be normal and have normal life spans. But this does not preclude an important indirect relationship. The mammalian Bcl-2 gene is homologous to the worm’s ced-9 gene and plays an important role in apoptosis, as described above. Bcl-2 was first identified because its elevated expression in a mouse mutant prolonged the life span of B cells and caused lymphomas. The inappropriate inhibition of cell death shortened organismic longevity. However, mice without a functional Bcl-2 gene die as embryos, exhibit massive cell death in their neural and hematopoietic systems, and thus show that the inappropriate stimulation of cell death also shortens organismic longevity (Motoyama et al. 1995). The process must be precisely controlled. It is possible that one major function of apoptosis is to serve as a precisely targeted defense mechanism against dysfunctional and/or potentially immortal cells. This possibility is supported by the observation that some of the genes involved in proliferation control (such as the c-myc and c-fos oncogenes, or the p53 tumor-suppressor gene) are actively involved in regulating apoptosis as well. Given that senescent cells often exhibit an inappropriate regulation of cell activities, the possibility also exists that inappropriate age-related implementation of apoptosis could result in the death of essential cells and thereby contribute to senescent decline. Table 12.6 lists some diseases associated with the abnormal induction or inhibition of apoptotic cell death. To the extent that these diseases arise as a result of the age-related breakdown of the cell’s programmed cell death mechanisms, these diseases highlight a common age-related failure mode of somatic cells. Warner et al. (1995) have presented a model in which they attempt to link the roles of apop-

12.4 The Death of Terminally Differentiated Cells

447

Table 12.6 Diseases Associated with Abnormal Apoptosis Inhibition of apoptosis

Activation of apoptosis

Cancer Follicular lymphomas Carcinomas with p53 mutations Hormone-dependent tumors Breast cancer Prostate cancer Ovarian cancer Autoimmune disorders Systemic lupus Immune-mediated glomerulonephritis Viral infections Herpes virus Pox virus Adenoviruses

AIDS Neurodegenerative disorders Alzheimer’s disease Parkinson’s disease Amytrophic lateral scherosis Retinitis pigmentosa Cerebrellar degeneration Myelodysplastic syndromes Aplastic anemia Ischemic injury Myocardial infarction Stroke Reperfusion injury Toxin-induced liver disease Alcohol

Source: from figure 4 of Thompson (1995).

tosis, caloric restriction, and oxidative damage into a unifying hypothesis to explain the retardation of aging by caloric restriction. If the somatic cell’s ability to undergo apoptosis decreases as the cell approaches the end of its replicative life span, the proportion of oxidatively damaged cells in a given tissue will increase with increasing age. But caloric restriction up-regulates the incidence of apoptosis. Therefore caloric restriction not only would reduce the incidence of oxidative damage suffered by the cells according to its ability to upregulate antioxidant defenses and attenuate the formation of reactive oxygen species (see chapters 7 and 10), but it also would improve the tissue’s ability to get rid of defective cells. The result would be a longer period during which tissues were maintained and a slowing of the aging process. This model may be viewed as an independent elaboration of the de Grey (1999) model discussed in chapter 10. It is also an integral part of the mechanisms involved in extension of the health span, as shown in figure 9.6. Some evidence suggests that individual cells cultured alone have a short survival time. Cells cultured in vitro with others of their kind live longer. Raff (1992) has suggested that there are important social controls on cell survival. Just as a cell seems to need to receive secreted signals

from other cells in order to proliferate (Edgar and Lehner 1996), so too it seems to need to receive signals (either positive or negative) from other cells if it is not to activate apoptosis and kill itself. This observation suggests that we could view our cells as integral members of a population that are continuously being questioned as to the appropriateness of continuing to live in that population. Cells that do not receive the signals of approval kill themselves. One advantage of such a cell-signaling mechanism is that it is a simple way to eliminate cells that are in the wrong place. In addition, competition among the cells for limiting amounts of secreted signals could result in the continuous survey of the population by its members and provide a mechanism for the control of cell numbers in particular tissues and organs, as well as the automatic elimination of abnormal cells and/or adjustment of normal cells to the changing internal environment of the organism (Ruhe et al. 1997). Either alternative would provide a mechanistic cell-level explanation of homeostasis. Widespread failure of the signaling process might lead to a loss of tissue function and the decline of reserve capacity that is typical of many aging organs (Shock 1983). Alternatively, as Raff (1992) points out, cells that can acquire the ability to

448 Chapter 12 Senescence as a Breakdown of Intracellular Regulatory Processes grow without relying on extracellular signals will give rise to tumors and the age-related pathologies associated with unrestrained growth. Fossel (2004) discussed the contribution that senescent cells, deficient in several aspects of signal transduction, make to the loss of function in aging adults. The converse of this argument has been put forth as a possible explanation for the amazing ability of embryonic stem cells to remedy defects in the adult host’s somatic cells. The stem cells may exert their effects not just because their descendants will repopulate the affected tissue, but mostly because the stem cells’ secretion of therapeutic growth factors triggers desirable changes in the gene expression patterns and the eventual phenotype of the neighboring host somatic cells (Chien et al. 2004; Fraidenraich et al. 2004). In this scenario, the cell-level phenomenon of aging is due to the attenuation of the normal signaling processes. Restoration of the signals restores (to some extent) the cell. This view provides us with a mechanism for the reversibility of aging. Both normal and abnormal growth are viewed as being under the control of a variety of intercellular signals, and this topic is discussed in more detail in chapter 13. The importance of the apoptotic process to normal aging is that it provides a controllable process that is clearly of some importance in regulating cell numbers. Apoptosis and mitosis are controlled by complex gene-based signaling systems. These systems can interact at the population and cell levels to bring about the net gain or the net loss of cells in the particular tissue. Organand organism-level physiological decrements can presumably begin to be observed either when the loss of cells begins to impinge on the tissue’s reserve capacity (Shock 1983) or when mutated cells begin to produce the signals that allow them to proliferate and to escape the bonds that restrain their normal neighbors and thus give rise to the cancers so prevalent in late life.

12.4.3 Death of Cellular Organelles: Autophagy Autophagy (i.e., self-feeding) is a third form of cell death, sometimes referred to as the type II form of programmed cell death, in which an autonomous process may result in the isolation and destruction of selected cytoplasmic organelles (see Shintani and Klionsky 2004 for review). It differs from apoptosis in the origin and mechanisms of formation of the autophagosome, the organelle that digests and recycles the selected intracellular targets, as well as in the details of the regulatory pathway regulating its activity. The autophagosome is formed from the cell’s own endoplasmic reticulum which blebs off, surrounds, and encloses the targeted structures to form a double-membrane vesicle. This vesicle then fuses with the cell’s own lysosomes, the enzymes of which then digest the cargo enclosed within the autophagosomic body. Autophagy seems to be controlled in part by the insulinlike signaling pathway, in that there is an interaction between the Akt/PKB step (see figure 7.32) and the TOR pathway (see figure 7.28), which initiates a side reaction that may result in the activation of autophagy. The data indicating an inverse relationship between the level of autophagy and the length of life suggests that there may be some sort of causal relationship between these two processes. One possible link may be that the increase in the cell’s resistance to oxidative stress, characteristic of longer-lived organisms, may come about via the autophagous removal of damaged mitochondria, and/or by the role of autophagy in regulating the rate of turnover of damaged DNA, proteins, or lipids. Another possible link is the demonstration that the autophagic mechanism may act as an innate defense mechanism and protect the cell against invading pathogens (Nakagawa et al. 2004).

13

Senescence as a Breakdown of Intercellular Regulatory Processes

13.1 Basic Assumptions All multicellular organisms face the problem of coordinating the activities of their different specialized cells and tissues. The problem becomes more pressing as the number and types of cells increase. The homeostatic mechanisms of the vertebrate body are controlled mostly via the neuroendocrine-immune (NEI) system. These intercellular signals exert their influence within the cell via intracellular signal transduction (ST) pathways, and I include here some details of ST not covered in the previous chapter. In this chapter I consider how alterations in the functioning of the NEI-ST system act to accelerate or retard the senescence-related loss of function. There is an enormous literature dealing with this broad topic; I present selected examples. The harmonious control of the body’s many functions depends on the efficacy with which the coordinating and integrating NEI-ST system uses its feedback loops and antagonistic processes to regulate these activities. On the basis of his long experience with the Baltimore Longitudinal Study on Aging, Shock (1983, p. 137) suggested that the “impaired effectiveness of these coordinating mechanisms may be the primary factor involved in aging.” Such coordinating mechanisms, or pacemaker tissues and organs, are not the only processes affected by senescence, as the prior chapters have shown. However, it is reasonable to consider the proper functioning of these integrative mechanisms as an extraordinarily important factor in organismic aging.

The importance of regulatory and homeostatic processes to the normal functioning of the adult was widely recognized some time ago, and their importance to the aging process has been reinforced by data collected in various long-term longitudinal studies. These integrative mechanisms can be affected by both internal and external sources of stress. The underlying assumption of such theories is that homeostatic failure of the organism plays a key role in the onset of senescence and death and that this failure may be traced to a more or less well-defined group of particular cells in each subsystem that have specific integrative functions. I also examine the effects induced when some stochastic insult brings about an inappropriate signal, which then shifts the entire NEI-ST system into a detrimental trajectory. The study of regulatory systems capable of integrating social and physical effects into wellknown biological mechanisms is a fruitful path to understanding how genetic and environmental factors interact in the normal aging process. Because this chapter emphasizes the integrative aspects of control mechanisms, it is appropriate to point out that modern biologists have moved beyond assembling a “parts list” of the molecules involved in the NEI-ST system and now views the molecules as constituting an interaction network (Bray 2003). The virtue of viewing these systems as networks is that it allows us to deduce the underlying functional architecture of the system by perturbing the highly connected nodes (glands, cells, molecules) and cataloging the effects observed in other nodes (target organs, receptors, pathways).

449

450 Chapter 13 Senescence as a Breakdown of Intercellular Regulatory Processes

13.2 Neuroendocrine Theories of Senescence In chapter 4 I briefly described species that die shortly after reproduction as a result of rapid senescent changes brought about by neuroendocrine mechanisms. But these senescent changes are not inevitable. Castrated salmon can continue living and growing for at least twice their normal life span. If Antechinus (the marsupial mouse) is castrated or socially isolated before mating, its life span is 2–3 years, as might be expected for rodents of its size. The normally rapid senescence of the marsupial mouse is not built into every one of its cells, as a consideration of only the insulinlike signaling pathway data might superficially suggest. Rather, senescence it is a direct consequence of alterations taking place in the neuroendocrine system. There is a huge literature on the neuroendocrine alterations that take place with aging, and the recent emphasis on trying to understand neurodegenerative diseases has stimulated even more interest in this area. It is impossible to adequately cover the entire field; instead, I focus on a few topics that illustrate if and how the loss of integrative ability might be associated with the course of senescence. Excellent reviews of this topic have been presented by Finch (1987), Finch and Landfield (1985), Merry and Holehan (1994a), Nelson (1995), Mobbs (1996), and G. M. Martin (2002). The proceeding discussion draws mostly from these sources. Mobbs (1996) pointed out that the neuroendocrine system operates in two different integrative domains, the homeostatic and the homeodynamic. The main function of the homeostatic domain is to determine and regulate average physiological set points (e.g., glucose levels), which are functions of the sensitivity of the neural components to hormonal signals. The main function of the homeodynamic domain is to regulate the temporal organization of physiological systems, largely by modulating the rhythms or pulses in these systems. This latter function is mediated mostly via the hypothalamus, the neurons of which are sensitive to circulating hormone

levels. Thus, alterations in either the number or the sensitivity of various neuroendocrine receptors could in principle give rise to the homeostatic or homeodynamic changes commonly associated with senescence. This analysis focuses on the role of the neuroendocrine system in reproductive aging and in the response to stress.

13.2.1 Reproductive Senescence Female reproductive aging has long been used as a model for the study of neuroendocrine senescence because it includes an early, obvious, and nonlethal failure of an important physiological system. Its cyclic nature provides a sensitive assay of functional status. The female reproductive system also offers the experimental advantage that it can be easily manipulated by means of surgical and/or hormonal treatments without major effects on other physiological systems. In humans, the median age at menopause is about 50 years and appears to be remarkably similar across various racial and ethnic groups (Merry and Holehan 1994a). Regardless of the age of entry into menopause, the menopausal transition takes about 6 years. The few populations that have an earlier median age of onset suffer the confounding effects of poor nutrition, health, and/or socioeconomic situations. Direct positive and negative effects of gonad and germline cells on life span have been demonstrated in invertebrate model organisms (see chapter 7). Do similar processes occur in the mammalian female as well? As described in chapters 5 and 7, the depletion of ovarian follicles in females of many species seems to be the primary endocrine-linked age-related change and might be the primary cause of various senescence-related syndromes. Follicular depletion is correlated with the simultaneous decrease of the plasma estradiol, increase of the basal luteinizing hormone (LH) to castrate (i.e., ovariectomized) levels, and impairment of the surge LH levels. Cells and tissues that depend on ovarian steroids such as estradiol and progesterone flourish at puberty and wither at menopause, for their function is tied to the hormonal ebb and flow. The fact that estrogen

13.2 Neuroendocrine Theories of Senescence

Frequency of estrus (cycles/mo)

replacement therapy can reverse these tissue hypoplasias shows that there is no problem with the estrogen receptor in the tissues and that these senescent changes may be due mostly to the lack of sufficient hormonal stimulation. These and other observations led Finch and colleagues (1984) to develop and test the hypothesis that the ovarian secretions have a cumulative effect on the pituitary and hypothalamus, which, when they pass a certain threshold, initiate senescent changes in these key regulatory components. Finch et al. (1984) were testing, in other words, the concept that aging is driven by certain key physiological events and is not propelled by the mere passage of time. When young ovaries were transplanted into old, postreproductive mice, the grafts failed to reinitiate the normal level of estrous cycles per month (figure 13.1; Finch et al. 1984). The fact that the young grafted ovaries had sufficient follicles and were capable of normal functioning if transplanted into young mice suggested that in the older mice there is a problem with the neuroendocrine cells of the pituitary and hypothalamus. However, when young ovaries were transplanted into old animals that had been ovariectomized when young and allowed to

Y

6

Y

4 Y

MA

2

0 0

4

8

12

16

20

24

Age (months)

Figure 13.1 The effect of age on the ability of an ovary transplanted from a young donor to maintain estrous cycles. When a young donor ovary was transplanted into a young (<15 months old) host female (Y→Y), the transplanted ovary functioned normally. If loss of function is due to host damage, later transplantation of a young ovary to a middle-aged (15–24 months old) cycling mouse (Y→MA) yields few if any additional cycles. Thus, a young ovary that can function well in young hosts cannot reactivate reproductive cycles in 16-month-old hosts. (After Finch et al. 1984.)

451

age without their ovaries, these animals had the ability to reinitiate regular estrous cycles. These results suggest that the ovary in situ normally has an adverse and irreversible influence on the hypothalamus and pituitary. This conclusion was supported by another series of experiments, in which reciprocal ovarian grafts were made between young and middleaged mice (figure 13.2; Mobbs and Finch 1992). If the middle-aged mice retained their own functioning ovaries until the young ovary was transplanted, their ability to maintain youthful estrous cycling was impaired. If the middle-aged mice were ovariectomized when young and then not further exposed to ovarian hormones until the time of the ovarian transplant, they were capable of maintaining youthful cycling. But if the middle-aged mice that were ovariectomized when young had also been given daily doses of estradiol since that time, their ability to maintain youthful estrous cycling was also impaired. Finally, when intact young mice were given daily doses of estradiol, their ability to maintain cycling was severely impaired. These results lead to the conclusion that reproductive senescence is the consequence of cumulative exposure to ovarian hormones and/or estradiol, in agreement with the data of figure 13.1. However, when ovaries from middle-aged intact mice were transplanted into young ovariectomized animals, the grafts displayed an impaired ability to maintain estrous cycling (figure 13.2). This result shows that the aging ovary has autonomous defects not present in young ovaries. This might be because older ovaries have fewer ovarian follicles and thus a decreased ability to synthesize estrogens and other compounds. Taken together, the data suggest that reproductive senescence arises from both ovarian and neuroendocrine impairments and that the neuroendocrine impairments are at least partly the result of the cumulative exposure to estradiol. Other data suggest that the ovary-induced neuroendocrine damage is done mostly during the period when the estrous cycles are becoming prolonged and irregular. In support of this hypothesis, Finch (1987) showed that many of the age-related changes that take place spontaneously

452 Chapter 13 Senescence as a Breakdown of Intercellular Regulatory Processes 30 (I)

Number of estrous cycles

25 (I) 20

15 (II)

(II)

(II)

10

(II)

5 (III) 0 Y

Y

Y MA Y MA Y MA Y MA MA Y MA Y (Intact) (LTO) (LTO+EV) (Int+EV) (Int+EV) (Intact)

Figure 13.2 The effect of the presence of ovaries and/or a single injection of estradiol (EV) on the ability of mice to support young ovarian grafts. Letters in parentheses indicate the state of middle-aged (MA) mice. Young (6-month-old) mice (Y) were either ovariectomized (LTO) or left intact, then given either 10 mg/g body weight of EV in oil or just oil. Six months later, ovaries from intact mice were grafted into young ovariectomized mice (MA→Y), and all MA mice were given young ovarian grafts beneath the kidney capsule (Y→MA). As controls, young ovariectomized mice were given young ovarian grafts (Y→Y). A bar represents the mean number of estrous cycles (+ SEM) that occurred after the ovarian graft in each group of six to eight mice each. Groups with the same Roman numerals (I, II, or III) differed (P > .05) by ANOVA followed by Neuman-Keul’s test. (After Mobbs and Finch 1992.)

in old rodents can be induced in young animals given sustained, high levels of estradiol for several months. When 5-month-old mice were ovariectomized, kept on high levels of estradiol for 3 months, and then given an ovarian graft, they responded with estrous cycles characteristic of old animals. In contrast, the controls, which were also ovariectomized but were not medicated with estradiol during the 3 months before they received an ovarian graft, responded with estrous cycles characteristic of young animals. This experiment not only proved the role of estradiol, but also allowed Finch (1988) and colleagues to quantify the relationship between the total amount of hormone exposure and the rate of aging of the reproductive system. Mice of the strain they used underwent a maximum of about 50 estrous cycles before they became nonreproductive—a state that is by definition the end point of aging in this system. Various calculations suggested that 5000 units of hormone—the amount of hormone needed to induce 50 estrous cycles—constitute the life-

time exposure of certain mice to estradiol and that irreversible damage to the hypothalamus and pituitary would follow once that exposure level was attained. In other words, the main aging event in this system was the organism’s attainment of a cumulative lifetime threshold level of 5000 units, the consequence of which was irreversible damage to the cells of the hypothalamus and pituitary. The mechanisms that cause the damage are not yet known, but there are several possibilities, including the regulation of gene activity patterns. If this hypothesis is correct, it should be possible to induce the same sorts of irreversible effects of aging by treating an animal with 100 units of hormone for 50 days, or with 20 units of hormone for 250 days, or with some other combination. The hypothesis suggests that aging, in the female rodent reproductive system at least, is the result of a quantitative dose response and that one should be able to dissociate aging from the mere passage of time (figure 13.3). This hypothesis

13.2 Neuroendocrine Theories of Senescence

453

Change (0/100) 100 Functional impairment Threshold Cryptic damage

Change (0/100)

50

100 10 0 50 Dose 10 0

Time Change = f(dose, time)

Life span

Figure 13.3 Age-related phenomena, represented on a three-dimensional surface whose axes represent time, dose, and change. This example is drawn from the effects of age and estradiol treatment on the reproductive ability of female mice. Estradiol is a cause of age-related change in the ovary-dependent neruoendocrine syndrome described in the text. Time is measured in months; dose is the total cumulative amount of estradiol administered to the mice, and change is the measured impairment in reproductive ability for each treatment. In many cases, the change may be cryptic (without overt functional consequences) until a threshold is reached (shaded background). Three trajectories at different doses are shown. (After Finch 1988.)

supports and is consistent with the idea that aging is an event-dependent process, a conclusion remarkably similar to those independently derived from studies of other systems in which it was established, for example, that the onset of cell senescence is related to the amount of oxidative damage the cell has undergone (see chapter 12). Those studies also indicate that senescence is an event-dependent but not a time-dependent process. These types of analyses have begun the process of converting time from an independent variable to a dependent variable in the analysis of aging. Of course, estrogen has many beneficial effects as well, some of which are mentioned below. A more accurate depiction of the concept embedded in figure 13.3 would take this into account, perhaps by plotting the net benefit or damage to the tissue, if such a metric could ever be worked out. Young ovaries transplanted into older (11 months old) hosts ovariectomized before puberty alleviated the symptoms of senescence otherwise present in the hosts and extended the host mean, but not maximum, life span (Cargill et al. 2003). This is essentially the same experiment as indi-

cated by the third bar in figure 13.2, but with life span being measured in addition to estrous activity. The increase in life span was not observed if the same treatment was done using younger hosts (see table 13.1 for details of the experiment). These findings can be interpreted in two ways, each of which is consistent with the invertebrate studies described in chapter 7. To recapitulate, the germline cells are thought to emit a signal that shortens life span, while the somatic gonad cells emit a longevity-extending signal. Removal of the entire gonad ablates both signals and has no effect. The first explanation of the mouse data is based on the observation that the transplantation process inevitably results in the loss of ovarian germline follicles by atresia as a result of the operation, and so the signal balance of a transplanted ovary is presumed to be tilted in favor of longevity extension. However, the signal is not strong enough to affect the robust somatic cells of the young female but is strong enough to affect the less robust senescent cells of the older female, and this effect is sufficient to revitalize them as indicated in table 13.1. An alternative but not contradictory explanation is that there is an optimal stage and dose

454 Chapter 13 Senescence as a Breakdown of Intercellular Regulatory Processes Table 13.1 Submaximal Exposure to Estrogens Enhances Longevity in Mice

Treatment

Life expectancy at time of treatment (months)

Intact control

19.9

Ovariectomized control Ovariectomized transplant at 5 months Ovariectomized transplant at 8 months Ovariectomized transplant at 11 months

18.0 19.2 18.3 20.9

Effect on survival curve and on estrus Continuous decline, some cycles in senescent phase standard values for qx Earlier onset of senescence, no effect on qx Earlier onset of senescence, no effect on qx, few cycles in senescent phase Early onset of senescence, some slowing of qx, and more cycles in senescent phase Early onset of senescence, most slowing of qx, and most cycles in senescent phase

Total time exposed to estrogens (months) 19.9 0.7 14.2 8.5 10.2

Source: adapted from data of Cargill et al. (2003).

level of estrogen at which its beneficial effects outweigh its harmful effects. Both the absence or presence of high levels of estrogen during the first year cause early onset of senescence and so must be harmful. The harmful effects of estrogen absence are likely related to the fact that estrogen inhibits the production of mitochondrial reactive oxygen species (ROS) in female rats (Borras et al. 2003; see table 8.5). Because ROS levels modulate the age of onset of senescence (chapters 7 and 12), ovariectomized females would be expected to enter senescence earlier, and they do. In addition, ovariectomized middle-aged rats suffer from a significant reduction in the capillary density in the frontal lobes when compared to intact controls (Jesmin et al. 2003). This decrease in blood supply is associated with marked decreases in growth factors and other proteins and may lead to increased neuronal apoptosis. As these data indicate, estrogen depletion is capable of bringing about the deleterious effects associated with senescence. The harmful effects of high estrogen levels are demonstrated in the data of figure 13.3. The presence of lower (optimal?) levels of estrogen in the second year of life of the transplant animals leads to positive effects on survival and estrus, with the magnitude of the effect being proportional to the amount of estrogen delivered to cells in the senescent portion of the life span (table 13.1). This quantitative explanation is consistent with the qualitative explanation dis-

cussed above, with the data of figure 13.3, with the Drosophila data showing that there is an optimal dose of ecdysone that has a longevityextending effect (see figure 7.26), and with the mostly beneficial effects of low-level estrogen replacement therapy observed in postmenopausal women. Human females who undergo an early natural menopause (at an age less than 44 years) have a significantly higher probability of dying early than do women who report a natural menopause at ages 50–54 years (Snowden et al. 1989). Snowden et al. did not investigate the role of estrogen at that time, and there are no data proving that it plays a role. Correlative data supporting a role of estrogen in longevity consist of the fact that longer lived women maintain active ovarian function longer than shorter lived women (Muller et al. 2002). Reproductive senescence may imply organismic senescence in one sex but not the other, as is suggested below. The explanation for such a difference probably resides in the details of the sex-specific life-history strategies of each species. Older men generally have lower concentrations of circulating testosterone, but these crosssectional studies did not allow for the exclusion of variables other than age as a contributory cause of the decline. A large-scale longitudinal study showed a progressive age-invariant decrease in both total and free testosterone levels beginning

13.2 Neuroendocrine Theories of Senescence

at about age 40 and continuing at a rate of 0.110 nM/l thereafter (Harman et al. 2001). The incidence of men with very low testosterone levels (i.e., hypogonadism) increased from 9% at age 50 to 91% at age 90, a tenfold increase. The definition of male hypogonadism is based on statistical but not on clinical criteria, and so it is difficult to obtain a precise view of the physiological changes from this decreased hormone level. It is also difficult to determine whether this gradual decrease in testicular function has any obvious effect on longevity. Because low testosterone levels are not suspected of bringing about an increase in mitochondrial ROS levels, as may be the case for estrogen in women (Borrass et al. 2003), such a direct effect is unlikely. There is no indication of an agerelated decrease in male reproductive capacity similar to that described here for females (see chapter 5). This might suggest that the utility of female reproductive senescence as a model of aging is quite limited because it doesn’t apply to half the members of the same species. Nonetheless, the study of female reproductive mechanisms has allowed us to uncover detailed interactions between the hypothalamus–pituitary axis and the target organ that are similar to those operating in other processes regulated by the neuroendocrine system, such as are described in the section on stress responses later in this chapter.

13.2.2 Gene Expression and Neuroendocrine Control Table 7.5 summarized the differences in gene expression patterns of the mouse hypothalmus and pituitary that are induced by caloric restriction. These changes probably play a key role in bringing about a coordinated shifting of bodily functions to a state compatible with increased somatic maintenance. Age-related changes in the gene expression changes of the hypothalamus– pituitary axis and the limbic system will be of great interest because they may give us insight into the changing regulation of the body’s physiological systems with age. Such gene expression changes will likely either be due to changes in peripheral hormone levels or to changes in the

455

trophic hormone levels. Early studies showed that age-related gene expression changes did occur in the brain, but the studies did not reach firm conclusions. As Johnson and Finch (1996) reported in their review of the early literature comparing mature versus old adult mouse brains that the mRNA levels of 20 genes showed age-related decreases, 14 showed no change or the results of different studies were contradictory, and only 3 showed consistent increases. Such results are difficult to interpret in the absence of any knowledge of the functions of the individual genes and the overall processes affected by their alteration. In the several years since then, the introduction of new high through-put techniques such as DNA microarray analysis has resulted in an explosion of new expression data. When these data are coupled with the extensive gene function (or gene ontology) data derived from the analysis of genome sequence data, some useful insights can be generated. The first impression from the data is the very large number of genes expressed in the brain, and the remarkable overall stability of the gene expression patterns. A comparison of about 7000 genes expressed in 6 brain regions in 2 different mouse strains revealed that only 24 of them were differently expressed in all 6 regions in both strains (Sandberg et al. 2000). Only about 1% of the genes were differently expressed in at least one of the regions. Can there really be so few differences in the brains within and between two strains, one of which is robust and the other which is weak, and in a structure considered to be the most complex in the solar system? The answer likely reflects the technical problem that our global assay techniques have a difficult time detecting small gene differences in a large pool. Restricting our examination to particular structures known to play an important role in the process under study will certainly restrict the pool size and probably uncover more informative observations. The more unpleasant aspects of senescence involve the loss of our physiological integration and its affect on our memory. Accordingly, I review data pertaining to the hypothalmic–pituitary– adrenal axis as well as to the hippocampus region of the brain.

456 Chapter 13 Senescence as a Breakdown of Intercellular Regulatory Processes The data of table 7.5 summarize the changes in gene expression with aging in the hypothalmus. These changes are consistent with a decrease in neuronal structure and synaptic activity, possibly derived from a increasingly inefficient energy supply and leading to decreased neural function. These are coupled with increased neural degeneration and compensatory up-regulation of mitochondrial energy processes (Jiang et al. 2001; Shimokawa et al. 2003a). An independent DNA microarray analysis detected 454 expressed genes in the hypothalmus, of which only 7 (1.5%) showed identifiable changes with age which were also consistent with increased oxidative stress and decreased neural function (Kappeler et al. 2003). This same study examined the expression patterns with age of 116 genes in the pituitary (Kappeler et al. 2003). Growth hormone (GH) mRNA accounted for 85% of the total gene expression in the pituitary. Only six genes in the pituitary (6%) showed identifiable changes with age. As might be expected, GH showed the largest decrease in the pituitary. It was unexpected that this decrease occurred even though the hypothalmic genes regulating GH expression showed no significant change. But, of course, it is not the mRNAs that are the important molecules in the NE system but rather the proteins. An examination of age-related alterations in the pituitary protein expression profile of mice over their life span led to the conclusion that only five known and seven unknown proteins varied with age (Marzban et al. 2002). Since more than 1000 reproducible proteins were assayed, this means that the small number (1.2%) of altered pituitary proteins is apparentely compatible with the small number (6%) of pituitary genes that show an agerelated alteration in expression. GH is present in several isoforms and declined with age as expected, but it is still the most plentiful protein present in the gland. The other four genes increased with age, and their identities are consistent with the presence of increased oxidative stress and cellular damage taking place in the older glands. The number of hypothalmic–pituitary genes and proteins showing age-related changes in expression is unexpectedly small for such important integrative organs. As with the overall brain study

mentioned above, this conclusion probably is an artifact arising from the somewhat insensitive statistical procedures routinely used in the field. However, the processes of increased cell stress and decreased cell function identified as occurring during the aging process are consistent with everything we know about aging. It is of some interest that these same processes are involved in the age-related loss of cognitive abilities. Given that information, then the question has to do with the mechanism by which these degradative processes in the NE system are initiated and whether they begin early or late in life. The hippocampus plays an integral role in memory formation and retrieval. Its normal functioning is central to cognitive abilities, for memory appears to be at the root of all problemsolving abilities. Older rats have age-related oxidative damage to their mitochondrial DNA (see chapter 11). The correlation between the age-related loss in cognitive function and the agerelated increase in oxidatively induced hippocampal damage is indicative of a causal relationship. Is there any other evidence to support this relationship? A well-designed study examined the gene expression changes taking place in the rat hippocampus as a function of both age and cognitive ability (Blalock et al. 2003). The inclusion of cognitive ability as a variable meant that one could directly tie gene expression changes to cognitive changes and thereby impart a robust meaning to the final results. Large numbers of rats of three different age groups were tested on two different memory-dependent behavioral tests. As expected, their performance on each test reflected an age-dependent impairment. The hippocampal A1 region of each animal was then dissected out and individually examined by microarray analysis. A sensitive and discriminating process of statistical analysis was used to first identify genes with significant expression changes and then to determine whether the gene in question was significantly correlated with changes in the individual animal’s age and/or changes in the individual animal’s behavioral test performance. Of the 4681 genes initially assayed, about 12% showed significant age-dependent changes. Again, this suggests that most of our genome has a rather stable expres-

13.2 Neuroendocrine Theories of Senescence

sion pattern. Functional information was available for 146 of these age-dependent genes. Only 11 of these were finally characterized as being highly significantly correlated with age but not behavior. In contrast, 135 were correlated with age as well as with the performance on one and/or both behavioral tests. Of these genes, approximately 40% were down-regulated and approximately 60% were up-regulated with age. What is more informative than the numbers is the identity of the cellular processes affected by these gene expression changes (table 13.2). The fact that pathways for energy metabolism, synapse formation, chaperone activity, extracellular matrix, and other processes supportive of neural activity are downregulated suggests that neural maintenance and memory-associated processes decline with age. The up-regulation of pathways involved in inflammation and oxidative stress, along with Ca2+ regulated signal transduction and myelin-related processes are informative, for they suggest that the hippocampal neurons are trying to compensate for severe oxidative stress by up-regulating a variety of compensatory processes affecting both neurons and glial cells. The above findings are generally compatible with those of table 7.5. What was most interesting about these results is that they showed that nearly all genomic alterations began before midlife, while many, particularly among the up-

457

regulated category, continued to advance between midlife and late life. One interpretation of this early start to loss of function is that the losses cannot be due to a widespread senescent deterioration of the brain in late life but probably represent a regulated response to early alterations of neuronal structure. The hippocampal network, in other words, undergoes a change in input parameters in the early/midlife phase. This induces a set of compensatory alterations in other systems that provide short-term relief but are incapable of long-term restoration of function, and so during the mid/late-life phase the system goes into a positive feedback spiral, leading eventually to impaired function and even cell death. A diagram of one possible model by which these empirical results could lead to such an outcome is depicted in figure 13.4. Note that the model involves at least two positive feedback loops, each of which leads to inherently unstable situations. Also note that the altered outputs of one subgroup of cells (e.g., myelin fragments from neurons) serve as inputs to another subgroup of cells (e.g., microglia and astroglia) and that these altered inputs bring about a qualitative change (e.g., autoimmune response against neurons) in the behavior of the affected subgroup, which induces a corresponding qualitative change (e.g., defensive reactions against oxidative stress) in the initial neuron subgroup. Note that the postulated changes

Table 13.2 Functional Processes Affected by Hippocampal Gene Expression Alterations Down-regulated

Up-regulated

Extracellularly regulated cell signaling Energy metabolism Transcription factors Biosynthesis (nucleus) Extracellular matrix Synaptic plasticity Chaperone

Inflammation/oxidative stress Growth and maintenance Protein processing Glial cell/structural Ca2+ regulated signal transduction Transcriptional regulation Lipid metabolism Myelin-related proteins Iron utilization Amino acid/monoamine metabolism

Source: after data of Blalock et al. (2003) Note: The table is based on the assay of those 146 genes significantly correlated with aging and/or cognitive performance on either or both of two memory-based behavioral tests.

458 Chapter 13 Senescence as a Breakdown of Intercellular Regulatory Processes

Altered Ca2+ Signaling in Neuron or Input from Cholinergic Neurons

Neural Activity

Neuronal Energy

Altered Extracellular Matrix

Neuronal Biosynthesis

Demyelination of Neuronal Axon

Regression of Axon

Synaptic Plasticity

Remylineation Process Activated in Oligodendrocyte

Autoimmune Rx Triggered in Glial Cells

Glial-mediated Inflammatory Cascade against Neurons

Glial Cell Hypertrophy

Oxidative Stress Response of Neurons

Impared Cognition, Neuronal Vulnerability

Figure 13.4 An integrative model of brain aging based on various experiments focusing on relating inflammatory damage to the rat hippocampus with cognitive impairment in the same individuals. Insults that result in a decreased neural input or altered Ca2+ signaling yield a decrease in neural activity. The resulting down-regulation of signaling pathways initiates a series of synthetic, genomic, and functional changes in the cells, resulting in increased neuronal vulnerability and impaired learning and memory performance. (Redrawn from data and figures in Blalock et al. 2003.)

disrupt neural activity but do not necessarily result in neural death, which is consistent with the fact that neuronal atrophy is not seen in the aging hippocampus (D. E. Smith et al. 1999). From a network perspective, the hippocampus system may be viewed as consisting of two major subroutines (neurons and glia), each of which is intimately connected to each other by reciprocal input/output links, and at least one of which (neurons) has important input links (perhaps Ca2+ metabolism) connecting them with the rest of the NEI network. These input links allow the system to exhibit considerable plasticity:

Network functions can be reset to new (more optimal or less optimal) levels by inputs from external factors not inherent in the hippocampal network itself. In the present case, this plasticity to (perhaps) Ca2+ metabolism brings about undesirable outcomes; I present data later in this chapter that illustrates the more beneficial aspects of hippocampal network plasticity to other different inputs. But the hippocampal network does not exist in a vacuum. It and four other brain areas (posterior cingulate cortex [PCC], bilateral inferior parietal cortex, left inferolateral temporal cortex, and ventral anterior cingulate cortex) are

13.2 Neuroendocrine Theories of Senescence

believed to constitute a default mode brain network that show correlated changes in metabolic activity when confronted with a cognitively simple sensory-motor task. Thus, the output links of the hippocampus must constitute some of the input links to these other regions comprising this default mode brain network and vice versa. Noninvasive imaging techniques allowed the analysis of brain activity in people when presented with various mental tests. Presentation of simple tests to normal individuals resulted in the activation of both the hippocampus and the PCC, but people with mild Alzheimer’s disease (AD) showed significant inability to activate their PCC (Greicus et al. 2004). This finding was interpreted as indicating a decreased connectivity between these two brain regions, and suggests the possibility of brain connectivity as a diagnostic test for AD. It is known that hippocampal damage precedes the dysfunction of the PCC, and this is consistent with the model (Grecius et al. 2004). But what causes the hippocampus to become destabilized? There was some indication that upsets in Ca2+ metabolism were responsible (figure 13.4), but this is likely to be secondary to some other insult. The hippocampus receives all its cholinergic innervation from the subcortical basal forebrain region, and these inputs are thought to play an important role in regulating neuronal activity in the hippocampus and other regions (Bartus et al. 1982). Deficits in the basal forebrain region have been correlated with memory deficits. It is plausible that this basal forebrain region is the source of some of the up-

stream insults that affect the hippocampus. These neurons undergo decreases in size but not in number as a function of age (D. E. Smith et al. 1999). At the time of examination, the neurons were atrophied but not dead. It is plausible that atrophied cells have a corresponding decrease in their cholinergic output signals to the hippocampus and other structures. It has been shown that treating the basal forebrain region of aged animals with autologous fibroblasts genetically modified to secrete human nerve growth factor (hNGF) can reverse atrophy and restore cell size, but not number (table 13.3; D. E. Smith et al. 1999). It would be interesting to know if the hippocampal function in the these old animals is restored by NGF treatment such that the events depicted in figure 13.4 either do not occur or are significantly attenuated. If so, this would confirm the suggestion by Blalock et al. (2003; figure 13.4) that the usual effects of aging on memory processes are not intrinsic to the hippocampus but stem from reversible interactions with other brain regions. In network terms, the decline in NGF in the basal forebrain subroutine leads to abnormal output signals to the hippocampus subroutine, which leads to the processes described in figure 13.4 and thence to the degradation of hippocampal outputs, which are part of the preferential input data for the PCC region. Loss of these inputs results in the inability of the PCC to activate itself in coordination with the hippocampus. The network suffers from localized damage. The increasing lack of coordination between these two (and other) subroutines composing the default mode

Table 13.3 Effect of HNGF on Number and Size of Cholinergic Neurons Upstream to the Hippocampus of Rhesus Monkeys Treatment Non-aged Aged intact Aged control graft Aged hNSF graft Probability

459

Mean age

Mean ± SD cell number

Mean cell size

9.4 25.1 23.3 22.6

60,300 ± 3,400 57,200 ± 3,900 55,000 ± 5,600 53,100 ± 2,400 0.62, NS

496 mma 449 mm 435 mm 484 mmb a vs. b = NS

Source: after D. E. Smith et al. (1999). Note: hNSF, human nerve stimulating factor.

460 Chapter 13 Senescence as a Breakdown of Intercellular Regulatory Processes brain network may result in a lack of appropriate stimulation to the neurons of these other centers. In that case, we may find that those regional subroutines will also suffer if the lack of stimulation results in the construction of a positive feedback loop, similar to that shown in the top portion of figure 13.4. The network suffers from diffuse localized damage, which has the effect of isolating the subroutines from one another. Those complex outputs of the NEI network, which are dependent on coordinated operations of different subroutines, will become increasingly defective. Physiological functions dependent on those outputs will be increasingly affected. The system will eventually crash. The Nun’s Study (Snowden 1997) made clear the importance that early and sustained neural activity has on the maintenance of one’s cognitive abilities (see chapter 5.9.6.4). If one is not mentally active all through life, then one might not have built up the reserve capabilities that will sustain one against the depredations of time. Not every change in gene expression involves quantitative alterations in gene expression. Some appear to involve unusual changes in the cell’s biosynthetic mechanisms. A fascinating example is the homozygous mutant Brattleboro rat, whose genetically based diabetes insipidus is due to a frameshift mutation in the vasopressin gene. As the animal ages, the expression of the wild-type vasopressin molecule gradually increases during the first 79 weeks of life, even though the animal is a homozygous mutant. This paradoxical situation appears to be due to frameshift deletions at two hot spots in the gene. Even more interesting is the observation that infusion of the animal with wild-type vasopressin during a 40-week period reduced the incidence of the deletions. It is unlikely that the vasopressin DNA is changing in any way. Instead, it is thought that the RNA splicing machinery of the cell is able to generate the wildtype sequence downstream of each mutation site by splicing together undamaged portions of the mutant mRNA into one functional mRNA molecule (Evans et al. 1994). It is also likely is that the vasopression levels function in some sort of a physiological feedback scheme that somehow enhances the mutant cell’s splicing ability so as

to make more wild-type spliced molecules downstream of the original mutation. If such a process actually does occur in this one aspect of the NE system, then it might well occur in other parts of the system as well. This example cautions us to consider that alterations in the cell’s posttranscriptional mechanisms can also yield significant changes in the NE proteome, which could affect the onset or duration of senescence.

13.2.3 Stress Responses The organism’s ability to respond to stress represents a major portion of its ability to adapt to environmental changes. The hypothalamic–pituitary –adrenal (HPA) axis is responsible for the neuroendocrine response to stress via the adrenocortical secretion of glucocorticoids. These steroids mobilize energy from storage sites, increase cardiovascular tone, and suppress various other functions that are not required during an emergency—the well-known fight-or-flight response. Even though a large variety of stimuli can initiate a stress response, the salient fact is that all these stimuli induce a complex but stereotypic response. The animal’s response takes place on both the neuroendocrine and the molecular levels. Each level depends on accurate detection of feedback signals for its effective functioning. The same adaptive features of acute glucocorticoid action can give rise to various pathologies when prolonged stress results in a chronic excess of glucocorticoid levels. Because senescence is often thought to involve a loss of the ability to retain homeostatic balance in the face of stress, an investigation of the effects of age on the HPA axis might shed some light on whether the degenerative aspects of stress and senescence each act to accelerate the other. The answer to the question is species-specific: In rats, the endocrine dysfunctions noted in the HPA axis may lead to observable functional abnormalities in the animal; in humans the HPA abnormalities are much more subtle and lead to an increased vulnerability but not necessarily an observable abnormality (Sapolsky 1990). These findings illustrate the need for caution in extrapolating findings from one species to another.

13.2 Neuroendocrine Theories of Senescence

The aging rat has the ability to start a normal stress reaction to any of various standard experimental stresses, but it is impaired in its ability to terminate the secretion of glucocorticoids. The hippocampal region of the brain is rich in glucocorticoid receptors and is thought to be responsible for mediating the negative feedback portion of the HPA axis. During aging, there is a preferential loss of mRNA for the type II glucocorticoid receptor in the rat hippocampus, with little loss observed elsewhere in the brain (van Eekelen et al. 1991). One experiment demonstrated that the cumulative exposure to stress-induced glucocorticoid secretion destroys specific hippocampal neurons, particularly in older animals (Landfield et al. 1980; but see Wickelgren 1996 for arguments that hippocampal neurons may undergo severe atrophy but not cell death). The observation that glucocorticoids atrophied or killed hippocampal neurons led to the idea that this neuronal loss desensitizes the HPA axis to glucocorticoid feedback regulation, producing further glucocorticoid secretion, which leads to further neuronal death, and so on (Sapolsky et al. 1986). This self-amplifying destructive feedback cycle appears to underlie the dramatic senescence of male marsupial mice (Antechinus), which die after a brief and aggressive mating period (see chapter 4). In this organism, the negative feedback system fails at the neuronal level, and the glucocorticoid concentration increases unchecked. The hormonal hypersecretion induces a fatal array of symptoms reminiscent of Cushing’s disease in humans, except that the reproductive ability of Antechinus is unimpaired. A similar situation is observed in the Pacific salmon (see chapter 4). However, this catastrophic feedback cycle may not be operative in every rat (see “Other Hormonal Phenomena,” later in this chapter) and is certainly not present in humans (Mobbs 1996). In people, both the hormone levels and the stress response show no significant alterations through age 80 years. After that age, some evidence suggests that humans show an increased glucocorticoid level and an impaired feedback process, although much milder than the situation in the aged rat (Sapolsky 1990). However, no evidence shows that the same mechanisms are involved. In

461

fact, the chronic use of glucocorticoids in current medical practice means that millions of patients have been treated with large doses over long periods with no reported adverse effects indicative of hippocampal damage (Goya et al. 1995). The molecular response to stress, such as the chaperone protein gene expression in various tissues of the animal, is modulated by the neuroendocrine system (Udelsman et al. 1993). Transplantation experiments have shown that young tissues put into an aged host have an “old” type of stress response, while old tissues put into a young host have a more nearly normal stress response (Udelsman et al. 1995). This demonstrates that the changes in the stress response with age are the result of prior changes in the neuroendocrine system. These neuroendocrine effects on the molecular chaperone molecules clearly have an effect on the organism’s transition from the health span to the senescence span, as was discussed in chapter 9 (see figure 9.5).

13.2.4 Factors Modifying an Individual’s Neuroendocrine Status and Rate of Aging The damages seen in the neuroendocrine system of older animals are not inevitable. The aging of the adult can be modified in a positive or negative direction by various developmental experiences. One example of a positive effect is that handling young, immature rats induces persistent changes in the HPA axis such that these animals when older do not have as severe a hippocampal neuron loss or learning impairments as do the nonhandled controls (Meany et al. 1995; Sapolsky 1990). In one study, the treated rat pups were handled in a defined manner for only 15 minutes a day during the first 3 weeks of life. After that, the handled and nonhandled animals were raised under similar conditions until maturity. When tested with various stressors, the handled adults secreted glucocorticoids at levels lower than those of nonhandled animals and were more effective in rapidly shutting down the stress response once the stressors were removed (figure 13.5). Relative to nonhandled controls, the handled adults also showed increased exploratory behavior, less

462 Chapter 13 Senescence as a Breakdown of Intercellular Regulatory Processes

50

40

30

20

10

* 0

STRESS

0

RECOVERY

1

2 Time, hour

3

4

Figure 13.5 Corticosterone titers in young and aged rats during 1 hour of immobilization stress, followed by 3 hours of post-stress recovery. Note that old rats (light gray symbols) do not return to the baseline as do young rats (dark gray symbols), and so are subjected to chronic corticosterone stress. (After Sapolsky et al. 1984.)

anxiety in unfamiliar situations, a reduced fear of new foods, and a reduced suppression of the immune system (Heuser and Lammers 2003). Even in the absence of stress, the aged, nonhandled animals had a higher basal level of glucocorticoids than that of their handled sibs, and this increased hormone secretion was accompanied by higher levels of the trophic hormone-releasing factors necessary to stimulate the adrenal gland. The whole HPA axis of nonhandled rats showed elevated activities. This resulting higher glucocorticoid level was estimated to be the functional equivalent of undergoing 12 hours of mild stress every night (Meany et al. 1995). If hippocampal neurons can be killed or atrophied by prolonged exposure to glucocorticoids, and if the nonhandled animals have higher cumulative exposure to these hormones than do handled animals, then one would predict that the nonhandled animals should have higher levels of hippocampal dysfunction than the handled ones. The pathway by which glucocorticoids exert their harmful effects on the hippocampal neurons may or may not be similar to the model of figure 13.4, but the two must be similar in that the detrimental effects must start in the early–midlife period, and they must exert a cumulative effect on out-

put of these neurons. These predictions are upheld, since the aged (24 months old), nonhandled animals have a 40% loss of hippocampal neurons than that of the aged, handled animals (Meany et al.1995). These neuron losses have functional consequences, for the aged nonhandled rats do much poorer in tests of spatial memory than do aged handled rats, and such cognitively impaired adults exhibit elevated basal and stress levels of glucocorticoids, as well as a significantly greater loss of glucocorticoid receptors in the hippocampus (Meany et al. 1995). Just as exercise alters NE gene expression in the rat brain, so the organism’s developmental experience modulates its hormonal response to stress. It is likely that chronically stressed animals will have a significantly different trajectory of senescence than will unstressed animals. There are also some developmental experiences that have negative effects on the eventual aging of the adult. When nonhandled, adult, pregnant rats are subjected to daily restraint stress during the last week of their pregnancy, the offspring exhibit impaired learning and a chronic alteration of their NE system when tested as adults (Lemaire et al. 2000). These offspring have relatively larger adrenal glands throughout their

13.2 Neuroendocrine Theories of Senescence

life span (table 13.4). An increase in the adrenal gland/body weight ratio is considered indicative of chronic glucocorticoid overload. When grown, these offspring exhibit a significantly higher error rate and slower learning rate upon testing with a standard type of spatial learning/water maze problem. These same animals also exhibited a substantial reduction in the rate of neurogenesis beginning at maturity and continuing thereafter (table 13.4). Hippocampal-mediated learning depends on the generation of new cells which arise in the dentate gyrus, migrate to the granule cell layer, and differentiate into neurons whose projections extend to the CA3 hippocampal region (Carlen et al. 2002). Such newly formed cells functionally integrate themselves into the existing neural circuits. Neurogenesis in adult rats is inversely proportional to the glucocorticoid levels. Maternally induced prenatal stress can alter the fetal NE system such that the HPA axis shifts to a state of future chronic glucocorticoid overexpression. This state has no apparent effect on the pups but becomes apparent in young adults and thereafter (table 13.4). This increased HPA activity suppresses neurogenesis in the dentate gyrus and subsequently alters the animal’s spatial cognitive abilities. Developmental experiences can bring about nongenetic transmission of the ability to handle stress and of the types of senescent morbidities expressed (Champagne and Meany 2001). I have already discussed the effects of maternal nutrition in humans on the child’s eventual senescence and morbidity (see figure 3.14). A substantial proportion of unhealthy age-related

morbidities in rats and humans may in fact be developmental or familial, not genetic, in origin. Pre-natal and postnatal developmental experiences can alter the senescent trajectory for better or worse. Both effects testify to the plasticity of the NE system and perhaps of the brain as a whole. Is this plasticity limited to the young animal? The decline in learning ability commonly observed in old rats has been viewed as ultimately due to the decreased input activity of the basal forebrain cholinergic neurons to the hippocampus (Bartus et al. 1982). This in turn is thought to be the result of a decreases in NGF activity. Applications of NGF via intracerebroventricular administration led to a reversal of the age-dependent atrophy of the cholinergic neurons and a significant improvement in spatial memory retention in 18- and 24-month but not 30-month-old female rats (Fischer et al. 1991), and this work has been repeated in other systems (Rose 1999). This suggests that the hippocampal neurons retain sufficient plasticity to react to changes in their input variables throughout most of adult life. The hippocampus will also react to environmental factors. Rats raised in an enriched environment show higher levels of neurogenesis, an increased complexity of arbors and synapses in the hippocampus and cerebellum, and an enhanced memory and learning ability (see Mattson et al. 2001b for references). The synaptic density usually decreases in an age-dependent fashion in rats, but maintenance of the adult animals in an enriched environment prevents synaptic loss and improves the outcome after an induced stroke.

Table 13.4 Effect of Prenatal Stress on Presumed Chronic Hypothalmic–Pituitary– Adrenal Stress and Hippocampal Neurogenesis Age (months) Variable Adrenal weights Prenatal stress Control Stress/control ratio Neurogenesis Stress/control ratio

1

3

10

22

297.9 ± 12.91 260.5 ± 7.74 1.15

159.4 ± 6.10 105.4 ± 8.51 1.51

101.0 ± 14.24 64.7 ± 1.49 1.55

nd nd —

1.0

0.86

0.90

0.75

Source: after Lemaire et al. (2000).

463

464 Chapter 13 Senescence as a Breakdown of Intercellular Regulatory Processes These findings suggest that the enhanced neuronal reserve induced by the enriched environment attenuates neurodegeneration and/or enhances recovery. Physical activity also affects the rat brain, for neurogenesis is improved by running on an exercise wheel. Nutritional changes, including the incorporation of antioxidant-rich blueberry or spinach extracts (see table 10.7) into the diet of older rats, can inhibit or even reverse the animal’s age-related loss in spatial memory (Galli et al. 2002). Caloric restriction implemented in middleaged and older animals still yields a beneficial effect (see figure 6.3). The beneficial effects of enriched environment, physical exercise, and caloric restriction on the brain are mediated by various trophic growth factors, and the various animal studies suggest that these interventions can benefit cognitive function to some degree in rats throughout most of their life span. Does a similar sort of plasticity occur in humans? The animal studies have focused on spatial memory, which is rooted in the hippocampus, but primate behavior depends also on “explicit” or “declarative” memory, which is based in the frontal cortex and is not addressed by the rat studies. In addition, spatial memory processes may be somewhat different because the rodent and human HPA axes differ in enough aspects so that one must translate, and not just transfer, data from the shorter lived rodent to the longer lived human. Yet there are some interesting similarities. In general, increased HPA activity is not normally associated with aging in humans. But increased glucocorticoid levels are associated with certain aging-related neuropathologies such as Alzheimer’s disease or Cushing’s disease. In the Cushing’s disease, increased glucocorticoid levels were shown to be associated with impaired cognitive function. Among normal aged (70-yearold) humans, patients that were apparently normal but had elevated cortisol levels performed worse on some learning tests than did patients with low cortisol levels (Meany et al. 1995). The existing data on humans thus are generally consistent with the rat data, but less extreme in their expression. Clinical studies have adduced a number of links between early life events and adult function.

Childhood abuse, neglect, and other dysfunctional familial behaviors increases the risk that the children will suffer from depression, post-traumatic stress disorder, or major anxiety disorders as children (see Heuser and Lammers 2003). Some clinical evidence supports the involvement of the HPA axis in bringing about these syndromes, although the data suggest that this may arise in humans from a chronic adaptation (i.e., increased sensitivity) of the HPA system rather than from a higher than normal level of cortisol (the adrenocortical hormone in humans). Depressed people have lower levels of neurogenesis, and clinical studies suggest that many antidepressants may act via the normalization of an initial HPA dysregulation. Finally, there are a host of documented cognitive and behavioral differences observed in adults who as children were raised in deprived versus enriched environments. The brains of both rodents and humans can be affected by physiological and environmental variables, and many of these seem to act through the HPA system. Although there are significant differences between the two, we would be foolish not to pay attention to the obvious similarities regarding NE plasticity, especially when formulating best-practice child-rearing advice. For example, if rat pups have improved cognitive functions and age more successfully as a result of a total of only 5.25 hours of handling during their lifetime, what implications does this observation have for the role of social phenomena in modulating adult function and the trajectory of senescence in humans? Babies raised in orphanages where their physical needs are met but their emotional and/or social needs are neglected often show an abnormal developmental pattern called “failure to thrive.” Humans are more complex than rodents, and so there might be some logical opposition to a mechanistic interpretation of these phenomena, coupled with a desire to base human practices on human evidence. Much of what is called “successful aging” may depend on our intrauterine and developmental history, our educational levels, our diet, and our physical activity, as well as on our genetic heritage (e.g., see figure 3.14). Based on the animal and human studies to date, it is probable that our genes do

13.2 Neuroendocrine Theories of Senescence

not act in a vacuum but rather in an environment indirectly constructed by these other variables. How would one obtain the data in future necessary to prove this hypothesis without doing irreparable harm to innocents? The directive to “first, do no harm” is of value here. There is no evidence that children are harmed by being raised in a loving manner in an enriched environment, and so perhaps we should err in that direction when we translate the existing data.

13.2.5 Other Hormonal Phenomena 13.2.5.1 Dehydroepiandrosterone

Ovarian steroids are not the only hormones that exhibit major decreases with age. The hormone dehydroepiandrosterone (DHEA) is produced by the adrenal gland and (in the sulfate form) is the most abundant steroid in the blood of humans. It is the only adrenal steroid that shows a strong agedependent alteration in its concentration, suggesting that its synthesis is regulated by changes in the responsible P450 enzymes other than those known to regulate glucocorticosteroids (Nawata et al. 2002). In humans, DHEA increases at puberty, reaches a peak level in the late teens or early twenties, and then steadily decreases such that older adults show a 90% decrease relative to the peak values (Orentreich et al. 1984). DHEA is one of the few mortality-tested longevity biomarkers available in humans and calorie-restricted primates (see figure 3.21), and it is presumed to play an important role in maintaining homeostasis. There is good reason to believe that DHEA-S (sulfate) inhibits the secretion of corticosteroids, which inhibit immune activity. Thus it is possible that the correlation of higher DHEA-S levels with extended human longevity may be an indirect one, actually reflecting the role of enhanced immune function on longevity (see figure 3.21). The loss of DHEA is not correlated with cognitive decline in older men (Moffat et al. 2000). A variety of other beneficial effects have been reported for this molecule. It cannot have a direct endocrine effect because no specific receptor for DHEA has yet been found. However, it has been suggested

465

that it may act as a prohormone and undergo intracellular conversion to a variety of active androgens or estrogens (Orentreich et al. 1994). Evidence of such intracellular conversions has been reported (Nawata et al. 2002). A third possibility is that DHEA may bind to some important enzymes and alter their activity. DHEA is an uncompetitive inhibitor of the enzyme glucose-6phosphate dehydrogenase (G6PDH) and thereby reduces the formation of NADPH, which, in turn, is thought to reduce the generation of ROS by inhibiting the activity of NADPH-dependent oxidases (Schwartz and Pashko 2004) Free-living baboons show a similar pattern of DHEA concentrations as seen in humans, providing the possibility of a primate model to eventually test the role of DHEA in aging (Sapolsky et al. 1993), and this has been verified by the caloric restriction/biomarker studies described in chapter 3. Some rodent strains do not have detectable levels of DHEA. Thus, the observation that shortlived mutant mice suffering from obesity, diabetes, and tumors showed improvement in these conditions, as well as an extended life span, after treatment with DHEA (Coleman et al. 1984) is difficult to interpret, especially in view of the report that lifetime feeding of DHEA to a healthy low-tumor strain of mouse (C57BL/6J) altered neither median nor maximum life span, nor affected the rate of aging (Orentreich et al. 1994), although DHEA has been reported to improve immune function (R. A. Miller 1996). The latter finding is consistent with the ability of DHEA to inhibit corticosteroid secretion, as discussed above. 13.2.5.2 Caloric Restriction, Stress, and Neuroendrocrine Hormones

In Chapter 6 I reviewed the evidence demonstrating that caloric restriction (CR) is a highly effective method, particularly in mammals, of slowing the rate of aging and increasing longevity. The data of chapter 7 also suggested that CR triggers the insulinlike signaling pathway (ISP) to alter the animal’s gene expression patterns and physiology. CR also increases the resistance of neurons to a wide variety of insults and, in this manner, it maintains plasticity and delays the onset of cognitive

466 Chapter 13 Senescence as a Breakdown of Intercellular Regulatory Processes loss. What specific mechanisms account for this increased neural resistance to stress? CR decreases the amount of energy available to the cell and alters the gene expresson patterns of the neurons. The genes altered in the brain fall into several categories but mostly include the heat shock or cellular stress proteins and various neurotrophic factors. Some evidence suggests that neurodegenerative processes may begin with the onset of alterations in the synaptic terminal which activate harmful apoptotic and excitory cascades. The postsynaptic complex is composed of a network of about 100 interacting proteins and is thought to play a key role in cognition (Grant 2003). For example, subtle alterations of hippocampal synaptic function are among the earliest signs of impending Alzheimer’s disease, and these alterations may be due to ROS-inducing Ab42 oligomeric proteins diffusing into the synapse area (Selkoe 2002). Brain synaptosomes prepared from rats on a CR regime, or from rats fed CRmimetic drugs, contain higher levels of cell stress proteins (e.g., HSP70, GRP78) than controls and are more resistant to various metabolic and oxidative insults (Mattson et al. 2001b). A localized increase of these neuroprotective proteins may maintain the neuron’s ability to resist injury, such as the increased Ca2+ flux of figure 13.4, and otherwise resist the effects of senescence. Higher levels of neurotrophic factors play an important role in maintaining the neuron’s plasticity. CR increases the level of brain-derived neurotrophic factor (BDNF), and BDNF is known to increase the survival rate of newly born neural cells. Raising animals in an enriched environment, as discussed above, is believed to bring about the same or similar changes in these neuroprotective molecules as does CR. Earlier in this chapter, I discussed the harmful effects of high glucocorticoid levels on the hippocampal neurons. CR induces low-level increases of glucocorticoid, which may have a beneficial effect on the organism. This hormesis effect was first suggested by Masoro (2002). CR also decreases leptin levels. Leptin is a hormone secreted by the adipose cells that signals their nutritional status to other body systems. Leptin also normally inhibits glucocorticoid syntheses; thus,

decreased energy intake results in the indirect leptin-mediated low-level activation of glucocorticoids. Any detrimental effects of this intial stress respone are thought to be outweighed by the beneficial effects of the increased cell stress proteins and neurotrophic factors (Bernal and Stern 2004; Patel and Finch 2002). Given the important role that the endocrine glands play in regulating neuroendocrine responses, one might initially conclude that the loss of function of any single gland has deleterious effects on the organism’s physiological status. However, the removal of the anterior pituitary (hypophysectomy) appears to maintain good health status in rodents at ages when control rodents appear senescent. Table 13.5 summarizes the reported effects of this treatment (Finch 1990). This alteration in hormonal stimulation resulting from the surgical removal of the gland or the genetic disruption of its function results in a delayed onset of senescence in the treated animals due to the significant reduction in insulinlike growth factor-1 levels, as discussed in chapter 7. Finally, it should be noted that age-related changes have been described in neural structures other than those few described here. If we regard our essence as being our mind and memories, then what we have learned suggests that the physical substrate that constitutes us is constantly changing us even as we change it. Some of these changes must adversely affect the integrative activities of our body and set in motion a chain of events that eventually culminates in our inability to surmount some challenge that we once could easily have overcome.

13.3 Immunological Theories of Senescence 13.3.1 Immunosenescence and Aging In chapter 5 I described some of the age-related changes that take place in the human immune system. There are actually two bodily systems conveying immunity to infection. Innate immunity is the phylogenetically older system and uses

13.3 Immunological Theories of Senescence

467

Table 13.5 Hypophysectomy and the Retardation of Senescence Trait

Improved by hypophysectomy

Worsened by hypophysectomy

Collagen

Tail-tendon tensile strength

Immune

Thymus weight, delayed hypersensitivity, graft rejection, in vivo phagocytosis, primary immune response

Kidney disease

Proteinuria, thickened basement membrane

Metabolism

Responsiveness of O2 consumption to thyroxine, O2 consumption in vivo, liver RNA synthesis

Neuropathy

Hind-limb paralysis via spinal neuron degeneration

Vascular

b-adrenergic mediated relaxation of smooth muscle, aortic thickness

Tumors

Reduced incidence in endocrine glands and other organs

Wounds

Healing improved

Primary immune response

Source: from table 10.9 of Finch (1990).

proteins encoded in the germline to recognize carbohydrate components on the surface of foreign invaders. The intestinal epithelium, for example, guards against the incursion of gut microflora into the rest of the body by means of the innate immune system’s macrophages and dendritic cells, which can recognize and defend against specific pathogens newly encountered by the host (Kraehenbuhl and Corbett 2004). In Drosophila, and probably in other organisms as well, the effectiveness of the innate immune system in containing bacteria seems to be directly related to the existence of particular polymorphisms in a number of different innate immunity genes (Lazzaro et al. 2004). Adaptive immunity is found only in vertebrates and can generate a huge number (~1011 different clones) of novel B and T lymphocytes that can recognize both protein and nonprotein antigens. These lymphocytes depend more on combinatorial permutation of existing genes for their potency than on the appearance of new heritable genetic polymorphisms. Excellent reviews of adaptive immunity have been published (R. A. Miller 1995, 1996, 2001b). Longitudinal studies of the age-related changes that take place in laboratory primates such as the macaque monkey suggest that the numbers and types of lym-

phocytes change with age, and these changes may be responsible for the defective maintenance of immunological memory with age (Bowden et al. 1994). Some of these senescent changes may be autonomous to the immune cell; most are not. The conventional belief is that these decreases in the quantity and quality of the immune response are due in part, directly or indirectly, to the early involution and senescence of the thymus. The thymus seems to be involved in regulating the quantitative extent of an immune response. The thymus also appears to determine early in life the qualitative array of immunological responses by the T cells that any individual can mount. After the involution of the thymus, an age-related change affects the extent to which the immune cells can proliferate in response to an appropriate stimulus: The spleen or lymph node or blood cells of older animals respond to a lesser extent than those of younger animals. This decreased vigor of the immune response seems to be due to a dramatic age-related shift in the proportion of naive and memory T cells (R. A. Miller 1995). T lymphocytes that leave the thymus and reach the peripheral immune system have not yet been immunologically triggered by a specific antigen, so they are referred to as “naive.” Once the cell has been triggered and

468 Chapter 13 Senescence as a Breakdown of Intercellular Regulatory Processes proliferated, some of the progeny cells return to the resting state and are called “memory” cells. Such memory cells enable the organism to mount a rapid secondary response if the same antigen is encountered later. However, the involution of the thymus puts a cap on the number of naive cells in the immune system, and the continual activation of new memory cells implies that the proportion of naive cells must decline—which it does. This shift in T cell types is thought to account for some of the age-related decline in immune function. This cannot be the only explanation, because there is also an age-related decrease in the ability of activated T cells to secrete and respond to lymphokines (such as interleukin-2 [IL-2], which is required for growth of both T cells and B cells), and this decrease is accompanied by an increased ability to produce interleukin-4 (IL-4; R. A. Miller 1995). The data suggest that the T cells are functionally subdivided into several different types with different synthetic capabilities. Thus, the age-related decline in T cell responsiveness is due not to a progressive change affecting every cell but rather to a change in the proportion of cells that can respond versus the proportion that fail to respond at all. The suggestion has been made that the immune system should be viewed as a mosaic of responsive and hyporesponsive cells in which the latter become more frequent with age (R. A. Miller 1995). Although multiple factors may be involved in immunosenescence, much attention has been paid to the progressive age-dependent involution of the thymus. Thymic involution precedes or accompanies the usual age-related functional changes seen in the immune system and thus might logically play an important role in bringing about immunosenescent changes. Structural changes in the mouse thymus begin quite early. Subtle signs of atrophy of the lobules and involution of the organ can be seen in the 6-month-old animal, as can a slight reduction in the staining of cortical thymocytes (Takeoka et al. 1996). Substantive changes begin later, however. The cortex and medulla of the thymus in the 12-month-old mouse have separated from each other and contain larger numbers of atrophying cells and other signs of structural degeneration (Takeoka et al. 1996).

The cortical cells of the thymus seem to be primarily affected at first; the cells of the medulla appear normal until the later stages of involution (Nabarra and Andrianarison 1996). Macrophage numbers increase and increasingly show signs of phagocytic activity (see chapter 11). By 18 to 20 months, the organ architecture has disappeared, and the number of lymphocytes has decreased drastically. By 25 months and later, the terminally involuted thymus is much reduced in weight, has lost the cortical–medullary separation as well as the presence of multiple differentiated cells, and is composed mostly of adipose cells and connective tissue. The maintenance of immune function until relatively late in life suggests that the small number of thymic cells remaining in the old organ are immunocompetent (Nabarra and Andrianarison 1996). Nabarra and Andrianarison concluded, from the correlation of thymic involution with age-related immune dysfunction that the thymus should be regarded as the aging clock or pacemaker of the immune system. They also suggested that the involution of the thymus is not intrinsic but is linked to an extrinsic cause, probably in the neuroendocrine system. This suggestion has been made by other authors and is still being vigorously debated. I explore it in some detail here. The age-related degenerative changes that take place in the normal mouse thymus are not identical to those seen in animals suffering from autoimmune disease, suggesting that the loss of function in these two syndromes arises from different causes and that autoimmune models are not completely equivalent to the situation in normally aging animals. The normal structural changes observed during thymic involution are accompanied or even preceded by changes in the functional capacity of the thymus. The mouse thymus shows a sharp drop in its ability to promote T cell differentiation within the first month of life but remains steady thereafter (Utsuyama et al. 1991). Removing the thymus from adult mice has a less obvious effect on cell-mediated immunity, possibly because of the continuing existence of spleen and lymph node lymphocytes. The exact role of the thymus in mediating adult immune function still needs to be worked out.

13.3 Immunological Theories of Senescence

Nonetheless, the composition of the T cell subpopulations induced by transplanted thymus does change with age, reflecting the changing capacities of the gland to support and sustain the peripheral immune system, and this change probably has functional significance. Another system in which to explore the role of the thymus in immunosenesence would be one in which the thymus either does not develop or does not degenerate. The Buffalo/Mna rat presents one such “natural” experiment in which the thymus does not involute but continues to grow with age, so it presents a useful model with which to explore the role of the thymus in aging. The Buffalo/Mna rat dies at about 20 months as a result of thoracic compression resulting from the thymic overgrowth. The animal’s immune functions are well maintained in middle and old age. Hypophysectomy performed on 8- to 9-monthold mice gave rise to animals with larger thymuses and higher ratios of cortex–medulla areas than the age-matched controls, illustrating the close connection between the neuroendocrine and immune systems (Harrison et al. 1982). Another “natural” experiment that addresses the question of thymic function involves the use of the nude mouse, a genetic mutant that has no thymus. Such mice have very low or no T cell functions and are very susceptible to infections. The thymic remnants in the head and neck region can give rise to a small number of functional T cells (Ikehara et al. 1987). Implantation of a thymus restores T cell function, at least in part (Furukawa et al. 1988). Nude mice have impaired neuroendocrine functions, suggesting that the two systems are intimately intertwined (Daneva et al. 1995). The B lymphocytes responsible for humoral immunity have been thought to be independent of the thymus. Yet one report correlates the age of thymic involution in mice with the age at which certain genes (for example, Rag-1) required for B lymphocyte formation are maximally active in the bone marrow (Ben-Yehuda et al. 1994). This correlation, and the demonstration that a diffusible factor obtained from T cell cultures can induce Rag-1 activity in B-cell precursors, suggests that thymic involution may affect both T and B cell functions. If so, then senescence

469

of both major components of the immune system may originate in the loss of thymic function. However, debate on this point continues, and there is still no generally accepted, unambiguous and definitive explanation for the dramatic functional changes seen in aging rodents and humans (R. A. Miller 1995). Thymic involution is presumed to play a critical role in the age-related loss of function, but the crucial details still elude us. A partial restoration of specific immune functions has been achieved by three different types of interventions. The first intervention, micronutrient supplementation, is low-tech but potentially very useful. Mocchegiani and colleagues (1995) have shown that oral zinc supplementation in old mice for a 1-month period reverses the age-related decline in zinc plasma levels. This treatment also leads to a recovery of thymic functions, a partial regrowth of the organ, and a partial restoration of peripheral immune efficiency. Zinc is a required cofactor for thymulin, a thymic peptide important for the maintenance of cellmediated immunity. The low thymulin activity characteristic of older animals, and the age-related loss of immune functions may be due in part to remediable deficiencies of zinc and perhaps other micronutrients (Fabris and Mocchegiani 1996). The second intervention is caloric restriction. With a 60% caloric restriction, the thymus is initially much smaller than in the ad libitum–fed animals, but it nonetheless has a larger proportion of active thymocytes by 6 months of age. Caloric restriction started at later ages also results in an apparent reversal and/or sparing effect on thymic size and function. The underlying mechanism is still unknown, although it is presumed to involve the neuroendocrine system, as discussed earlier. The third intervention involves the use of synthetic oligopeptide analogues of thymic proteins (R. A. Miller 1995). Administration of these compounds to older mice led to a large and significant increase in the production of IL-2. IL-2 is produced primarily by helper T-cells. It regulates the growth and function of a variety of cells involved in both cellular and humoral immunity and normally shows an age-related decline in its transcription (Pahlavani and Richardson 1996).

470 Chapter 13 Senescence as a Breakdown of Intercellular Regulatory Processes Thus, restoring higher levels of expression of IL-2 should lead to a generalized reversal of agerelated losses. Other thymic peptides demonstrate equally interesting effects (see R. A. Miller 1995). The effectiveness of other thymic proteins such as IL-7 is still unclear (Henson et al. 2004) Taken together, these three interventions suggest that the age-related losses in immune function are reversible, at least in part. The same data can be interpreted as suggesting that the neuroendocrine mechanisms are involved in bringing about—or modulating—these age-related losses in immune function. This is a potentially important topic, which I explore in the next section. A fuller understanding of the mechanisms involved in the maintenance of thymic and lymphocyte functioning would offer a promising base for the development of more effective interventions.

13.3.2 Inflammation and Senescence I have already discussed the important role of oxidative stress in the aging process. Various components of the immune system are a major source of ROS, and the immunological process of inflammation is a major underlying cause of the aging process (McGeer and McGeer, 1999). A number of recent reviews on this topic are available, and this discussion draws mostly from those sources (Chung et al. 2000a,b; Kim et al. 2002). The inflammatory response is described as a three-step process involving (1) endogenous events within the cell that activate proinflammatory gene expression; (2) tissue damage arising from the infiltration of immune surveillance cells and the accelerated production of pro-oxidants; and (3) inflammatory responses leading to altered tissue permeability, protein leakages, and eventual apoptosis or necrosis. When these events happen as a response to harmful bacteria or infected cells, then the process appears to be adaptive. When these events happen in response to our own healthy but aging cells, then the process may be viewed as a maladaptive one in which the immune system has escaped from its normal controls, and its damage to the body obviously contributes to senescence.

ROS and reactive nitrogen species (RNS) obviously can damage the macromolecules of the cell, and this is probably their major immediate effect. But ROS and RNS can also activate harmful genes, and this may be an even more damaging response in the long run. Due to the accumulation of unrepaired cellular damage, the aging organism gradually slips into a redox imbalance brought about by the difference between their increasing ROS production and the decreasing effectiveness of their antioxidant defense systems. This redox imbalance can initiate the beginnings of an inflammation reaction by affecting prostaglandin synthesis. The prostaglandins are signaling molecules secreted by cells and enzymatically destroyed near their site of synthesis; they thus serve as local hormones. Their synthesis is dramatically affected by a variety of specific environmental changes, including an increase in ROS levels. The prostaglandin biosynthetic pathway is a major ROS-generating source, contributing up to 30% of total ROS levels measured (Kim et al. 2000). The increased ROS levels (from any source) induce the activation of the NFkB transcription factor, which then induces the synthesis of a number of proinflammatory cytokines. Cytokines are signaling molecules that provide links within the cells of the immune system as well as connecting them to the NE and the ST systems, and the proinflammatory ones induced here are essentially signaling the existence of an inflammation to the rest of the immune system and thus activating it for the defense of the tissue. There is a striking correlation in the biosynthetic patterns of prostaglandins and cytokines between the usual inflammatory process and the aging process, which led Chung et al. (2002) to suggest that part of the loss of function associated with senescence may be due to a chronic uncontrolled inflammatory response (table 13.6). If senescence is associated with inflammation, then one would expect that factors that extend longevity would also alter the inflammation response in the animal. CR was until recently the only treatment known to increase the longevity of mammals, and its effects have been described in chapters 6 and 7. All the proinflammatory cytokines are activated by the NFkB transcription

13.3 Immunological Theories of Senescence

471

Table 13.6 Correlated Changes in Inflammatory Parameters with Aging Parameter

Inflammation

Aging

Caloric restriction

⇑ ⇓

⇑ ⇓

R R

⇑ ⇑ ⇑

⇑ ⇑ ⇑

R R R

⇑ ⇑ ⇑

⇑ ⇑ ⇑

R R R

⇑ ⇑ ⇑ ⇑ ⇑ ⇑

⇑ ⇑ ⇑ ⇑ ⇑ ⇑

R R R R R R

Redox state ROS/RNS ADS Proinflammatory enzymes Inducible NO synthetase Heme oxygenase-1 Cycloxygenase-2 Proinflammatory cytokines IL-1b IL-6 TNFa NFkB activation NFkB binding activity NIK/IKK activation Phsophorylation of IkBa Degradation of IkBa/b in cytoplasm NFkB-dependent gene expression Activation of STPs (ERK, JNK, etc.) Source: from table 1 of Chung et al. (2002).

Note: ROS, reactive oxygen species; RNS, reactive nitrogen species; ADS, antioxidant defense system; IL, interleukin; TNF; tumor necrosis factor; NFkB, nuclear factor kappa beta; NIK, nuclear factor kappa beta-Inducing kinase; IKK, inhibitory kappa kappa; IkB, inhibiting kappa beta; STP, signal transduction pathway. R, change is reversed by caloric restriction (CR).

factor, which is very sensitive to ROS levels. Thus, the inflammatory response depends on the activation of the factor. CR eliminated both the usual age-related increase in ROS levels and the usual age-related increases in COX2 and NFkB activity relative to the ad libitum fed controls (Kim et al. 2000, 2002). Thus, the large ROS load generated by prostaglandin synthesis does not occur, and the NFkB transcription factor is not induced. This factor normally exists in the cytoplasm in an inhibited state due to its binding to IkBb (Inhibitor-kappa B beta) and IkBa (Inhibitor-kappa B alpha) molecules. Phosphorylation of these IkB molecules causes dissociation of the complex and relieves the NFkB inhibition. The kinase enzyme that performs this phosphorylation step is activated by relatively small increases in redox levels. The CR-induced maintenance of low ROS levels maintains this proinflammatory transcription factor in the inactive state. This also constitutes evidence showing that senescence is an event-dependent, not a time-dependent, process.

NfkB plays a central role in the pathology of many common age-related diseases (Tracy 2003). Atherosclerosis, for example, was once thought of simply as a “plumbing problem.” Later the underlying role of inflammation was appreciated, but it was considered to be just a reflection of the underlying disease. As more data accumulated, it became apparent that the inflammation was not just a symptom but was actually the cause of the disease. The current model incorporates positive feedback cycles with these earlier views to reflect the role of inflammation in this senescent disease (figure 13.6). This model is not limited to atherosclerosis. The brain lesions associated with Alzheimer’s disease, for example, are characterized by the presence of a wide variety of inflammatory markers and mediators. These molecules are synthesized by the brain cells in the region of the lesion, suggesting that these neurons are under oxidative stress. These inflammatory effects exacerbate the original lesion and hasten the progression of the disease (McGeer and McGeer 2002). In both of these cases, the long term use of nonsteroidal

472 Chapter 13 Senescence as a Breakdown of Intercellular Regulatory Processes

Aging, Lifestyle

Redox Imbalance Vascular Sources Inflamed atheroma Hypertensive arteries

Nonvascular Sources Proinflammatory Cytokines

Adipose tissue Chronic infection

IL-6

Blood Vessel Atheroma

Liver CRP Fibrinogen

Endothelial Cell Altered cell adhesion molecules

Figure 13.6 One probable manner by which lifestyle choices or aging changes can induce an accelerating inflammatory cycle, which will give rise to serious cardiovascular pathologies.

anti-inflammatory drugs (NSAIDs), which inhibit the synthesis of prostaglandins, reduces the risk of diagnosis and/or progression of the disease. It is now known that activation of NFkB in mouse hepatocytes leads to hepatocellular carcinoma (Pikarsky et al. 2004). In mice with inflammatory hepatitis, the inflamed stromal cells produce tumor necrosis factor-alpha, which induces the neighboring hepatocytes to activate NFkB. The NFkB, in turn, indirectly inhibits these precancerous cells from entering apoptosis. This then encourages the proliferation of all hepatocytes, including those in a precancerous state. Eventually the presence of other carcinogenic agents would push tumor progression further toward malignancy. Hypertensive rats have higher than normal levels of NFkB activity, and the resulting inflammatory responses are thought to play a causal role in the genesis of their hypertension. Feeding such rats a normally occurring metabolite (sulforaphane), which induces various defensive enzymes, results in significantly lower oxidative stress levels, a better endothelial-dependent

relaxation response of the aorta, and lower blood pressure (Wu et al. 2004). Sulforaphane is normally found in young broccoli sprouts, and offers one more piece of evidence that a dietary shift may have a significant effect on one’s health status. Inflammation plays a central role in defenses against pathogens, but if the response is not ordered and timely, then chronic inflammation results. The data presented above demonstrate that chronic inflammation may well play a role in bringing about the onset of major age-related chronic diseases such as hypertension, atherosclerosis, or cancer. The available evidence shows a correlation between various inflammatory mediators and mortality in elderly populations, although the complexity of the lifestyle variables makes it difficult to draw a casual conclusion at this time (Krabbe et al. 2004). Interactions that selectively inhibit the NFkB pathway may provide a useful intervention for such diseases. The recognition of inflammation’s major role in the genesis of these pathologies should also make us aware of the potential of dietary and/or lifestyle

13.4 Integrative Aspects of the Neuroendocrine–Immune System

alterations to provide an upstream inhibition of generating disordered and untimely inflammatory responses in later life.

13.4 Integrative Aspects of the Neuroendocrine–Immune System I pointed out in chapter 5 that the nervous, endocrine, and immune systems are integrated in a single neuroendocrine–immune system in the sense that each of them affects the proper functioning of the others. Consider, for example, the differences in the functioning of the male and female immune systems as a result of their different hormonal milieus. Women usually mount a more vigorous immune response than men, yet women show a higher incidence of autoimmune disease than men. In addition, women suppress any immune response against their fetus but generate substantial postpartum antibody levels while they are nursing a baby. Recent research suggests that the answers to these and other puzzling questions involve the hormonal differences between the sexes. A variety of studies have shown that estrogens and/or estradiol have a stimulatory effect on the levels of hormones such as prolactins and growth hormones, which then increase the production of T and B cells. Progesterone and testosterone, in contrast, are reported to have a suppressive effect on the immune system. The net effect of these relationships in the context of the female rat estrous cycle is that estradiol secretion peaks before ovulation (see chapter 5) and increases the immune response of the uterus, cleaning it of bacteria and preparing it for implantation. In the rat’s cervix and vagina, however, the same estradiol induces a tissue-specific suppression of the immune response, presumably so that the sperm will not be attacked. After ovulation, when estradiol decreases and progesterone increases, the female’s immune response is partly suppressed, presumably so that it does not attack the fetus. The drop in progesterone and the increase in prolactin and/or estradiol after birth stimulate the immune system

473

and end its temporary suppression. The female’s cyclic hormonal levels are reflected in her cyclic immune functions. Postmenopausal women are no longer cyclic, and their overall immune system responses are low and constant, much like a man’s immune response. More evidence of the intimate interactions between the neuroendocrine and immune systems is the simultaneous appearance in cells of components of both systems or, more convincingly, the demonstration that the secretory product of one system will significantly affect the functional status of the other system. The following several examples support the view that the three systems are integrated into one: 1. The Schwann cells of the peripheral nerves synthesize progesterone, which promotes myelin formation during nerve regeneration (Koenig et al. 1995). 2. Functionally impaired neurons, but not normal active neurons, express major histocompatibility complex Class I genes and thereby make themselves available for immunosurveillance by cytotoxic T cells (Neumann et al. 1995). 3. Bone resorption by osteoclasts is activated by various cytokines (IL-1, IL-6, tumor necrosis factor, and so on) secreted by thymic stromal cells and by peripheral blood monocytes. These activation processes are inhibited by estradiol acting directly on these cells (Horowitz 1993). 4. The saliva contains different antibodies, predominantly IgA and IgM. The levels of IgM are significantly affected by the light–dark cycle: The peak values in individuals overwintering in Antarctica are temporally correlated with the absence of daylight (Gleeson et al. 1995). It seems reasonable to assume that neuroendocrine factors involving the pineal gland and melatonin secretion are involved in this light–dark modulation of an individual’s humorally mediated immune function. 5. Both single and extended periods of exercise have been shown to significantly alter T cell–

474 Chapter 13 Senescence as a Breakdown of Intercellular Regulatory Processes mediated immune function (Mazzeo 1994; Nash 1994), reducing the incidence and severity of infection. Again, the neuroendocrine system provides the most likely mechanisms through which physical exertion can affect immune function. 6. Patients suffering from hyperthyroidism do not show the age-related decline in thymulin levels found in normal individuals, but patients suffering from hypothyroidism have levels of thymulin that are significantly lower at all ages than in the normal controls (Fabris et al. 1995). 7. Cytokine treatments decrease the levels of thyroid hormone in rats and humans, and treatment of old mice and humans with thyroid hormone or thyroid-stimulating hormone gives rise to increased thymulin production, lymphocyte proliferation, and T cell–mediated immune function (Fabris et al. 1995). 8. Treatment of old mice with thymosin yields an increase in the release of adrenocorticotropic hormone and glucocorticoids, and treatment of old mice with growth hormone yields increased levels of IL-2 relative to controls. 9. One excellent example of the integrated functions of the NEI system can be found in the cyclic activities of the female, but not the male, immune system. Females generally have a more variable but occasionally stronger immune response than do males, and this generally seems to be advantageous. Before ovulation, estrogen up-regulates the secretion of both prolaction and growth hormone. These hormones in turn increase the production of both T and B cells to a very high level, thus giving the female enhanced immunological protection relative to males. But during the second half of the menstrual cycle, the drop in estrogen and the increase in progesterone leads to a partial suppression of the immune system. Adaptive reasons having to do with not encouraging the immunological rejection of foreign bodies such as sperm or embryo have been put forth to account for the coordinated cycling of the NEI/reproductive system (Morell 1995). This cyclic activity means that

women may be more or less vulnerable to infection than men, depending on the stage of their cycle when exposed. After menopause, the female’s level of circulating antibodies loses its cyclicity and becomes low and steady, much like that of the male. Other examples are available in the literature, but those above suffice to illustrate that the neuroendocrine and immune systems are functionally integrated into one NEI system. The findings need confirmation and elaboration, yet they appear persuasive. If confirmed, they imply that the age-related immune dysfunction is not intrinsic to the immune cells but arises as a result of agerelated defects in other neuroendocrine functions. The reciprocal may also be true. If so (and the reader should be cautioned against undue optimism), the potential for effective intervention in the aging process may be larger than was previously suspected. Psychological factors have also been suggested as impinging on the modulation of these functions (Tricerei et al. 1995). The complex interactions of the NEI system are also well illustrated by the role of the major histocompatibility genes in the aging process. The topic has been reviewed by Lerner and Finch (1991), Yunis and Salazar (1993), and Crew (1993), and this discussion draws heavily from those sources. Our knowledge of the histocompatibility genes originated in the genetics of transplantation. If skin from an inbred strain of mice is grafted to another individual of the same strain, it is accepted and will grow normally. But if it is grafted to an individual of a different inbred strain, the graft is eventually rejected. To find out which genes were involved in this tissue recognition and rejection process, the genetic differences between the strains were gradually minimized by inbreeding until one particular region of the mouse chromosome 17 could be identified as causing the rejection. Because of its role in tissue recognition and rejection, this region was named the major histocompatiblity complex (MHC), or H-2 region. A homologous region in humans is located on the short arm of chromosome 6 and is termed human leukocyte system A (HLA). All vertebrates that have been examined contain such a gene cluster.

13.5 Signal Transduction

The MHC complex in mice and the HLA complex consist of highly polymorphic series of a dozen or more loci, comprising at least 1 million base pairs, which code for proteins found in the blood serum as well as on cell membranes. The genes involved encode membrane glycoproteins and have been sorted into two classes. The class I genes code for antigens that are expressed ubiquitously throughout the body and are the main basis for graft rejection. The class II antigens occur mostly on cells derived from bone marrow, including B and T cells, macrophages, and other antigen-presenting cells. Thus, the MHC genes are intimately involved in immunological identity. In humans, there are 30–60 alleles of each class I gene and slightly fewer of each class II gene. These polymorphisms are widespread in all human populations, such that any one allele rarely exceeds a gene frequency of 10%. In addition, the complex contains genes coding for several serum complement factors (class III genes) and a number of genes with no obvious role in immune function. The complex affects reproduction. In mice, specific MHC genes and alleles are associated with mate selection, patterns of estrous cycles, fertility, fecundity, rate of embryonic development, and reproductive aging and senescence. For example, a comparison of three different H-2 haplotypes tested on a B10 genetic background shows that the H-2 r haplotype differs from the other two haplotypes in number of litters, number of pups per female, and age at last litter. Because that data also show a remarkable concordance between age at last litter and maximum life span, we must conclude that the MHC genes are exerting some kind of controlling influence on both reproduction and longevity. Another interesting aspect of the MHC and HLA complexes is the association of certain MHC or HLA genotypes with certain traits, some of which have a direct immunological connection but many of which have only an indirect connection or none. For example, an individual carrying the HLA-B27 genotype is 87.4 times more likely to develop ankylosing spondylitis (a degenerative bone disease) than is an individual who

475

does not carry this genotype. The percentage of women with the HLA-B8 genotype declines with increasing age, an indication that such individuals may have a shortened life expectancy. An examination of centenarians and normal-lived controls revealed that three particular HLA alleles (DR7, DR11, and DR13) were significantly overrepresented in the long-lived groups (Ivanova et al. 1998). Two of these allele-specific effects appear to be gender-specific, with DR7 mostly affecting men and DR11 mostly affecting women. It appears to be advantageous to be heterozygous: there is a decreasing proportion with age of HLAhomozygous individuals. The mechanisms underlying these associations are far from clear, but they suggest that the MHC/HLA gene complexes may exert an effect on longevity because their effects are not limited to the immune system but are more widespread and affect basic body processes.

13.5 Signal Transduction All of the complicated intercellular signaling mechanisms we have discussed above come to a common destination at the cell membrane, where each of these different extracellular signals need to be translated into specific sorts of intracellular signals that will specifically affect the appropriate function or structure of the cell. This system of intracellular signaling is called signal transduction, and it plays an important role in homeostasis and senescence. For example, the discussion of the genetic mechanisms underlying longevity determination and senescence focused on the ISP, which is a classic instance of the crucial role signal transduction pathways (STP) play in providing the genome with the environmental signals necessary for the regulation of gene expression. Logically, the discussion of STPs belongs in the previous chapter dealing with intracellular phenomenon, but it is included it here because it provides the ultimate target of all the signaling networks in the body. Reviews on this topic are available (Martindale and Holbrook 2002; Pawson and Saxton 1999; Pires-da Silva and Sommer 2003), and this discussion draws on those sources.

476 Chapter 13 Senescence as a Breakdown of Intercellular Regulatory Processes The STPs are typically activated by the binding of a ligand to a transmembrane receptor, which in turn leads to the modification of various cytoplasmic transducers, which activate certain transcription factors, which in turn alter specific aspects of gene expression. Given the complexity and specificity of the signals impinging on the cell, it is surprising that only a few pathways are involved. And this is more surprising when one realizes that these same dozen or so STPs function to control embryonic development as well as to regulate adult homeostasis and senescence. This raises the key question I address in this discussion: how can so few STPs allow the cell to express such a large and diverse repertoire of specific responses appropriate to each of the extracellular signaling molecules impinging on its membrane? The common names and abbreviations of some STP families that I consider are (1) hedgehog (Hh), (2) wingless related (Wnt), (3) transforming growth factor-beta (TGFb), (4) receptor tyrosine kinase (RTK), (5) notch, (6) janus kinase (JAK), (7) signal transducer and activator of transcription (STAT), (8) extracellular-signal regulated kinase (ERK), (9) c-Jun N-terminal kinases (JNK), and (10) nuclear hormone receptor (NHR). Each family is composed of a number of related molecules, each of which has slightly different functional specificities (table 13.7). All STPs are important, for the failure of any one usually leads to the death of the cell. However, from the standpoint of aging, the most interesting STPs are those involved in regulating the cell’s responses to ROS and other stresses (Martindale and Holbrook 2002). Oxidative agents can evoke a wide spectrum of responses from cells, ranging from proliferation to growth arrest to cell senescence to cell death (Finkel and Holbrook 2000). What accounts for this multiplicity of responses? Some of it undoubtedly depends on the particular cell type being tested because the particular genes preferentially activated in various differentiated cells will dictate to some degree the cell’s response. Some of it undoubtedly depends on the particular conditions or microenvironment surrounding the tested cells. Some of it seems to

depend on the strength of the oxidative stress. Low oxidant concentrations tend to promote cell proliferation, whereas high oxidant concentrations tend to result in growth arrest and/or cell death. And finally, some of the outcome variability seems to depend on the relative balance between the various signaling pathways activated by any particular oxidative stress. The signaling pathway allows the cell to carry out a highly nuanced response to variable environmental conditions, as I point out below. One pathway promoting cell proliferation and survival in response to oxidant stimulation is that leading to activation of the ERK pathway (Li and Holbrook 2003). ERK is a subfamily of a larger group of kinases known as mitogenactivated protein kinases (MAPK; Martindale et al. 2002). The ERK pathway is mostly concerned with the regulation of cell proliferation and can be activated by a variety of growth factors as well as by low levels of oxidants; the binding of a growth factor to its receptor allows the receptor to initiate a signaling cascade that eventually activates ERK. For example, either oxidants or the epidermal growth factor (EGF) can bind to the EGF receptor. The resulting signal cascade activates a membrane-localized Ras, which then indirectly activates the Raf, MEK and ERK pathways. Their activation usually leads to a proliferative response on the part of the cell. On the other hand, different growth factors and/or higher oxidant levels can lead to the activation of the JNK pathway and/or the p38 pathway, the activation of which generally leads to growth arrest and apoptosis. The JNK family is mostly concerned with stress responses and is activated by a wide range of stresses including cytokines, radiation, osmotic shock, heat stress, mechanical injury, and oxidative damage. This family includes, for example, the PI3K/Akt STP, which is an integral part of the ISP and plays a major role in longevity determination (chapter 7). Both the ERK and JNK pathways are activated by a complex cascade of MAPK enzymes. Each of the enzymes involved in this cascade appears to have its own specificity, and this certainly contributes to the ability of a few pathways to generate specific responses.

13.5 Signal Transduction

An inspection of table 13.7 reveals that the number of signaling molecules in the STPs dramatically increases as the complexity of the organism increases. Yeast appears to have a minimal number of active pathways compared to the multicellular model organisms. Humans appear to have the most complexity, with the exception of the nematodes’ NHR family, which evolution seems to have fully exploited. Clearly, the increased numbers of different molecules within a pathway allows for the transmission of specific input and output signals. The 48 receptor tyrosine kinases (RTK) ligands acting on the 25 RTK receptors could allow the human cell a minimum of 1200 specific combinations, which one might think sufficient to allow one pathway to carry a large number of different messages. However, the numbers in the table are minimum values, for they are based on an in silico analysis of the several genomes and reflect the number of genes that contain each STP families’ characteristic DNA sequence. But proteins are the operative molecules in the cell, and proteins are often formed via differential splicing from a number of different exons. Many proteins have multiple STP domains, and so their ability to act in some sort of a combinatorial manner might greatly increase the number of unique messages each pathway could encode. In addition, other STP molecules have different effects depending on their extent

Table 13.7 Numbers of Signaling Molecules in Selected Signal Transduction Pathways (STPs) of Model Organisms STP Ligand RTK TGF-b Wnt Notch STAT NHR Receptor RTK Wnt NHR

Human

Fly

Worm

Yeast

48 29 18 3 7 47

3 6 7 2 1 17

4 4 5 2 1 147

0 0 0 0 0 0

25 12 59

6 6 25

1 5 270

0 0 1

Source: after data presented in Subramanian et al. (2001).

477

of phosphorylation. For example, one particular STP receptor (the beta platelet-derived growth factor receptor, or bPDGFR) is a dimer with five different phosphorylation sites. The number and type of each phosphorylated site dictates which particular STP molecule the receptor will interact with. The net result is that this one receptor can activate or repress about six different signaling pathways (Pawson and Saxton 1999). The ultimate goal of the STPs is to activate or inactivate some transcription factor and thereby alter the cell’s gene expression pattern. In many cases, a group of genes participate in the same cellular process and must be co-expressed. The simplest way to bring this about is to regulate their transcription by a transcription factor that binds to common promoter elements in each of the required genes. The binding of the NFkB factor to the promoters of the several cytokines that it induces is an example of this strategy. But this strategy is also the signature of a scale-free network. The limited number of intracellular STPs interact in a combinatorial manner to build scale-free networks that allow diverse cellular responses to a great variety of extracellular signals. One advantage of a signaling network with multiple intersecting pathways is to allow for a coherent response to numerous, potentially conflicting signals. A cell being subjected to oxidative stress may be receiving mitogenic signals, apoptotic signals, stress signals, inflammatory signals, and the like. If the different signals act synergistically, then their effect will be to enhance one pathway and thus yield one sort of response. If the different signals act antagonistically, then perhaps their relative intensities will strengthen or weaken their mutual inhibitory intersections, allowing the cell to identify the strongest signal and react to it. When one summarizes the existing data to determine which STPs result in an enhanced cell survival or result in cell death, then one often finds that the same pathway can deliver either result on occasion (table 13.8). This result suggests that a given input signal (e.g., oxidative stress) does not automatically result in survival or death but is likely to be the outcome of the cell’s weighing of the various input signals and making a measured response.

478 Chapter 13 Senescence as a Breakdown of Intercellular Regulatory Processes Table 13.8 Probabilistic Cellular Outcomes of Activation of Some Signal Transduction Pathways (STPs)

The same network of signaling proteins downstream of the initial receptors needs to coordinate many cell functions. Eukaryotic cells have evolved scaffolding proteins whose function is to simulataneously bind multiple components of an STP. Apparently different scaffolding proteins bind different members of the same family and allow for some degree of specificity in this network. This then allows for the grouping in one physically linked structure of the appropriate proteins from each level of the pathway. A summary diagram of the MAPK cascade and the different transcription factors it can activate is shown in figure 13.7. The diagram has five (horizontal) levels indicating the sequential layers (top to bottom) in which the different members of each category of the STP are located. The MAPK

Cellular outcome STP

Enhanced survival

Cell death

+ +++ ++ +++

+++ ++ +++ –

NFkB ERK JNK JAK/STAT

Source: after Martindale and Holbrook (2002). Note: (–) there is minimal or no evidence that this pathway influences this outcome; (+) there is some evidence for this outcome; (++) there is much evidence for this outcome; (+++) this is the predominant outcome for this pathway.

Stimulus

Growth Factors Cytokines, cellular stress ceramide

GTPase

Ras

Oxidative stress osmotic shock

TGF-

Rac1 Rho

MAPK4

PKC

MAPK3

Mos Raf Tpl2

MAPK2

MEK1

MAPK

ERK1 ERK2

Transcription c-Myc factors

Src

GCK

PAK

MEKK3 MEKK1/2 MUK

MEK2

Sap1

MLK

MKK7

MKK4

c-Jun

TAB1

???

TAK

???

MKK6

MEK5

MKK3

JNK

Elk1

???

ATF2

p38

ERK5

MAX

MEF2

Figure 13.7 The mitogen-activated protein (MAP) kinase cascade in higher eukaryotic cells. A variety of stimuli, including both reactive oxygen species and antioxidants, initiate the cascades. Signal integration mostly takes place at the level of transcription factors that in some cases can be regulated by several MAP kinase cascades. The genes activated by these transcription factors include detoxifying enzymes as well as those involved in stress resistance. These genes constitute an antistress network but one that is crucially dependent on the functioning of their signal transduction pathways. (After Yoon et al. 2002.)

13.5 Signal Transduction

subfamily includes, for example, the five listed members (from ERK1 to ERK5). The fact that there is more than one entry in each sub-box at the MAPK level (i.e., ERK1 and ERK2) means that one or the other molecule is involved, but not both. Scaffolding proteins likely function so as to incorporate one member from each level in one structure. Considering only the ERK1–ERK2 column of figure 13.7, there are 36 unique cascades that can be formed by the 10 listed molecules and 77 such cascades for the figure as a whole. The combinatorial nature of the system allows for a large number of specific and sensitive cascades and responses. What happens to the STP in aging cells? There has been a long-standing hypothesis that aging is associated with a reduced ability of the cell to initiate prosurvival STP in response to oxidative stress, and that this may underlie the greater vulnerability of the aged cell to oxidative damage. Evidence to support this concept has been obtained (Ikeyama et al. 2002). Hepatocytes from old mice or old rats are significantly less able (threefold less) than young animals to phosphorylate and activate their ERK or Akt pathways. It appears that the age-related decline in both these species arises from a common failure to phosphorylate a particular amino acid in the EGF receptor (Li and Holbrook 2003). It is

479

possible that this failure might arise as a result of an age-related membrane alteration. The ability of CR to prevent this age-related decline in STP activation suggests that modifiable factors lie at its root. Looking at the function of our NEI and ST systems, we can conclude that our regulatory processes are composed of a hierarchial system of interacting networks that serve to detect and transmit environmental signals to nuclei and thus bring gene expression profiles and physiological functions into line with the world in which we live. Such a nested set of interacting networks is very efficient at transmitting messages from one level to another, and this serves us well during development and the adult health span. By the same token, this network is also very efficient at spreading the effects of unrepaired damage at any level up and down the network. Attenuated or altered extracellular hormone levels or cytokine secretion or cell membrane receptor density or phosphorylation activity of intracellular kinases will eventually bring about a loss of function in the cell or tissue, and this lessened function will spread through the networks and further influence them in the direction of senescence. Identifying the early decreases in critical STPs and intervening so as to halt or reverse them is the simple goal of anti-aging interventions.

This page intentionally left blank

Change (0/100) 100 Functional impairment Threshold Cryptic damage

Part V 50

Change (0/100) 100

10 0

An Integrated Theory of Aging 0

Time Change = f(dose, time)

Life span

This page intentionally left blank

14

A Theory of Aging over the Life Span

14.1 Biological Theories of Senescence I have reviewed the biological facts and theories of the aging process. What can we say in the way of a conclusion? One thing that this tour of the field should have made obvious is that we know a great deal about the biological changes that accompany aging, much more than we did just some few years ago. We have characterized a few senescent mechanisms in great detail, but there is still much that we do not understand about the others. We have identified and characterized some of the longevitydeterminant mechanisms, and now we are faced with the task of integrating these bits and pieces into some coherent tale. The material covered in part IVof the book clearly implicated the involvement of certain genes, certain types of damage-causing processes, and certain types of repair or synthetic processes as being central to the aging process. Not only has the large number of plausible mechanisms been reduced over the years to a probable few, but the nature of the research questions now being posed has been honed to address very specific problems. A variety of good experimental animal models are available, , each with characteristic strengths and weaknesses. There is every reason to believe that the current progress being made in deciphering these processes can be maintained and even increased in the future. Several of the theories are supported by good data. Some of the postulated mechanisms might well dovetail into one another; others seem to be quite incompatible. This disparity is less of a

problem than it might appear at first. There is no reason to expect there is only one biological mechanism responsible for aging in all of the different species and kingdoms of living organisms. The differences in life-history strategies alone impose very real reproductive and physiological divergences even on the few species used in research on aging. In primates, life span is correlated with brain size; in birds, life span is correlated with energy efficiency (see figure 4.12 and table 11.4). These different statistical associations imply that the same outcome (an increased life span and/or a decreased rate of aging) is governed by multiple and varied inputs. Sacher (1978) suggested that mechanisms of senescence have not been selected for directly per se, but that natural selection has acted to develop “longevity assurance” mechanisms that can guarantee vigor in animals until the time of sexual maturation and reproduction. In that context, it is easy to imagine that there might be a wide variety of mechanisms that could be altered to maintain the optimum functioning of an organism for a limited period of time, and this diversity may be reflected in the disagreements we have noted between sets of experimental data obtained from different species and orders. Despite the existence of diversity, one lesson that biology suggests is that there are fundamental structural, functional, and biochemical similarities of cells and tissues. Among animals, at least, it seems reasonable to suggest that the diversity of aging processes is more apparent than real. Different long-lived strains of Drosophila, for example, may share a deep biological unity in that

483

484 Chapter 14 A Theory of Aging over the Life Span each strain may have altered its energy metabolism to better enable them to withstand physiological stresses, and their apparent differences are quite secondary, simply reflecting the manner in which each strain has organized the details of their metabolic changes. It is quite possible that the different mechanisms of aging operate in a variety of disparate ways to produce distinctive types of cellular damage, the end results of which, however, will be similar in any animal cell. It is a reasonable and interesting exercise, although an admittedly speculative one, to search for such common themes, as we will do now.

14.1.1 Senescence and the Maintenance of Function Our cells and tissues are thermodynamically unstable structures. Their highly ordered molecular structures are under continual attack by stochastic degradative events and processes. Even a cursory study of cell biology lets us see that the typical animal cell is a highly integrated construct of exceedingly complex and intricate structures, under which there is layer upon layer of intricate molecular regulatory processes, all of which appear to be important for the continued smooth and homeostatic operation of the cell. Regulatory processes are both the strength and the weakness of a cell. They allow for an impressive degree of subtlety in responding to a large variety of physiological changes. However, any structure becomes more susceptible to being disrupted by stochastic damage as the number of potential targets—the regulatory processes and/ or molecules—increases. As with any complex structure, the cell not only requires a continuous and stable supply of nutrients and raw materials to maintain itself, but it must also be able to combat the stochastic degradative processes that threaten the integrity of its regulatory mechanisms. If our cells were not able to detect, repair, and replace their damaged structural and functional components, they would probably have a limited life span. Consider, for example, our red blood cells. As highly specialized and enucleated cells, they have no repair capabilities to speak of.

Consequently, they die within about 90 days as a result of irreparable damage to important membrane components. I have examined numerous postulated stochastic mechanisms of aging. The discussion of the evidence has suggested that the free-radical or oxidative-damage theory is a probable explanation for aging. The oxygen-induced free radicals are produced as a consequence of aerobic respiration. The mere act of breathing oxygen to produce the energy needed to take another breath also produces the free radicals that threaten to impair our ability to take that breath. These highly active free radicals cause substantial amounts of oxidative damage to the subcellular components. The cell can maintain itself in the face of these insults only by expending substantial amounts of energy on antioxidant defenses and repair processes. Much of the genetic evidence that I have reviewed also points to the involvement of free radicals in the aging process, and the evidence that seems to point in another direction (such as the alteration of gene activity patterns by caloric restriction) may in fact be pointing toward the necessity of maintaining active genetically based defense or repair processes in order to delay the onset of senescent damage. The large body of evidence dealing with in vitro studies can also be interpreted as suggesting that the efficiency of the repair and replacement mechanisms is adversely affected in differentiated cells, perhaps because they are linked to the processes of cell division, which seems to be inhibited in such cells. Finally, in large, multicellular animals such as humans, some of the more important aspects of senescence likely do not primarily involve changes in autonomous and general cell functions per se, but rather the intercellular processes that regulate our complex physiology and provide us with our intercellular homeostatic mechanisms. Certainly, both the immunological and the neuroendocrine theories of aging rest on concepts of gene action in one cell affecting the outcome of processes in another cell. The only effective technique we have at present of extending mammalian life spans is caloric restriction, and this intervention appears to exert its effects on gene activity via the neuroendocrine system. The study of reproductive aging

14.1 Biological Theories of Senescence

in mammalian females has shown that aging may best be viewed as a genetically based, event-driven process in which the fundamental events occur on one physiological level (the hypothalamus) but are translated by the target cells into an alteration of gene activity patterns. The genetic activity patterns implicated by these several different theories may be related to the repair activities of the affected cells. The term “repair” should be defined to include more than just the concept of DNA repair; it should be broadened to refer to the maintenance of all cellular activities. In fact, the most fundamental aging process that we can yet identify may well be the decrease in the organism’s ability to maintain its functional integrity.

14.1.2 Three Types of Integrative Models: Qualitative and Quantitative Approaches The central role of repair is the essence of the disposable-soma theory of aging. Two different investigators independently made this concept the centerpiece of their integrative models of aging. I now explore three such models, the first two of which were developed by Lamb (1977) and by Kowald and Kirkwood (1994, 1996). The third model that I discuss at some length is the hierarchial network model. The first model is a verbal, or qualitative, approach to organismic aging; the second is a mathematical, or quantitative, simulation of the repair and damage events central to aging and senescence at the cellular level; the third is a mathematical model in which connectivity and fluxes are of major importance. 14.1.2.1 A Linear Model

Lamb (1977), following a review of the literature, developed her seven-step model as a chain of events that might lead in many cases to senescencence and death. First, stochastic damage is normally caused by a wide variety of diverse mechanisms and can affect any component of the cell from the DNA level to the cell membrane. Similar types of damage may arise from different

485

types of mechanisms. A large part of my discussion of theories of aging was devoted to a recital of the mechanisms known to cause significant and diverse damage to the cell. The fact that damage is caused may be more important than the mechanism by which it is produced. Perhaps all the damage-causing mechanisms I have discussed play a role in the aging process; if so, that might explain why there are plausible data for all of the mechanisms but no overwhelming proof in favor of any single one. Second, the cell usually has several different types of repair processes that are normally operating at levels sufficient to cope with the stochastic damage and thereby maintain the structural and functional integrity of the cell. The cell has a sufficient reserve capacity so that even if repair processes are not perfectly efficient there are no operational problems, at least while the organism is young. Third, for reasons that are still imperfectly understood, these repair processes become less efficient with age. This situation may arise as a result of event-driven alterations in gene transcription and translation, or of decreased energy production, or of a shift in the cellular economy to maintain the specialized functions of the cell at the expense of the repair processes, or of another plausible process. In any event, the mechanism(s) responsible for this temporal decrease in repair efficiency are likely to be the key step(s) in the aging process. Fourth, the combination of a more or less constant rate of damage coupled with a declining efficiency of repair processes ensures that the number of structural abnormalities will increase with time. The rate of increase will depend on the kinetic details involved, but it will increase. The persisting damage will affect both structural and regulatory components. Fifth, as a result of unrepaired damage to its homeostatic regulatory mechanisms, the cell will no longer be able to perform the same functions and maintain the same homeostatic balance with the same efficiency as it has in the past. Early signs of this decreased cellular efficiency may be noted in an increased variance of certain functions, in an altered response time between functional

486 Chapter 14 A Theory of Aging over the Life Span periods, in a decrease in the rate of synthesis of important components, or in a decrease in the functional output of the cell. Sixth, the decreased cellular efficiency leads to a decreased functional capacity of tissues, organs, and organ systems. This would be especially evident if the cell were a key member of a homeostatic regulatory network, for its decreased efficiency would then cascade to all the cells that it controlled, even if those latter cells were not affected by autonomous cell damage. Seventh, the ability of the organism to cope with the changing environmental demands becomes progressively diminished. The probability of dying increases to near certainty. It is likely that any one of these processes can be significantly altered by environmental interventions, and I have given many examples of how the genotype and the environment interact to produce a characteristic life span. Such interactions would add the necessary individual variability to this model. This is a pioneering model, one of the first to view aging as the outcome of a hierarchal cascade, and so it is not surprising that the model does not encompass the specific gene/ protein interactions picked up by recent data. The linear model is not wrong, but it is certainly too simple. The fact that it is qualitative rather than quantitative is somewhat less of a criticism. Nonetheless, all subsequent work is built on Lamb’s (1977) hierarchial cascade model. 14.1.2.2 An Early Network Model

A good start toward a quantitative systems approach was presented by Kowald and Kirkwood (1994, 1996). They developed the idea that aging and longevity may be understood as the outcome of a network of maintenance processes that control the capability of the system to preserve homeostasis. The other important advantage of their analysis is that it is a quantitative model. It has been said that the beginning of true understanding lies in being able to measure what you are talking about, and a numerical analysis does permit one to make specific predictions and thus continually refine the accuracy of one’s thoughts. Kowald and Kirkwood (1994) considered that the

free-radical theory and the protein error theory were each important but incomplete descriptions of the aging processes within a cell—incomplete because each could interact with another at particular points and provide sources of damage and/ or protection not specifically predicted by either theory alone. As one example of such interaction, Kowald and Kirkwood suggested that free radicals could damage enzymes and thereby provide another source of abnormal protein not specifically foreseen by the original theory. To the extent that such abnormal proteins included abnormal antioxidant enzymes, the level of protection against free radicals would be reduced because of this protein error, another source of damage not specifically foreseen by the original theory. Other sorts of interactions may be imagined. Kowald and Kirkwood (1994, 1996) first considered that a simplified system comprising these two theories of aging could be composed of free radicals, antioxidant enzymes, antioxidant scavengers, and the ribosomes, synthetases, and/or mRNAs involved in protein synthesis. Figure 14.1 illustrates the reactions and interactions of their MARS (mitochondria, aberrant proteins, radicals, and scavengers) model. In this model, the free-radical production rate is the key variable and depends on the kinetics of their production in the free-radical source (mitochondria) and their destruction in their sink (superoxide dismutase and other antioxidants). The production of free radicals thus depends on the level of energy provided by the mitochondria and on the synthesis, turnover, and degradation of the mitochondria. The model also takes into account the diffusion of free radicals from the mitochondria into the cytoplasm and the damaging effects such radicals might have on different classes of cytoplasmic components. After making several literature-based assumptions regarding the concentrations, reactions, synthesis, turnover, and energy consumption of each of the several components, Kowald and Kirkwood (1994, 1996)derived a series of 15 differential equations to describe the interactions of the system components. Subsequent computer simulations yielded some interesting results. High levels of free radicals, insufficient levels of free-radical protection, and high levels of protein error each lead to an inte-

14.1 Biological Theories of Senescence

487

mRNAs

Radicals

Radicals

ATP

Mitochondria

Scavengers

Antioxidants

Normal Abnormal

Normal Abnormal

Normal Abnormal

Ribosomes and synthetases Normal Abnormal

Figure 14.1 The reaction scheme of the MARS (mitochondria, aberrant proteins, radicals, and scavengers) network model of aging. For clarity, the reactions of ribosomes and amino-acyl transfer synthetases, as well as the two different types of proteolytic enzymes, are presented in one functional compartment each. Radicals can damage all components and are removed by antioxidants; this is indicated by the interrupted reaction path for the radicals. (After Kowald and Kirkwood 1996.)

grated breakdown of homeostatic process and cell death (figure 14.2). This particular simulation suggests that the main targets of free-radical damage in the typical cell are the mitochondria, damage to which results in decreased energy production, an increased damage rate to various cytosolic proteins, and a shift in the proportion of damaged or erroneous proteins present in the cell. The net result of these insults is the breakdown of cellular homeostasis, In effect, the stability of the cell is undermined by the instability of one of its major components (the mitochondria), which initiates an eventually disastrous positive feedback cycle of freeradical production and ensuing damage to mitochondria and proteins. The model is obviously an incomplete description of the real cell, as its authors are the first to acknowledge. Yet the fact that the model yields predictions regarding changes in the mitochondrial and protein populations which are entirely consistent with the empirical data described in the preceding chapters is a sign that this simulation is on the right track. The MARS model also provides us with a detailed look at the processes underlying the failure to repair, which is the critical step in Lamb’s model, described above. In that sense, this is an elaboration of the former model. Given this simulation, is it possible to avoid cell death by devoting more energy to repair pro-

cesses? Kowald and Kirkwood have done this simulation and find that a virtual immortality might be achieved if 55% percent of the total energy of the simulated cell were devoted to repair and/or prevention of free-radical and oxidative damage. Below that level of expenditure, the loss of cellular homeostasis was inevitable, although it could be significantly modulated by various treatments. For example, simulating the effects of dietary restriction by decreasing the rate of free-radical production led to a significant delay in the time at which cellular homeostasis was lost. The specific value of the number of calories needed for indefinite repair may not be as important as its general magnitude. The 55% value is so high that it becomes questionable whether an organism that devoted so much energy to long life would also be able to participate in other energy-requiring activities, such as reproduction. Thus, this computer simulation brings us back to the evolutionary precepts on which modern biogerontology is based. Why are these outcomes so important when we already know these results to be empirically true? The point is that the computer simulation accurately reproduced the empirical results, and this fact testifies to the accuracy of the equations and of the postulated interactions. The way is now clear to use computer models to suggest

488 Chapter 14 A Theory of Aging over the Life Span

8x105 AM05 6x105

M15M05 HalfM

4x105

ATPSOD RadC

HalfP RM05

SODc ReRc

RiRc

Scaling: SODc M0 Rad0 RadC ATPdt ATPSOD HalfP HalfM ReRc RiRc M15M05 RM05 AM05

= = = = = = = = = = = = =

1 1000 20 1E5 50 500 400 0.1 34E5 7E5 1E6 10 60000

Rad0

2x105 ATPdt M0 1x107

2x107

3x107 Time (sec)

4x107

5x107

Figure 14.2 A computer simulation showing the changes in selected variables in the MARS model during the course of a cellular collapse casued by insufficient free-radical protection. The relative scaling of each of the 13 variables in this simulation is given; note that there may be as much as six orders of magnitude separating any two related variables. As time passes the cell shows a decrease in the proportion of undamaged mitochondria (M0), coupled with an increase in the proportion of damaged mitochondria (M15M05). The free radical production in intact mitochondria (Rad0) decreases as a result, while the radical concentration in the cytoplasm (RadC) slowly increases. This leads to an increase in the half-life of mitochondria (HalfM) and to the increase in radical production/mitochondria (RM05) and the decrease in ATP production/mitochondria (AM05). The increase in the raio of erroneous (i.e., damaged) ribosomes to correct ribosomes (ReRc) is presumably related to the significant increase in the ratio of inactive to correct protein (RiRc) and to the increased half-life of proteins (HalfP), reflecting the difficulty in degrading aberrant proteins. The ATP consumed/protein molecule synthesized (ATPdt) slowly decreases as well, although more energy is expended on antioxidant protection (ATPSOD). There is only an initial and terminal decrease in the level of correct antioxidants (SODc). (After Kowald and Kirkwood 1996.)

relationships we did not know beforehand. The value of the quantitative MARS model over the qualitative model of Lamb (1977) does not lie in its conceptual complexity, for they are both subtle, and they both postulate the same sorts of events and interactions; rather, the value is in the ability of the computer model to expose hidden causal links between diverse phenomena and thereby provide us with valuable information in the design of further studies. 14.1.2.3 Aging As a Hierarchy of Networks

Genes do not act by themselves but rather act as one component of an integrated gene circuit, which is thought to function in a manner analogous to a logic circuit. The early work done with lambda phage, Escherichia coli, and Saccharomyces cerevisiae supported this view and demonstrated

the biological role of various regulatory circuits (Ptashne and Gann 2002). At least 95% of singlegene mutations in yeast affect not only their own expression but also alter the expression of many other genes. These data suggest the presence of what has been called a gene expression network— a set of interactions between genes (or gene products) that together dictate gene expression in the cell (Featherstone and Broadie 2002; Ideker et al. 2001). Such networks exist in multicellular organisms as well. Davidson (1999) and colleagues (2002) have characterized the gene expression network responsible for the temporal and spatial development of the gut in the sea urchin. In this case, once the early interactions are specified, the network is so organized that it locks in its successive regulatory states and proceeds inexorably forward with development of the gut and the setting up of the input signals for the succeeding

14.1 Biological Theories of Senescence

developmental steps under the control of other linked networks. This stability and forward progress are what enable these gene expression networks to produce what we perceive as a program of embryonic development. Senescence is not a programmed response in the same sense as embryonic development. Senescence has many characteristics of a disorderly or stochastic progressive loss of function. So the complex circuit arrangements of genes involved in development cannot serve as a direct simulation of aging. We now have a problem: how can we explain the progression of the aging process— its apparent similarity among all members of the species—without a genetic program? It is possible to explain the aging progression using neither time nor program, but rather using the concept of a stochastic (but not random) breakdown of a hierarchial system of networks. Most recent work has focused on characterizing the intracellular networks characterizing gene– gene and protein–protein interactions, and I touched on this in the discussion of signal transduction pathways (STPs) in chapters 12 and 13. But the loss of body function involves more than one cell; it involves the interactions of that cell with others in the same tissue, with other tissues and organs, and with the neuroendocrine–immune (NEI) system. Each of these levels can be viewed as composing its own network, as discussed regarding the neuroendocrine system in chapter 13. Thus, a full description of senescence would require us to organize our knowledge within the minimum framework of a hierarchal four-level series of connected networks (cell-tissue-organsNEI). Representation of the complex topologies and interconnectedness of cellular networks presents a difficult visualization problem for bioinformatics (Sharom et al. 2004); representation of a four-level network is probably beyond our present capabilities. But it should remain our goal. In the meantime, let us review what is known of the intracellular gene interaction networks and how they might relate to senescence. Gene Interaction Networks. The networks involved in aging are nonhomogenous, by which is meant that in such networks, most genes have but

489

few connections, while others are highly connected. Such a network is also called “scale-free,” which means that the average number of steps between any two arbitrarily chosen genes involves relatively few connections (Albert et al. 2000). A simple diagram of such a model is shown in figure 14.3. We are surrounded by such networks: the Internet and the World Wide Web, for example, are silicon versions of such a nonhomogeneous, scale-free network. Hub-and-spoke airlines are real-world examples. Such networks are fundamentally modular in that the interacting hubs have a tendency to combine into subgroups within which they are highly interconnected but which have relatively few links to outside hubs. The hallmark of such networks is that they are resistant to the loss of any ordinary component but they are very vulnerable to the

X

A B

Y C

Figure 14.3 A model of the architecture of a scale-free network. There are 27 hubs or nodes with 32 interactions in all, for a mean of about 1.2 links/node. Nodes A, B, and C are highly connected and have about 10 links/ node. These three highly connected nodes and their links each constitute a submodule within the network, and each of these is moderately connected to the other with about 4 links between each pair. There is one input link from node X and one output link to node Y.

490 Chapter 14 A Theory of Aging over the Life Span in the fruit fly by the Groucho (gro) protein and C-terminal binding protein (CtBP). Note that the gro protein predominantly regulates the genes in groups 1 and 6; the CtBP protein predominantly regulates the genes in groups 2 and 3; and both the gro and CtBP proteins cooperatively regulate the genes in groups 4 and 5. Clearly the gro and CtBP proteins each represent highly connected hubs, while the other genes have a much lower level of connectivity. The gro and CtBP subsets are characterized by a high internal connectivity within each subset but a low connectivity between subsets. Single events disabling the genes coding for the gro or CtBP proteins would isolate their respective subsets from the information flowing over the rest of the network, the subsets would soon lose their effectiveness, and the cell would undergo a loss of specific DNA repair functions. The cell would, in other words, begin to undergo the loss of function characteristic of senescence. Other gene systems, such as the mitogenactivated protein kinase module described in

loss of any highly-connected component (hubs or nodes). Thus, the loss of even one or two highly connected hubs will have major effects on the interconnectivity of the network and could even fragment one network into two or more pieces. Storm closures of key hub airports rapidly immobilize entire airlines, including their operations in “spoke” airports enjoying lovely weather. For the World Wide Web, it has been estimated that loss of less than 5% of the hubs will lead to the loss of network connectivity and the abrupt failure of the system (Albert et al. 2000). When one assays which genes or gene products preferentially interact with each other and then displays the results in a graphical format, one obtains a diagram of a gene interaction network such as that shown in figure 14.4. This study of the genome and proteome interaction data of Drosophila (Giot et al. 2003) shows that empirically determined gene networks do resemble the theoretical construct of Figure 14.3. Figure 14.4 depicts the regulation of transcription repression

1

2

4 6

5

3

Figure 14.4 A detail from the protein interaction maps of Drosophila. Two transcription regulatory circuits involving the well-characterized co-repressors CtBP (C-terminal binding protein) and Gro (groucho) are depicted. The CtBP repressor mostly regulates subsets 1 and 6, while the Gro repressor mostly regulates subsets 2 and 3. Subsets 5 and 6 seem to be under shared control. The binding partners of the two co-repressors are largely non-overlapping, which concurs with existing evidence that CtBP and Gro repressors act as hubs that independently mediate short- and long-range transcriptional repression of many spokes. (After Giot et al. 2003.)

14.1 Biological Theories of Senescence

Werner’s syndrome gene, WRN, is a highly connected hub in that subnetwork (de Magalhaes and Toussaint 2004). The concept of hubs and spokes seems to be empirically validated. The foregoing discussion may lead one to suppose that the genetic architecture of all longevity mechanisms must resemble that shown in figure 14.5. Lieb and Mackay (2002) used quantitative trait loci (QTL) analysis to identify the chromosomal loci significantly involved in the extended longevity phenotype of 12 different Drosophila strains. Their data showed that, although each of the 12 strains possesses a functional ISP, each use different upstream and downstream genes to achieve the same longevity phenotype. For example, the QTL region on chromosome 3 at 66–68 (which includes the small heat shock genes and CuZnSOD gene and is the major QTL in the long-lived La strain described in figure 7.14) is used by a majority (8 out of 12) of the strains, but then each one of these uses a different pattern of other genes in addition. Thus we cannot talk of there being only one genetic method of extending longevity, but we can talk of there being multiple mechanisms probably arranged about a common core such as the ISP. So the ISP core of figure 14.5 would be constant, but there would be significant differences in the identity and connectivity of the downstream genes regulated by the dFOXO transcription factor. The variable nature of aging probably arises in part from these unique patterns of genetic

figure 13.7, are also characterized by high internal and low external connectivtiy. The hippocampal subsystem of the NEI was described in similar terms in chapter 13. Chapter 7 presented data showing that the insulinlike signaling pathway (ISP) plays a key role in translating environmental signals on the availability of food into regulatory signals that would activate or repress specific sets of genes. If those data are correct, and if our views of gene interaction networks are correct, then the ISP should also be characterized by high internal and low external connectivity. If one examines the genes and proteins directly associated with aging in mice and humans, one can construct a protein–protein interaction map such as that depicted in figure 14.5. This drawing represents the network associated with the growth hormone/insulinlike growth factor-1 (IGF-1) axis. These gene products include those composing the ISP and constitute a sort of multinode hub connecting the low connectivity “lateral arms” to each other, and of course to other cell functions. Because of this connectivity, signals reaching the ISP via the IGF-1 protein, for example, are rapidly disseminated throughout the cell and change it from a progrowth pattern of gene expression to a promaintenance pattern or vice versa. This network provides us with the mechanism by which a prolongevity intervention such as caloric restriction is able to bring about widespread coordinated systemic changes in cell function. A similar analysis involving the mammalian genes involved in DNA repair shows that the

YW HAZ IGF 1 PLA U

LRP 2

PIK3 R1

IRS1

GRB 2

IRS2 IGF1 R

INS

INS R IGF 2

SHC 1 PTP N1

491

STA T5B

GH RH

Il7R

GH1 1

HES X1

GHR PTP N11

STAT 5A

PRO P1 NR3 C1

STAT 3 JUN

POU 1F1 CRE B1

PXR 1

Figure 14.5 A gene interaction diagram of genes important in regulating mammalian longevity. The dark symbols represent genes with a connectivity ≥ 5, and so they represent the hubs in this diagram. Note that many of the hubs are actually part of the insulinlike signaling pathway. It is this high connectivity that leads to the appearance of coordinated effects of the activation/repression of this system on the body. (Reprinted from Magalhaes and Toussaint 2004, with permission of J. Magalhaes.)

492 Chapter 14 A Theory of Aging over the Life Span polymorphisms particular to each individual’s genome, and in part from the dynamic nature of regulatory gene networks. A computer-based analysis of genome-wide transcriptional regulatory information and gene expression data for multiple environmental conditions in the yeast revealed that diverse stimuli caused the transcription factors to alter their interactions to some degree and thereby effectively rewire the network (Luscombe et al. 2004). The network is stabilized, however, by the fact that evolutionary studies of the yeast genome conclude that the amount of expression polymorphism among genes that encode interacting proteins is quite similar (Lemos et al. 2004). This suggests that genes whose products are involved in complexes are precisely coregulated with their interacting partners and have transcription and translation properties that minimize noise in the system. Important components of the interaction network vary together if at all. Taken together, we can view the gene interaction network as being inherently stabilized by its interacting components but being altered by the polymorphisms unique to different individuals as well as being affected over time within the same individual by alterations in the strength and nature of the input signals to the insulinlike protein core. If much of the function of a gene interaction network is to regulate the activities of its component parts, then fragmenting the network into isolated pieces will interfere with these regulatory processes and lead to alterations in gene expression patterns and a concomittant loss of buffering capacity (Lipsitz 2002). If the hubs are actually highly connected genes with key roles in maintaining processes essential for adult somatic maintenance, then the loss of even one or a few hubs would disrupt the gene expression network and probably shift the molecular and physiological state of the cell toward a state of lower function. As individual cells lose hubs and shift to individual states of lower function, the tissue they comprise begins to show physiological signs of aging. It is not required that all cells in a specific tissue of one individual exhibit the same damage phenotype. Different cells might well fragment the network in different ways. The result would

be a population (cells, tissues, etc.) of individuals that all possess the same gene expression networks but not all of which would damage the network in the exact same manner. The net result would be a mechanism that provides for similarity while allowing individuals to age in a contingent manner. What damages the hubs? There are different ways in which stochastic damage can exert its effects on the hubs, and the prior chapters have described many of them. Computer simulation experiments led Promislow and Pletcher (2002) to suggest that mutant alterations to even simple networks would result in large changes in age-dependent phenotypes. This is now supported by empirical data showing that Mendelian disease genes preferentially inactivate the most highly connected hubs and exert their ill effects by causing a characteristic fragmentation of the network (Bortoluzzi et al. 2003). The altered gene interaction effects could yield complex (i.e., individualistic) compound phenotypes not immediately predictable from the action of the individual genes. One implication of this finding is that, although most network alterations would degrade the organism’s function, some alterations alone or in combination might lead to maintenance or even increase function. A mathematical analysis of yeast aging led to the development of a model in which the only variable was increasing change, which led to alterations in homeostasis (see figure 10.10). The model is not deterministic but is nonlinear; it allows for stochastic events and is dependent on the prior state of the system. The scale invariance of the equation indicates that loss of homeostasis would occur at all levels of biological organization. The model leads to a population undergoing epigenetic stratification and giving rise to a normally aging subset and a slowly or non-aging subset. Most important, the model “indicates change as a cause rather than a consequence of the aging process” (Jazwinski et al. 1998, p. 579). This analysis is fully compatible with the view of aging as arising from the stochastic alteration of a gene network. Perhaps the locus on chromosome 4, which is reported to be involved in the longevity of centenarians (see

14.1 Biological Theories of Senescence

chapter 8), might act via an allelic variant that interacts with a few highly connected genes (hubs) to allow an increased (or perhaps a more slowly decreasing) stability of the network. This interpretation leads explanation for the late age decrease in age-specific mortality rates based on stochastically induced heterogeneity (Arking and Giroux 2001). Thus, a stochastic (but not random) fragmentation of the gene expression network, when coupled with the complex and contingent gene interaction effects arising from such fragmentation, could be a nonprogrammatic, nonselectable mechanism by which organisms progressively lose various physiological functions at different times and in different combinations. Ideker et al. (2001) showed that deleting one gal gene involved in galactose metabolism brings about significant and different effects on the expression of other connected genes and on multiple physiological functions. Deleting other galactose genes yields radically different effects on the expression of the other genes and on functional physiology. The different consequences of two gene deletions on the organism can be interpreted in terms of the different connections binding each gene to the rest of the network. Giroux and colleagues (2003) focused on the nature of the response network of yeast nuclear and mitochondrial genes involved in the oxidative stress-resistance phenotype. The genes that responded to the treatments were those involved in the biochemical or metabolic pathways that were experimentally modified. Exposure of the yeast cells to different concentrations of oxidative stress-inducing compounds qualitatively altered the cell’s response, suggesting that the network is capable of sudden large changes of state in response to small changes in the concentration of the toxic substance. This may be the mechanism underlying the phenomenon whereby chronic insults eventually bring about a sudden loss of function. Aging was not examined in Giroux et al.’s study. However, oxidative stress is intimately involved in the aging process, and data suggest that oxidative stress is the major damage mechanism in the cell (Finkel and Holbrook 2000). Given the highly conserved nature of the oxidative stress

493

mechanisms in animals, it is likely that something resembling the yeast oxidative-stress generesponse network is operative in other organisms, including mammals. Indeed, the oxidative stress response of aging mice appears to suffer from early or upstream breakdowns in stress signaling pathways (Edwards et al. 2003; see table 7.9), and upstream breakdowns would most likely be the result of disconnecting a hub from its particular subset of spokes. A theoretical study suggested that aging networks would be affected by a loss of connectivity (Zhu et al. 2003). Other Intracellular Networks. Above the gene interaction network lies the protein interaction network, of which the STPs compose but one aspect of the proteome. The metabolic pathways of the cell, involving the various enzyme proteins, constitute another specific aspect of the proteome. An analysis of the interactions among 3575 yeast proteins revealed that proteins associated with senescence show a higher number of interactions (e.g., connectivity) than do proteins associated with non-aging traits (such as salt tolerance, cell size, UV sensitivity, and DNA silencing; Promislow 2004). This implies that longevity, more than specific cell functions, depends on a coordination of activities across a diverse spectrum of pathways. The discussion of the genetic mechanisms empirically involved in longevity (see chapter 7) led to a similar conclusion. Epicellular Interactions. A somewhat different approach to modeling the aging process was provided by Jazwinski (1996), who incorporated in one model all of the different epicellular relationships known to be important in regulating aging and senescence. Figure 14.6 pulls these relationships together into an interesting viewpoint that is not at odds with the models described earlier. The life maintenance reserve is defined by Jazwinski (1996) as composing both the organism’s metabolic capacity and its ability to respond to stress. Together, these are considered to be a genetically determined functional potential that allows the organism to survive at least until reproductive maturity. Metabolism results in the generation of pro-oxidants, which

494 Chapter 14 A Theory of Aging over the Life Span

Interacting epigenetically

Determined genetically Metabolic capacity

Stress responses

Environment during development • External • Internal (hormonal)

Life maintenance reserve Aging

Genetic instability

Adult environment • Damage • Stress • Disease Dysregulation

Death

Figure 14.6 One view of the determinants of aging and longevity. Phyiological relationships, not specific molecular mechanisms, are shown. Metabolic capacity, stress responses, and environment during development all contribute to the life maintenance reserve, where they interact. Genetic instability is genetically determined, results in genetic alterations, and is influenced by environmental factors. Although senescence is a cellbased process, it plays out at all levels of biological organization. (After Jazwinski 1996.)

call forth the response to oxidative stress. The life maintenance reserve is diminished by mitochondrial damage, and its potential expression can be significantly modulated by internal and external environmental factors. The postdevelopment environment exerts its effects through damage, stress, and disease. Epigenetic changes affecting the chromatin that occur during the lifetime of a somatic cell result in small changes in gene regulation that, if the cell enters a positive feedback cycle, can result in gene dysregulation. Together, the processes of gene dysregulation and environmental insults lead to increased stress responses. Such increased stress is deleterious in its own right and affects the organism’s ability to respond effectively to acute stress. The model in figure 14.6 postulates that the factors that affect aging are genetic, epigenetic, and environmental in origin,

whereas the factors that limit longevity are metabolic capacity, efficiency of stress response, and attenuation of signal. There is no contradiction among these models, for each is attempting to describe the interactions of positive and negative forces interacting over several different levels of organization. Each one is accurate; taken together, they afford us a good conceptual view of the aging process in all its complexity. We know that there are positive feedback effects from one stage to an earlier stage and that such effects accelerate the process of aging. An example of this is when damage to certain key cells, such as those in the pituitary or the hypothalamus, has cascading deleterious effects on the undamaged cells regulated by these key cells, whose decreased feedback functioning might further decrease the physiological functioning of the neuroendocrine cells. We need to be able to view the senescent process in four dimensions. The convergence of ideas based on independent data sets suggests that we are progressing toward a more realistic portrayal of the mechanisms underlying senescence. The systems analysis approach is better able to adjust theoretical constructs of gene action with the multiple dimensions of the aging process. The models of Lamb (1977) and of Kowald and Kirkwood (1994, 1996) represent the initial attempts to construct a systems analysis approach to understanding aging and senescence. The current hierarchial network model represents the next step in our attempt at understanding.

14.2 Biological Theories of Longevity The previous secton with its focus on cell-level biological processes applies to all organisms. But is it a full and complete explanation of longevity and senescence in all organisms? It is not. Although theories of senescence can explain why both mice and humans age and lose function, they provide no obvious explanation as to why a mouse lives but 3 years while a human lives for 100 years. It was pointed out some years ago that

14.2 Biological Theories of Longevity

aging and longevity are fundamentally different concepts (Cutler 1976; Sacher 1978). The evolutionary theory of aging is based solely on individual natural selection (chapter 4). However, a complete understanding of longevity will require in addition some knowledge of evolutionary processes such as sexual selection and kin selection, as well as a bioeconomic analysis of age-structured behaviors (e.g., division of labor, intergenerational transfers; Carey 2003). All sexual (and some asexual) organisms must interact with another member of their species to reproduce, and the species’ adoption of one or the other broad life-history strategy (e.g., r or K selected as in table 4.1) dictates much about their consequent longevity trajectory. But we are now asking why the biology of longevity and senescence as explained it to this point is not a full and complete explanation of longevity and senescence for all species which may have adopted a broadly similar life-history strategy. One approach to answering this question is to survey the life histories of taxonomically diverse species that are long lived, with the idea of determining whether they differ in some important variable. The results of such a survey showed that long-lived species generally adopted one of two different life-history/ecological strategies (Carey 2001, 2003). One group of long-lived animals lived a more or less solitary life in environments where resources were either scarce (e.g., deep water) or unpredictable (e.g., deserts). The other group exhibited extended parental care and/or lived in groups with complex social behavior. The extended longevity of the solitary species probably arose via the action of individual natural selection so that the individual animal can live long enough to intercept some of the rare food resources and thus be able to reproduce, as described below. However, the existence of the social group suggests that our discussion of aging to this point is incomplete, for it has not provided any explicit tools or concepts for demonstrating that social effects might be of importance in fostering the evolution of a biological trait such as longevity. Several reviews examining this particular topic are available, and

495

the following discussion is based on those sources (Carey 2003; Carey and Tuljapurkar 2003; Wachter and Finch 1997).

14.2.1 Environmentally Selected Extended Longevity Animals that live in sparse or unpredictable environments can only survive in an evolutionary sense if they successfully reproduce. Successful reproduction requires them to obtain sufficient extra resources (i.e., energy) from the environment to ensure that their matings will be productive and that their young will not only be born but that a sufficient number will survive to sexual maturity. Only then can they be said to have successfully reproduced. Reproducing in the absence of necessary resources yields a high probability of unsuccessful reproduction and would also likely increase significantly the chances of the parent not surviving the reproductive attempt due to the wasting of its resources in a resourcepoor environment. Successful reproduction requires that the animal reproduce only under conditions that ensure success. If these conditions are met only rarely or sporadically, then the animal must live long enough to live through at least one such season after it becomes sexually mature. The energy allocations to reproduction and somatic maintenance (figure 4.5) can be modified if the alteration results in a higher Darwinian fitness than if they are not modified. The animal’s best strategy to increase its Darwinian fitness under resource-poor conditions is to shift its current energy resources from reproduction into somatic maintenance because it must first survive until the environment becomes favorable to reproductive success. At that time, it can shift its energies to reproduction. The shift of energies into somatic maintenance means that the animal will likely evolve one of the gene-based longevitydeterminant strategies outlined in chapter 7. Which one it adopts will depend on the details of the animal’s genetic, life-history, and environmental parameters. The point is that the biological descriptions offered in chapter 7 are adequate

496 Chapter 14 A Theory of Aging over the Life Span to describe the evolution through natural selection of extended longevity. Examples of animals that have undergone environmentally selected extended longevity include tortoises, sea turtles, deep-water tube worms, some birds, and some beetles (Carey 2003).

14.2.2 Socially Selected Extended Longevity Species socially selected for extended longevity include those that exhibit extensive parental care and live in an interdependent social group. Examples of such species include humans and other primates, elephants, killer whales, dolphins, naked mole rats, brown and vampire bats, parrots, hornbills, albatrosses, and social insect (termite, ant, bee) queens (Carey 2003). The extended longevity of animals in this category comes about through the actions of natural selection as well as sexual selection and kin selection. More variables are operative in this case, and it is more complex. There are three foundational principles that guide this analysis (Carey, 2003): (1) evolutionary theory of aging, (2) intergenerational transfers, and (3) division of labor. The first principle was thoroughly discussed in chapter 4 and is alluded to by the term “natural selection.” Investment of resources in reproduction can be extended to investment of resources in parental care of the offspring, as well as to investment of resources in somatic structures that will permit greater Darwinian fitness later. If you think about the vocabulary used in the preceding sentence, you might conclude that economists and evolutionary biologists speak the same language, and you would be right. The combination of capital investment theory from economics combined with life-history theory from biology led Kaplan and colleagues (2003) to their “embodied capital theory,” which views the processes of growth, development, and maintenance as investments in stocks of somatic (embodied) capital and the production of new offspring as a return on investment. I use the evolution of human longevity as a case study with which to explicate this scenario is some detail.

14.2.2.1 Factors Influencing the Evolution of the Human Life Span

Diet and Intelligence. We are human in large part because our ancestors specialized in an ecological niche that emphasized the consumption of rare but nutrient-dense, low-fiber foods rich in protein and fat. We evolved as hunter-gatherers, and more than half the calories of that diet are typically derived from meat (Kaplan et al. 2000). This nutrient/ecological niche is difficult to exploit, and it requires the ability to learn and reason so as to efficiently develop the large skill set necessary to survive in this niche. Exploitation of this niche strongly selects for increased brain size. This results in a longer developmental time, but our gestational period is constrained by the fact that the infant’s head must fit through the birth canal. The compromise solution to these apparently contradictory conditions is that we are born premature and require extensive parental care. This imperfect solution then begets another round of adaptations needed to achieve the higher Darwinian fitness implicit in a larger brain size. Analysis of the human fossil data by both Cutler (1975) and by Caspari and Lee (2004) suggests that the long life characteristic of humans began with our (post-Neanderthal) ancestors in the Early Upper Paleolithic. The increased longevity both flowed from the distinctive niche our Homo sapiens ancestors began to exploit and contributed to more effectively making use of that niche. As I show below, longevity may be necessary for the transgenerational accumulation and transfer of information that allows for complex kinship systems and other social networks that are uniquely human. In this sense, then, longevity allowed us to become modern humans. Intergenerational Transfers of Food and Information. It takes humans a very long time until we can support ourselves, whether in a huntergatherer environment or in the modern world. As shown in figure 14.7, individuals in a particular hunter-gatherer society (the Piro; see Kaplan 1994) consume between 1000 and 4000 calories/ day, depending on their age. Children do not make any caloric contribution to their own nour-

14.2 Biological Theories of Longevity

10000 9000

Food consumption Food production Net productivity

Calories/day

8000 7000 6000 5000 4000 3000 2000 1000 0

497

Net surplus

Net deficit

-1000 -2000 -3000 0

5

10

15

20

25

30

35

40

45

50

55

60

65

Age

Figure 14.7 Food production and consumption by age in a hunter-gatherer tribe (the Piro) practicing a mixed economy of swidden horticulture, hunting, fishing, and gathering. Data from both sexes are combined. Note that children are not self-supporting until the age of 20, after which they generate a surplus. (After Kaplan 1994.)

ishment until they are 7 years old or so, and their net production continues to fall until about age 16 years. Only then does it begin to climb. Offspring do not become nutritionally self-supporting until they are about 20 years old. Until then, they are dependent on intergenerational transfers of resources from their parents. At that point, their increasing skill allows them to generate a significant surplus of calories. The fact that they reach their peak food production at age 55 indicates that their productivity depends more on skill and learning than on simple physical strength. Surplus food can now be transferred to the next generation. Independent studies on two other hunter-gatherer societies, as well as interspecific studies on chimpanzees, lead to the same conclusions (Kaplan 1997). This behavior may seem only natural or moral to us who live within a social system based on sharing. But most animals, including humans, are not altruistic. And morals apparently have precious little to do with Darwinian fitness, which is the final value that counts in evolution (but see Bronowski 1956, for an incisive essay on the ethical basis of human society). What are the evolutionary principles that would encourage such an

apparently altruistic transfer of resources? The transfer of resources from older to younger individuals is usually done only for genetic relatives and can be viewed as an investment by which juvenile mortality is reduced. During their long developmental period, children are taught how to gather and hunt by their parents, their sibs, their extended family, and the rest of the community. Learning takes time. Children do not attain even minimal plant-gathering skills (i.e., acquiring ~10% of the adult productivity) until age 10, and that attainment level for hunting is not reached until age 15 (Kaplan 1997). Social learning and community-level organization increase the volumes and pathways by which information is transferred from one generation to the next, and this broadband transmission decreases the learning time and increases the effectiveness of each lesson. Successful human reproduction requires the transfer of genetic, energy, and information resources to the next generation. This potential increase in Darwinian fitness is realized when the now healthier offspring survive to adulthood and reproduce. It also requires an extensive amount of parental affection, touch, playing,

498 Chapter 14 A Theory of Aging over the Life Span and just plain love. Recall that minimal handling of rat pups by their handlers had dramatic effects on their stress levels, memory, learning performance, and exploratory behavior (see table 13.5). The child must learn well and efficiently if he or she is to be an effective adult. The parent–child bonding stimulates the child’s brain, encourages the formation of complex neural circuits, enhances neurogenesis now and in the future, prepares the child to speak, and generally aids the ability to solve new problems. Children raised in the absence of such bonding are unfortunate beings who will have a difficult time being an effective adult. Bonding likely increases the Darwinian fitness of the parent and certainly increases that of the child. For optimal success, human development requires the presence of the child’s parents for about 20 years or until the child is self-sufficient. This, in turn, requires that the parents should live for about 20 years past the time of last reproduction. Menopause in humans occurs at about 45–50 years, and the mortality rate in this hunter-gatherer group does not significantly increase until about age 68 (Kaplan et al. 2003). These are not likely to be unrelated facts. Elsewhere in this book I briefly discussed the “grandmother hypothesis,” which speculated that our long life arose from the selection of long-lived grandmothers who could then help feed and raise their grandchildren. This hypothesis suggests that the increased productivity of adults relative to children is due primarily to the latter’s greater strength. The analysis of Kaplan et al. (2000) suggests an alternative hypothesis: Our long life may be the result of selection on adults living long enough to support all of their children until the offspring can fully support themselves. This parental selection (or embodied capital) hypothesis suggests that the increased productivity is due to the learning-intensive nature of human foraging (Kaplan 2002). Of course, both scenarios may have played important but unequal roles in our evolution. In fact, Hrdy (2002) has pointed out that there are a variety of evolutionary pressures that act on mothers, and so we need not choose between them here. However, the selective bene-

fit to the parent of nurturing a grandchild is only half that achieved by nurturing one’s own child. The parent gets a bigger payoff by saving one child than by saving one grandchild. This reproductive calculus is consistent with the observed facts of human behavior and may suggest that the parental selection/embodied capital hypothesis is the more important mechanism by which we became modern human beings. Division of Labor. Division of labor results in a specialization of function, which, by spreading different skills among different individuals, reduces learning costs and increases skill and productivity output. The more or less general division of food acquisition between hunting for meat and gathering of plant foods between the sexes may have other purposes, but it also allows each person to fully develop a limited set of skills. Kaplan (2002) points out that it is the protein- and lipid-rich meat that both supports and requires the extended juvenile period of humans, and this suggests that it is not just calories but rather nutrition which must be considered when evaluating our evolution. The gender-based division of labor with regard to food gathering also increases the interdependency of individuals on each other and on the group as a whole and strengthens the bonds holding the group together. 14.2.2.2 Demographic Transition

Once the confluence of factors is established, even in an early and inefficient mode, it begins to reduce the mortality rate in different aspects of the human life. These attributes are positively selected for, and so a positive feedback cycle is established whereby reductions in juvenile mortality result in fewer births needed to maintain the group at any desired or optimal size (i.e., to maintain the parent’s Darwinian fitness at some high level). Mothers with fewer offspring are generally healthier because they have had to transfer their somatic resources to fewer fetuses, and so that non-transferred energy is now available for them to use on somatic maintenance

14.3 Tying It All Together

without any penalty to their Darwinian fitness. They can also invest more of these spared resources in their offspring, raising a smaller number of higher quality offspring than mothers with many offspring. An increase in offspring quality decreases juvenile mortality even more. It also allows for the development of better adult skills and thus the accumulation of more resources than less well-educated adults. The increased available resources can be funneled back into improving the offspring, thus creating a virtuous positive feedback cycle in which reduced family size and health, health and longevity, longevity and productivity, productivity and wealth, longevity and knowledge, and knowledge and productivity are positively associated with one another. The model organisms with extended longevity discussed in chapter 7 are unable to modify their environment or their reproductive strategies, and so all the prolongevity adaptations must take place within their own cells. Humans, however, have adopted a life-history/nutritional strategy that emphasizes intelligence and manipulation of the environment. These traits, even when rudimentary, enabled humans over time to alter their environments and their reproductive schedules. Our ability to reconstruct our environments to our liking and to transfer resources to maintain and enhance that ability has allowed selection to reduce the extrinsic mortality rates over every part of our life cycle. Our environment-dependent mortality rates are plastic and decreasing.

499

Carey (2003) uses the term “aging theory” to delineate the proximate causes of senescence and “longevity theory” to designate the ultimate causes of longevity. A complete view of human aging requires the fusion of these two views. The hierarchal network theory of aging, presented in the first part of this chapter, was based entirely on the laboratory models and may be fairly viewed as representing (albeit crudely) our current view of aging theory, as used by Carey (2003). The left-hand side of figure 14.8 depicts a version of that hierarchal aging network, and the resource transfer and decreased mortality processes on the right-hand side represent the interaction of the longevity theory social factors with the biological factors. The extra resources permit increased stability of each network level, leading to healthier offspring and mortality reductions, which increase the amount of resources available for transfer, and so on. Thus the evolution of human longevity has occurred within a framework generally applicable to all animals but with certain resource transfer mechanisms that amplify the basis biological processes. Such transfer mechanisms are not unique to humans, for they are certainly found in other primates and probably occur in other social animals. In fact, the study of the social insects (ants, bees, termites) is advocated on the grounds that their analysis will give rise to a better understanding of the interactions between reproductive and nonreproductive units (be they cells or individual organisms). The need to efficiently transfer information probably supported the evolution of language, which of course resulted in another positive feedback system (see Dennett 2002).

14.2.3 Implications of Longevity Theory The model organisms that we depend on to illuminate the genetic mechanisms underlying longevity determination and senescence do not owe their longevity to social mechanisms. This means that the information we obtain from those organisms is incomplete relative to what we need to know about ourselves. The models give us excellent information about intra- and intercellular mechanisms but not about interorganismal mechanisms.

14.3 Tying It All Together Figure 1.7 presented an interim integrated theory of the life span in which a functional distinction between the health span and the senescent span was postulated based on the relative activity of maintenance versus senescent processes in each. We elaborated this concept in figure 9.6 so as to

500 Chapter 14 A Theory of Aging over the Life Span

NEI Network

Resource Transfers: Information Energy

Organ Systems Network Decreased Mortality

Tissue Networks

Other Cells

Cell STPs, Gene Networks

Figure 14.8 A conceptual model linking the hierarchal gene expression networks of the neuroendocrine-immune (NEI) system with the resource transfer processes that play an important role in human evolution and development. The transferred resources permit increased stability of each network level, leading to healthier offspring and decreased mortality.

incorporate the molecular and genetic network data. How would we update it to accommodate the social interaction concepts discussed above? Given the preceding discussion of the evolution of human longevity, it is obvious that the focus on the adult portions of the life span omit consideration of our most important learning phase. We must incorporate a developmental phase into this scheme. An attempt to do so is depicted in figure 14.9. I discard none of the biological processes listed in figure 9.6, nor do I omit any of the interactions implicit in the hierarchal net-

work structure of the organism as depicted in figures 14.5 or 14.6. These must be imagined as being implicitly present in figure 14.7 so that we do not clutter the diagram with already presented information. I add to that data the resource transfers of energy and information and expand the health span to include a developmental phase in which much growth and learning takes place. There is some resource transfer into the senescent phase, but this may be counterbalanced by an increased transfer of information and resources to descendants.

14.3 Tying It All Together

501

Resource Transfers Energy, Information

Parent

Developmental Span

Health Span

Senescent Span Adult Span Network Breakdown Mechanisms

Longevity Determinant Mechanisms Brain/NEI

Productivity

Knowledge

Resource Transfer to Offspring

Transmission of information to young

Health

Receipt of Resource Transfer

Darwinian Fitness Learning Mortality

Learning from experience

Longevity despite Mortality = Extended Senescence

Mortality Figure 14.9 An update of figure 9.6, in which I discard none of the information coded in that figure (although space reasons prohibit showing it) but add to it the resource transfers and positive responses as presented in the discussion of the evolution of human longevity. All levels of interactions from the intracellular gene interaction network of one cell to the interorganismal interactions implicit in a social animal living in an integrated community are incorporated.

We differ from other animals not in the possession of these mechanisms but in the extent to which we have pushed them. We started down this path with the assistance of natural, kin, and sexual selection. We have amplified our capabilities with the assistance of cultural evolution. We

are now in the process of supplementing biological evolution with purposive evolution in the form of biotechnology and are now on the cusp of controlling our own longevity. The implications of this control are addressed in the next chapter.

This page intentionally left blank

85

Part VI

Life expectancy in years

80 75 70

What Can We Do about Aging?

65 60 55 50 45 40 35 30 1600

1650

1700

1750

1800 Year

1850

1900

1950

2000

This page intentionally left blank

15

Aging-related Research and Its Impact on Society

15.1 Introduction A reasonable appraisal of the preceding 14 chapters is that aging is a nonprogrammed, almost inadvertent, biological process that can be modified by combined genetic and environmental effects and, as such, represents a legitimate and promising area of biological research. The old views, which held that aging was not susceptible to manipulation and/or was not worthy of serious interest because it could be due to nothing more than the wear and tear of the body, are no longer tenable and have been discarded, as indeed they deserved to be. The detrimental effects of the biological damage caused by senescence can be ameliorated to some extent by various behavioral, social, and cultural practices. Social organisms may receive enough support from their group to enable them to survive past the time they would have been able to survive on their own. Such supportive behavior was selected for in humans by both cultural and biological means (see figure 14.5). But such amelioration, even though it is an intrinsic part of our species’ life-history strategy, does not defer senescence as much as it provides an environment in which older individuals can thrive and still contribute to the clan. Our tradition of resource sharing and our strong bonds to kin made it inevitable that social support of aging individuals became an esteemed societal goal. The hunter-gatherer society in which we evolved is quite different from the individual and consumercentered society we created in the past century.

And so our heritage of intergenerational resource transfers is being pressed by new demographic and social realities at the same time that new scientific realities are bringing forth the promise of intervening in the aging process. Compared to 1998, we have a much increased and more integrated knowledge of the biology of aging, which I hope is evident in this new edition. Preliminary scientific tests aimed at determining whether caloric restriction has the same effect on human biomarkers and longevity as it does on rodents and nonhuman primates have already begun. It is almost certain that various means of slowing or delaying aging, such as pharmaceutical mimics of caloric restriction, will be eventually tested on humans once they pass clinical trials. The biogerontological community has its exhorting optimists and its skeptical realists, and no one can yet tell how the future will play out. In fact, two well-known scientists have made a bet as to whether the maximum life span can be pushed to 150 years by 2150, with the proceeds payable to their descendants in 2150 (McCann 2001). So, while there is not unanimous agreement about the efficacy of such interventions, almost all practicing biogerontologists would likely agree that such interventions will one day be tested, and this alone says much about the state of our knowledge. If successful, such anti-aging technology would be qualitatively different from the procedures used to extend longevity during the 20th century. As shown in figure 2.14, this past century saw public and private health measures used to increase the mean but not the maximum life span. We used culture

505

506 Chapter 15 Aging-related Research and Its Impact on Society and technology in the 20th century to increase the probability that many people would actually achieve the longevity potential inherent in the past human experience. In the 21st century, we will use the new insights and technologies to try to achieve all of the unexpressed longevity potential inherent in the human genome The realization that extending the life span is possible has encouraged much thinking and philosophizing on the matter, all of it speculative by definition, some of it spurious and motivated by greed, but much of it serious and worthy of consideration. Having come this far in our quest to understand the biology of the aging process, it is only reasonable that we should engage in this ongoing discussion. This is a current and timely discussion but it is not new. There have been intimations of immortality, usually couched in religious terms, for as long as there has been civilization. But, as Adams (2004) points out, this discussion took a distinctively different turn in Western civilization about a century ago. The logic of Darwinian thought gave rise to an awareness that social progress was neither an inevitable nor an ordained trend and that chance played a larger role in human history than divine guidance. These thoughts were not then and are not now accepted by all people, but they are so robustly supported by overwhelming empirical data that they are effectively true and so guide our thoughts. The rise of experimental biology and a dawning appreciation of the potential explanatory and manipulative power of modern science led to the idea that humans might be able to use science to control the future—that blind chance might not always hold us at its mercies (Binstock 2004b). The pros and cons of controlling the future were played out over the course of the 20th century, occasionally in scientific writings but mostly within the genre of literature we know as science fiction. This was an extraordinary conversation carried out with gusto by an extraordinary group of men: H.G. Wells, Ilya Mechnikov, Alexis Carrel, Serge Voronoff, Julian Huxley, Aldous Huxley, Olaf Stapledon, J.B.S. Haldane, C.S. Lewis, J.R.R. Tolkien, Robert Heinlein, Isaac Asimov, Arthur C. Clarke, and others. Our pre-

decessors debated the possibility of controlling human aging within the larger context of questioning just how much control we humans should exert over our fate. Should we take this rational step and run the risk of hubris, or should we acknowledge that there are limits to human action and gracefully accept our natural state? Readers who have enjoyed reading or watching The Time Machine, Brave New World, Methuselah’s Children, The Lord of the Rings, Beyond This Horizon, The City and The Stars, or 2001: A Space Odyssey have already been pondering some visionary answers written by these men to this new–old question. Standing on the shoulders of these giants, seeing as they could not the empirical possibility of altering human aging, it is our responsibility to emulate our far-sighted predecessors by carrying on this quest for a truth suitable for our times with the best of our knowledge and sensitivity and passion. In this new phase of the discussion, the anti-intervention position has been championed by Leon Kass, Francis Fukuyama, Bill McKibben and others, while the prolongevity position has been supported by Greg Stock, John Harris, Richard Miller, Michael Fossel, and others. The interviews, articles, and commentary on this contemporary phase of the discussion may be read on the website devoted to aging and public policy: http://www.SageCrossroads.net. In addition, a good review and critique of the scientific and ethical arguments surrounding the use of stem cells to possibly cure or alleviate various agerelated diseases has been provided by Juengst and Fossel (2000) and by Groopman (2004). Many of the same arguments regarding the deliberate extension of longevity are also found in the stem cell debate, so that discussion may help us. In this chapter, I discuss some of the demographic, social, and ethical implications of our present and future ability to manipulate the forces of senescence. The discussion is, of necessity, speculative. Speculations are normally not thought appropriate for textbooks. But what we are discussing is the intersection of biology with public policy. If biologists do not talk and think beforehand about the rendezvous of politics and science, then we will be ceding the field to people whose opinions, whatever they may be, will likely not be subjected to the critical scrutiny character-

15.2 Changes in Age Structure during the 20th Century

istic of modern science. Of course, science and scientists may well be mistaken, and so the following discussion should be viewed as one which is informative but not complete, critical but perhaps not correct. The main purpose of this chapter is to encourage readers to think seriously about the outcome of the impending effect of biogerontology on society and to join in the conversation. First, I describe the changes in age structure that characterized developed societies during the 20th century. Second, I describe the changes in the age structure that will likely take place during the 21st century. Third, I discuss the ethical issues concerning a prolongevity policy. Fourth, I provide links to various speculations as to what might be the likely social effects of a prolongevity policy. We need to be careful to distinguish between and not to conflate two scenarios. The first scenario of “extended senescence” is simply a continuation of the present trends in the developed world and can be viewed as the default outcome. The second scenario of “extended health” would be a qualitatively different way of aging brought about by the proposed anti-aging interventions and can be viewed as the experimental outcome. Some discussions on our future confuse these two scenarios by considering all forms of longevity extension as being equivalent, thus mixing up extended senescence and extended health span. This is not correct. I use the two-scenario format to make clear that they involve two very different strategies of longevity extension. It is likely that our society will be greatly transformed by the end of the 21st century under either scenario. And since the proposed interventions for the second strategy rely on pharmaceutical but not genetic engineering means, arguments based on an aversion to the genetic engineering of humans are excluded from the review, if only as a means of focusing the arguments on the essentials. The demographic and social changes already in play are reshaping our society in a way that disheartens some observers. As much as modern society changed in the first half of the 20th century, it was still recognizable as that which had gone before, but with electricity. It is likely that the society which will exist at the end of the 21st century will support a very different life span tra-

507

jectory than that characteristic of its progenitor. Maintaining our present society unchanged through this 21st century does not seem to be one of our options. The question is whether we can, or should, use the new scientific knowledge to shape a more inhabitable society for our children to live in than will be the case if we do nothing.

15.2 Changes in Age Structure during the 20th Century The extraordinary demographic changes that have already taken place in our society (see figure 2.14) did not result from any alteration of basic aging processes. They are due almost entirely to a mix of alterations in cultural habits and biomedical practices. These transformations are not limited to the developed societies found in North America, northern Europe, and Japan, but are global in their impact. Different societies are simply in different phases of this demographic transition. The relative ratio of birth rates, death rates, and net migration determines the age structure of a population. Advances in public health and biomedical interventions have greatly reduced premature mortality. The enhanced survival of such cohorts leads to a large increase in the number of people who reach reproductive age. When this large cohort, upon reaching adulthood, undergoes a decline in fertility rates, as has occurred in many developed countries, the proportion of older adult cohorts in the population increases relative to younger cohorts. Italy and several other developed countries in Europe now have lifetime birth rates of 1.5 or fewer children per women. This is substantially below the replacement rate (~2.1) which would, in a lowmortality environment, maintain a stable age structure in the population. This means that these low birthrate countries are dominated by onechild families. The United States currently has a higher lifetime birthrate (~2.3), but this includes the substantial contributions of new immigrants; if they are excluded, then the U.S. rate falls to about 1.9. Although relatively high, this is still below replacement levels.

508 Chapter 15 Aging-related Research and Its Impact on Society There are three implications of these numbers. First is the that the projected increase in the world’s population must be the product of population growth in those societies that are not developed and have not undergone the demographic transition. This is substantiated by the data shown in table 15.1. Thus the most effective way to deal with the human overpopulation problem is not to deny health care to elderly people in developed societies (e.g., see Callahan 1987, 1999), but rather to assist these undeveloped societies to grow to the point where they can undergo the demographic transition. However, these less developed regions are changing, for half of the world’s people now live in regions in which fertility is less than

Table 15.1 Comparison of the Age Structure of Developed and Less Developed Countries (Population Numbers in Millions) Less developeda % 42.7

52.2

4.9

0.2

Totals

Developeda

Number

Age

550.6 533.7 525.0 474.7 427.2 414.4 385.3 332.2 278.3 245.0 187.6 148.1 126.2 100.6 71.8 43.8 21.9 8.1 2.1 0.3 6.2 × 10–6 4876.9

0–4 5–9 10–14 15–19 20–24 25–29 30–34 35–39 40–44 45–49 50–54 55–59 60–64 65–69 70–74 75–79 80–84 85–89 90–94 95–99 100+

Number

%

66.6 72.3 25.2 79.9 81.6 82.0 85.5 87.0 91.8 90.5 60.5 85.3 77.5 62.1 61.0 51.9 47.2 34.9 12.8 19.2 12.0 4.4 1.5 1.0 1.1 × 10–4 1193.7

Source: data taken from U.N. Population Division, The 2002 Revision Population Database: http://esa.un.org/unpp/p2k0data.asp. a Both sexes combined. Note the pyramidal shape of the age distribution in the less developed region compared to the rectangular shape of the developed regions (albeit disturbed by the nigh number of baby boomers and the diminishing number of young children).

the replacement rate of 2.1 (Wilson 2004). The high population growth areas of the future are in sub-Saharan Africa and much of the Middle East; as well as in India and China, and it is a race to see whether they can begin or complete their modern development before their population pressures abort the process and bring grief to the entire world. The second implication is that societies with very low lifetime birthrates are undergoing a change in their age structure such that the ratio of older to younger people is increasing (figure 2.15). Because all societies have age-structured societal roles and obligations, often for valid functional reasons (see figure 14.9), a quantitative change in the proportion of people in each class can force qualitative changes in the way in which that society distributes opportunities and obligations. The effect of the “baby boomers” on U.S. society since 1946 and the alteration of traditional retirement policies by the increased health of the recent cohorts of 65–75 year olds are cases in point. The third implication has to do with the drive to reproduce. Not one country that has undergone the economic and social changes inherent in becoming a developed country has maintained a higher-than-replacement birth rate. The finding that people from a variety of different cultures and religious backgrounds are apparently content to trade many children for economic success is quite extraordinary. It suggests (but does not prove) that the “instinct” to reproduce, once it is separated from sexual activity by contraception, is much weaker than was generally assumed; although the travail of childless career women in their 40s suggest that it is by no means nonexistent. An individual’s decision to reproduce seems to be plastic and often can be modulated by the rewards built into the particular social structure. Because the age and social structure of a society that intervenes directly in the aging process (i.e., the experimental outcome) is likely to be different from one that does not (i.e., the default outcome), it is not logical to assume that this low birthrate is an inevitable fixture of all future developed societies. Low birth rate is not the only component responsible for the aging of the population. In most

15.2 Changes in Age Structure During the 20th Century

509

59 years. The life expectancy was never as high there as in the West, but this estimate represents about a 5- to 10-year decrease in life span over a period of two decades or so. As pointed out in table 5.1, the historical record shows that our societies took much longer than two decades to gain an additional decade of life. This is but one example of the dependency of a group’s life span on the social practices of that group. The same situation appears also to have been the case for life expectancy at age 65 in Australia and in eastern Europe, and for an increase in infant mortality in the United States. For humans, longevity depends to a large extent on social support. This statement would sound strange only to those who are unfamiliar with our evolutionary history (figures 14.7 and 14.9). The decreases in fertility, driven by socioeconomic considerations, and the decreases in adult mortality, driven by cultural and biomedical

developed nations, increases in survival at age 65 are now outpacing increases in survival at birth, largely because of the reduction of heart disease and stroke among older individuals (see figure 2.12) and in part because of decreasing exposure to infectious diseases early in life that have a delayed effect on longevity. Therefore, the already low mortality rates for older adults may be expected to continue to decrease (figure 15.1), and it is possible that the proportion of elderly might increase as long as this form of intergenerational resource transfer continues (see figure 14.9). Continuation of the recent increases in life expectancy is probably not a guaranteed phenomenon but depends on the existence of healthpromoting cultural and social factors. If these are not present, life expectancy may actually decline. At the end of the last century, the life expectancy of males in Russia was estimated to be about

1.0 Cohort mortality

Annual proportion dying

0.1

0.01 1751-60 1811-20 1871-80 1901-10 1931-40

0.001

9 -8 85

9

9

-7 75

-6 65

9 55 -5

-4 9 45

9 35 -3

9 25 -2

-1 9 15

5-

0

9

0.0001

Age Figure 15.1 A plot of age-specific mortality over the life span in Sweden over the period from 1751 to 1940 for cohorts of persons born in the specified years (semi-logarithmic plots). Mortality is estimated for the later years of the last birth cohort, assuming that mortality at adjacent age groups has the same relationship as that in the prior cohort, and is represented by a dotted line. Note that the qx values of the 50+ age cohort have significantly decreased as a probable result of decreased infections in early life. Data obtained from the Berkeley Mortality Database. (After Finch and Crimmins 2004.)

510 Chapter 15 Aging-related Research and Its Impact on Society considerations, mean that the proportion of old people may well continue to increase for some time. The significant growth of this segment of the population has important implications for both individuals and families, as well as for public-policy makers and planners. One of these implications will almost certainly be the desire to better understand the biological basis of aging so as to develop interventions. Such interventions will be needed not only to enhance the quality of life of the elderly but also to contain the economic costs implicit in a future where senescent changes are dealt with only after the fact.

15.3 Anti-Aging Interventions 15.3.1 Continuation of Present Trends: Extended Senescence It is true that there is not currently any clinically approved anti-aging interventions proven to be widely effective in humans. Fifty-one biogerontologists (including this author) signed such a public statement in 2002 (Olshansky et al. 2002). But although that statement is true today, it will likely not be true within a few years. Although all biogerontologists believe such effective anti-aging interventions will exist sometime, one group (the “realists”) doubts that interventions will appear in a useful form within the near future. Another group (the “optimists”) believes they will exist in the near future. Let us focus on the logic and facts put forth by each group to get a sense of whether such an intervention is feasible in the near future. If so, then the nearness of the event would justify an extended discussion. The core argument of the realists runs as follows (Carnes et al. 2003; Olshansky et al. 1990, 2001). During the 20th century, relatively small investments in public or private health yielded large increases in life expectancy. But these easy victories in which the environment was altered such that premature longevity was greatly decreased have been won already. They cannot be won again. There is every reason to think that comparable life expectancy increases cannot be

cheaply obtained by similar manipulations of the environment of older (i.e., postreproductive) individuals. The difficulty of continuing to increase the life spans of the elderly will ensure that the increases in life expectancy will slow and eventually come to an end. Figure 2.14 plainly shows that the increases in the mean longevity decreased in each consecutive population sampling during the 20th century. An essential aspect of the realists’ argument is that the continued extension of postreproductive life span in humans is fundamentally contradicted by the evolutionary theory of aging, which posits a close relationship between reproduction and life span (Rose 1991). It is unlikely that we can significantly increase the reproductive span. They conclude that functional senescence begins well before reproductive senescence (i.e., menopause) occurs. Those functional declines signal the onset of the age-related diseases that eventually kill us. It would take an 85% decline in all-cause mortality rates from the 1985 level to yield a life expectancy of 50 years at age 50 (Olshansky et al. 1990). Such a decline is beyond our present capabilities, and so they conclude that, “Barring major advance in the development and use of life extending technologies or the alteration of human aging at the molecular level, the period of rapid increases in life expectancy in developed nations has come to an end” (Olshansky et al. 1990, p. 637). Their analysis suggests that there is a lower limit to death rates, and thus an upper limit to life expectancy, and although that calculated limit of 82–97 years seems close to where we are today (the white female life expectancy at birth is now 79.9–82.0 years, depending on the actuarial assumptions used), that deceptively small gap will be difficult to close for various demographic reasons. The reliance on the evolutionary theory of aging appears to set real limits on the increased life expectancy that might be obtained from interventions on senescent (i.e., postreproductive) humans. If the optimists are wrong and we are more or less at the maximum longevity that is potentially possible, then what do we have to look forward as we age? There is good news and there is bad news (Baltes and Smith 2003). The good news is that, as discussed in previous chapters,

15.3 Anti-Aging Interventions

people in the 65–85 age bracket (which constitutes ~12.8% of the population of developed countries) are much healthier and productive today than they were in the past, that they have high levels of physical and mental well-being, and there is still substantial potential for better mental and physical fitness. For these reasons, people in the 65–75 age bracket, have been termed the “young-old” in recognition of their ability to maintain function into what was traditionally considered the frail and morbid years of old age. There is no reason that some of these “young-old,” could not continue to contribute to and enrich society by developing new social roles and by contributing to the costs of their own medical care. In fact, increasing numbers do so. The bad news is that the “oldest-old,” the people in the 85–100 and over age-bracket (which constitutes ~1.5% of the population in developed countries) are not in as good condition. Many have some sort of functional physical or cognitive loss, and they are losing their social contacts as spouses and relatives and friends die. These oldest-old are for the most part lonely women, most of whom will die alone in a hospital or nursing home. This group may not be numerically large now, but their numbers will substantially increase if we just maintain our current practices. They really are a harbinger of our future. Such data should give even an optimist pause, for the joys of living long are nullified by the loss of dignity and control suffered by the oldest-old. If all that our vaunted knowledge of the biology of aging can do is to increase the odds that one would live to 85 years and beyond but not devise any way to buffer against the inevitable functional losses seen in this age group, then perhaps it would be better for biogerontologists not to continue their researches. Maybe there should be limits to life.

15.3.2 Extended Health Span The optimists offer three arguments for believing that effective anti-aging interventions may not be so difficult to achieve. First, some demographers, illustrated by the views of Oeppen and

511

Vaupel (2002), contend that the entire history of estimating human longevity limits has been a dismal failure because almost every estimate has been falsified by events, and it is likely that the same will happen to current estimates. Life expectancy in the developed nations has been increasing steadily since 1840 at an overall rate of 2.5 years per decade (figure 15.2; but note that the curve is a composite drawn from different societies and represents an ideal rather than a real life expectancy schedule). This increase in life expectancy has been due to a multitude of continuing small improvements in various aspects of our environment. If this trend continues, a life expectancy of 100 years would be reached in about six decades. Oeppen and Vaupel make no predictions as to how this would come about, and so their optimism depends on a continuation of past events. Realists say this is not possible, and they point to the decreasing rate of mortality decrease (figure 2.14) as evidence for their view. What other evidence exists to support the optimist’s views? If a near-term limit to human longevity existed, the increase in life expectancies should be slowing down, and we should expect to see signs of an acceleration in the late-life qx values (Wilmoth 2000). This is not observed. Rather, what is observed is a continuing deceleration of the late-life qx values (Vaupel et al. 1992; Wilmoth 2000; see chapter 2). Given this contradiction between prediction and observation, Oeppen and Vaupel (2002) concluded that human longevity is not approaching a detectable limit, should one even exist. Because the life span is not rigidly fixed and because there appears to be no demographic data supporting the idea of mortality compression, it is plausible that an alternative survival curve that might also describe our future society might still resemble the 2000 curve in figure 2.14, but with the numerical values of the mean and maximum life spans significantly increased as a result of an extension of the curve to the right. Such a survival curve would suggest a different set of future societal health and social usefulness than would a continuation of our present efforts. It has been shown that one genotype can give rise to at least three different longevity phenotypes

512 Chapter 15 Aging-related Research and Its Impact on Society

85

Life expectancy in years

80 75 70 65 60 55 50 45 40 35 30 1600

1650

1700

1750

1800

1850

1900

1950

2000

Year

Figure 15.2 Female life expectancy in northern Europe showing an upward trend beginning in the early 1800s. The data before 1840 comes from nonmetropolitan parishes in England; the subsequent data comes from Denmark, Norway, and Sweden. Note the long-term stability of life expectancy at birth of 35–40 years from 1600 to 1800, which was altered by about 1840 and has increased steadily ever since. (After Oeppen and Vaupel 2002).

(figure 15.3). As shown in figure 15.3A, the first longevity phenotype (type 1) is a delayed onset of senescence, which leads to a significant increase in both mean and maximum life span of the experimental strain. The mortality rate doubling time (MRDT; see chapter 2) is reduced by 50% in these long-lived strains relative to normallived controls. The second longevity phenotype (type 2; figure 15.3B) is an increased early survival, which leads to a significant increase in mean but not in maximum life span. The third longevity phenotype (type 3; figure 15.3C), is an increased later survival, which leads to a change in the maximum (LT90) but not in the mean life spans. Neither of these latter two phenotypes show any alteration in their MRDT relative to the controls. All three longevity phenotypes are known to exist in mice. However, only two of these three longevity phenotypes occur in humans. There is every reason to believe that the missing longev-

ity phenotype class does exist in humans (e.g., see table 6.4), but that it can only be expressed when our cells are presented with the appropriate stimulus. For example, it certainly appears as if the type 1 phenotype is expressed in macaque monkeys (Ingram et al. 2001) and in humans (Walford et al. 2002) subjected to caloric restriction. This is almost certainly a conserved longevity mechanism and a conserved longevity phenotype. If so, it should be possible for us to induce it in humans using stimuli other than a 30% reduction in caloric intake, and these efforts are underway. For example, the data of table 6.4 and figure 7.6 not only support the idea that diet has a rapid and beneficial effect on the genome’s expression patterns, but they also offer a panel of gene expression whose accompanying physiological changes can be used to rapidly ascertain the effectiveness of pharmaceutical therapies designed to mimic the beneficial effects of caloric restriction (see chapter 6). The existence of this type 1 delayed

A

1

Treatment

Survival

Control

Ra La 2La 0

20

40

60

80

100

80

100

80

100

Age (d) B

Survival

1

Control Treatment

Ra PQR 0

20

40

60 Age (d)

C

Survival

1

Control Treatment

Ra RaHx 0

20

40

60 Age (d)

Figure 15.3 The same Ra strain of Drosophila was subjected to three different selection or environmental pressures, which resulted in the expression of three different longevity phenotypes. This means that a single population has within it the capability of changing its longevity in different ways; there are multiple longevity phenotypes, and this realization is important to the biology and ethics of future anti-aging interventions. (A) The two treated longlived strains (La and 2La) have a significantly delayed onset of senescence relative to the normal-lived Ra strain. Both mean and maximum life span values are significantly increased (see Arking et al. 2000, for experimental details). (B) Survival curves of the normal-lived Ra strain and the PQR strain selected from it by direct selection for paraquat resistance. The treated strain has a significant increase only in the mean but not in the maximum life span (see Vettraino et al. 2001, for experimental details). (C) Survival curves of the normal-lived Ra control strain and the longer-lived Ra heat-treated strain. The animals were subjected to a nonlethal heat shock (37oC for 90 minutes) early in life at days 5–7 after eclosion. The treated strain shows a significant increase in the maximum but not the mean life span (see Keuther and Arking 1999, for experimental details). (After Arking et al. 2004.)

514 Chapter 15 Aging-related Research and Its Impact on Society senescence phenotype presents the conceptual basis for a future extension of the human life span. The disagreement has to do with how long it will take to develop an intervention that will effectively induce this phenotype under realworld conditions. The importance of these data to the current debate is that they demonstrate that the failure of humans to express a type I delayed onset of senescence phenotype may be the result of not applying an appropriate stimulus rather than the result of an intrinsic inability of our bodies to respond. This also means that we are asking our bodies to evoke a response that is already built into our genome. This is by definition a natural response. Such a natural evocation of an innate response does not alter any essential aspect of our human nature. The foregoing discussion presents the rationale for believing that it is feasible to extend longevity by intervening in the healthy, presenescent adult. How would we go about accomplishing such a task? What is biologically feasible in the near term is to somehow increase the human health span (figure 9.6). Let us define the health span as that period of time beginning with the end of development (20 years in the case of humans; figure 14.7) and lasting until that time when the log qx values begin to increase. For the purposes of the present discussion, I arbitrarily define that age to be when 10% of the poulation will have died (LT10). Consequently, the senescent span will cover the time from the LT10 until the last animal in the cohort dies. Female mice genetically engineered to have a 50% reduction in the levels of their insulinlike growth factor-1 (IGF-1) receptor (IGF-1R) genes have a type 1 extended longevity due to a delayed onset of senescence (see figure 7.33). The health span of the control female mice covers the period from about 3 to 12 months and that of the experimental female mice covers the period from about 3 to 21 months. Death occurs at 27 and 32 months, respectively. Thus, the senescent span of the female control mice is 15 months and that of the female experimental mice is 11 months. Not only is there a 33% increase in the mean life spans (756 ± 46 versus 568 ± 49 days) and an 18.5% increase in

the maximum life span, but the health span of the treated female animals increased from 9 months to 18 months, while the senescent span decreased from 15 months to 11 months. These effects are summarized in table 15.2. If we assume, as did Carnes et al. (2003), that mice and humans will react rather similarly, then this means that such an intervention might increase the human female health span from 35 years (ages 20–55 years) to about 70 years (ages 20–90 years) while decreasing the human senescent span from 45 years (ages 55–100 years) to 33 years (ages 100–133 yrs). The point of this numerical exercise is to emphasize that the next quantum leap in life expectancy at birth will come from intervening in younger, healthy adults rather than in older, senescent adults. This intervention will be done by pharmaceutical means, not by genetic engineering of humans. The point of using genetically altered mice is not to develop genetic engineering techniques suitable for humans but rather to identify the genetic pathways influencing longevity so that we can devise suitable pharmaceutical interventions that will induce the expression of the same longevity phenotype. As described in chapter 7, pharmaceutical interventions capable of inducing the expression of a type 1 longevity phenotype are a reality. The FDA-approved drug 4-phenylbutyrate induces a delayed onset of senescence in Drosophila by inhibiting certain histone deacetylases and thereby activating various antioxidant and other genes (Lin et al. 1998). Caloric restriction will bring about a type 1 longevity phenotype in mice (figure 6.1) and humans (Walford 2002), and a similar phenotype can be

Table 15.2 Effect of Insulinlike Growth Factor-1 Receptor Reduction on Life Span of Mice Control Life span phase Development Health Senescent Totals

Experimental

Months

%

Months

%

3 9 15 27

11 33 55

3 18 11 32

9.4 56 34

Source: data from Holzenberger et al. (2003).

15.3 Anti-Aging Interventions

induced in rats by high doses of the FDA-approved glucoregulatory drug metformin (Spindler 2003). Related FDA-approved glucoregulatory drugs are believed to exert similar effects on laboratory animals. Other laboratories are searching for different drugs that mimic caloric restriction (Ingram et al. 2004). The effects of phenylbutyrate and metformin constitute a proof of concept even if they are not a practical clinical intervention. This class of pharmaceutical interventions applied to presenescent, healthy humans might bring about the next quantum leap in life expectancy at birth. It is true that this shift has not yet taken place. It is true that the necessary clinically approved biomarkers are not yet in place. It is true that theoretical possibilities are difficult to model. It is true that one could marshall a host of other practical objections (see, e.g., the debate between Richard Sprott and A. de Grey at http:// www.SAGECrossroads.net). Whether the clinical use of such interventions takes place in one decade or in five decades is a matter of conjecture at this moment, but that it will take place is highly probable. The animal studies suggest that it will likely be successful.

15.3.3 Compression of Mortality Before the development of the biomedical and public health interventions that succeeded in reducing premature deaths, most human populations had a survival curve resembling curve B in figure 2.11. As life expectancy is increasing rapidly in most countries, it seemed reasonable to suppose that continuing to decrease the number of premature deaths would eventually result in a human population having the shape of curve A in figure 2.11. This “rectangularization of the curve,” as it has been called, suggested to Fries and Crapo (1981) that we are heading for a society in which the maximum life span will be fixed as it is now at about 120 years, and the median life span will be fixed at about 100 years. This theory implies that the maximum life span is not plastic but is fixed by some unspecified biological process so that no one may live longer than 120 years. Note that such a fixed maximum life span cannot exist because

515

no death or aging program could have evolved under the aegis of natural selection (see chapter 4). There seems to be no robust evidence for compression of mortality (see figure 2.16). In chapter 7, I discussed the animal studies in which significant longevity extension was achieved by a proportional increase of both the mean and maximum life span. This is not the pattern one would expect if the life span was fixed. The compression of mortality hypothesis requires a fixed life span, and this does not appear to exist. As desirable as such a compressed-mortality society might seem, it is not likely to be our future. Absence of evidence is not evidence of absence, and so we cannot be quite sure of that conclusion. But that same absence of evidence also means that, even if a fixed life span exists, we are nowhere near running into those limits. We cannot avoid the issue but will be forced to choose between the consequences of the present scenario and those of the interventionist scenario.

15.3.4 Ethical Considerations Regarding Anti-Aging Interventions Not everyone believes that it is a good idea to live longer than we do at present. Since few of those people opposed to the idea of extended longevity also think it is a good idea to live a shorter life than we do now, this might suggest that the naysayers believe that we are now at the ideal life span. What is more likely is that the naysayers may not view the current state as ideal but argue that the dangers of proceeding far outweigh the benefits. We have already seen how proceeding without foresight has in some cases (e.g., industrialization) rendered severe harm to the environment. The critics are correct in that we should not knowingly repeat the exuberance of ignorance and just pass over the possible consequences. Their points are serious and worthy of consideration. But neither should the criticisms invalidate the applications of this science. Complaints were made and social/individual costs were paid for every social advance. Yet would our society be better off if the biological knowledge that led to Salk’s polio vaccine, for example, was not used for fear of potential side effects (which do exist)?

516 Chapter 15 Aging-related Research and Its Impact on Society There is a cost if we act; there is a cost if we do not act. Neither the risks nor the benefits are objectively known, and so the cost–benefit analysis is not an easy one. It is further complicated by our cultural differences, as witnessed by the fact that serious people in different societies examining the same data may well reach diametrically opposed conclusions, as has happened to the genetically modified food question in Europe versus the United States. We should keep in mind that what benefits an individual might not be viewed as a benefit to society. Yet our society places great value on individual expression and freedom, and so there are good reasons to use our society’s precepts to guide our discussion. Some objections are based on a rational opposition to using genetic engineering or enhancement techniques on humans for both safety and philosophical reasons. I mostly agree with their proponents, but because the feasible biological mechanisms discussed above are based on pharmaceutical effects on ordinary humans, the antigene engineering arguments are superfluous and will not be dealt with here. I only review the philosophical opposition to health span extension in this section. Not everyone believes that this is the best of all possible worlds. Those who do not generally hold that opinion believe that longevity is a positive good, that the past century’s increase in life expectancy has left humans and their societies better off today than they were in the past, and that the effort to extend longevity will certainly have mixed effects (as with any other intervention, be it planes or penicillin) but that the net effect would still be beneficial. I review the answers of these prolongevity partisans to the antilongevity arguments in this section as well. We are heir to at least two different religious traditions which likely inform both sides of this debate. One is the tradition that we live in a hierarchial world in which stewardship implies respecting and conserving the natural order. The other is the tradition that the world is unfinished and that it is part of our God-given responsibility as humans to “repair the world.” These different traditions likely account for much of the disagreement. It is also worth pointing out that all parties

to this debate are knowledgeable in their fields and are concerned that light, not heat, emanate from this discussion. My criticisms are aimed at their ideas, but not at their intent nor their creed. 15.3.4.1 The Philosophical Arguments against Intervening in the Aging Process

Finitude and the Ideas of Leon Kass. Leon Kass is well known as chair of the President’s Council on Bioethics. He holds an M.D. as well as a Ph.D. in biochemistry and is well educated in bioethical issues. Kass wrote, “the finitude of human life is a blessing for every human individual, whether he knows it or not” (Kass 2004, p. 311). In support of this tenet, he offers four benefits of mortality: interest and engagement; seriousness and aspiration, beauty and love; and virtue and moral excellence. Let us consider these attributes. By interest and engagement, Kass means that increasing the life span by, for example, 40% would not necessarily result in a 40% increase in the pleasures of life. “Even less clear are the additions to personal happiness from more of the same of the less pleasurable and less fulfilling activities in which so many of us are engaged so much of the time” (Kass 2004, p. 313). The pleasures of living an extra three or four decades may be eroded by the ordinary ennui of daily life. Taking out the garbage for an extra 2000 times might, perhaps, be compared to a movie that has run too long. To which a skeptic might respond that the argument is flawed by the implicit assumption that the extra years will be used only to do more of the same that was done before. In fact, the extra years will likely be used to add dimensions and new experiences to our life by enabling us to pursue multiple and sequential careers and vocations (table 15.3). Not more of the same, but more of the different may be the rule. New chapters can be added to one’s biography which will transform one’s life to that which could not have been predicted before the gift of years. And so taking out the trash will be the boring but necessary task it is without presenting us with an existential crisis. Yes, it is true that today many people lead banal or meaningless lives by Kass’s standard that the unexamined life is not

15.3 Anti-Aging Interventions

worth living. And it is likely true that just as many people will continue to lead banal and meaningless lives after they take the therapy. But the fact that many people lead a banal life today does not mean that many others do not lead meaningful lives today, for Leon Kass is the best refutation of that generalization. Thus, the probable existence of banal lives in the future cannot lead us to conclude that meaningful lives will not also exist in the future. The existence of banal lives today and tomorrow is an indictment of our practices of child raising and education, but one cannot logically use it as an indictment of longevity extension, for the two variables are not linked. In fact, it would be entirely reasonable to think that the gift of years and the term limits of society might encourage more people to challenge themselves with new experiences and to grow, even if only a bit, in knowledge and in depth. And maybe more of them will, in one chapter or another of their long and rich lives, move back into the classroom to teach the young child with the care and passion developed in them by their transformative biography. We may try to educate or persuade others of the importance of a meaningful life, but in a free society of autonomous adults it is not permissible to force them to live by our standards. It is legal to be banal, and banality is not grounds for denying longevity treatment to some individual. Nor does the banality of one individual constitute grounds for denying longevity treatment to other individuals. Regarding seriousness and aspiration, Kass points out that “Is not the limit on our time the ground of our taking life seriously and living it passionately? To know and to feel that one goes around only once, and that the deadline is not out of sight, is for many people the necessary spur to the pursuit of something worthwhile. ‘Teach us to number our days’, says the Psalmist, ‘that we may get a heart of wisdom”. . . . Mortality makes life matter” (Kass 2004, p. 313). To which a skeptic might respond by quoting Kass’s own words: There may be some activities, especially in some human beings, that do not require finitude as a spur. A powerful desire for understanding can do without external proddings,

517

let alone one related to mortality; and as there is never too much time to learn and to understand, longer, more vigorous life might be simply a boon. The best sorts of friendship, too, seem capable of indefinite growth, especially where growth is tied to learning. . . . But, in any case, I suspect that these are among the rare exceptions. For most activities, and for most of us, I think it is crucial that we recognize and feel the force of not having world enough and time. (Kass 2004, p. 313) The exceptions he concedes are more important than his disclaimer, for surely this is not a complete list of timeless activities. In addition to Kass’s counterargument, a skeptic could point out that the ability to write new chapters for one’s biography and to grow thereby also does not require finitude as a spur but rather the lack of premature finitude. And for those tasks that would benefit from finitude, death need not be the only spur. Prosaic term limits might provide the same incentive with less sorrow. And so what was written as a general statement is seen on reflection to be a partial truth at best, riddled with the exceptions that illustrate its incompleteness. By beauty and love, Kass means that “only a mortal being, aware of his mortality and the transience and vulnerability of all natural things, is moved to make beautiful artifacts. . . . that will bespeak and beautify a world that needs beautification. . . . our appreciation of its beauty depends on our appreciation of mortality. . . . How deeply could one deathless ‘human’ being love another?” (Kass 2004, pp. 313–314). To which a skeptic might respond that the world is indeed in need of beauty, and it is good of people to beautify it by objects or by acts. But beauty is in the eye of the beholder. The transience of spring flowers may be an intrinsic part of their beauty, but then what are we to say of the unchanging beauty of a mountain range, or the constantly changing beauty of an oceanscape, or the simple beauty of a well-designed object, or indeed of the elegant beauty of a mathematical proof? No, beauty and transience do not yield an equation. Beauty is more complex than that. And what

518 Chapter 15 Aging-related Research and Its Impact on Society about love? Kass thinks it depends on death. I think it depends on life. Half the couples in this country may divorce but that means that half do not. And what keeps these together? Of course, custom and habit and fear of change may be misinterpreted by outside observers as love. But I think that many couples stay together because they are continuously growing with and for each other. They continually reinterest the other as people. Perhaps fewer will stay together for life in the future if only because there will be more opportunity for one to grow differently from the other. But love exists between people even if it does not last a lifetime; and it is the hope for, and the memory of, that bonding that sweetens life and makes us vulnerable to the pangs of new love once again. By being vulnerable, we open ourselves up to another person. Sometimes we are rewarded, sometimes not. Love exists despite death, not because of it. By virtue and moral excellence, Kass means that “Through moral courage, endurance, greatness of soul, generosity, devotion to justice—in acts great and small—we rise above our mere creatureliness, spending the precious coinage of the time of our lives for the sake of the noble and the good and the holy . . . to trouble oneself for the sake of home, family, community, and genuine friendship, is truly to live, and is the clear choice of this exemplary mortal” (Kass 2004, pp. 314–315). To which a skeptic might respond that this argument might make sense if immortality was our goal, but it is not. The statement stands as a standard of behavior which we should try to attain, even if only occasionally, by spending the coinage of whatever time we have. Some number of people both today and tomorrow will live banal and mean lives; but tomorrow more will likely attain Kass’s exemplary stature if only because their extended biography should give them the opportunity to grow in character beyond what a shorter chapter of life would allow. And so their life—and not their death—may teach them “to strive, to seek, to find, and not to yield” (Tennyson, Ulysses, line 70). Kass considers procreation to be a fifth benefit of mortality, and it is likely the heart of the matter in his view. Humans seek to live longer

or forever, not because we do not want to die, but in his view rather because we thirst after a wholeness that cannot be fully satisfied in this life. Kass believes that the fullest measure of love or wisdom or redemption cannot be fully satisfied by the biomedical conquest of death; for “once we acknowledge and accept our finitude, we can concern ourselves with living well, and care first and most for the well-being of our souls, and not so much for their mere existence” (Kass 2004, p. 316). But, he continues, “Perhaps there is no soul. Certainly modern science doesn’t speak about the soul; . . . [but] Biology also teaches about transcendence, though it eschews talk about the soul. Biology has long shown us a way to rise above our finitude . . . [via] procreation— the bearing and caring for offspring. . . . reproduction as such implies both the acceptance of the death of self and participation in its transcendence” (Kass 2004, pp. 316–317). Children are needed, he concludes, if for no other reason than that they will look at the world with new eyes and see things not obvious to those whose eyes have been dimmed by habit. Having our own children challenge us in their own ways may both weary and renew us at the same time. Kass is right about children and the renewal of the world. But his writings make sense only if one assumes that it is the present normal-lived society that is hospitable to children, and the prospective long-lived society that is not so. There is simply no evidence to support that assumption; and if the assumption is gratuitous, then the argument for children as a case against extended longevity simply falls apart. It is the present developed society that has lifetime birth rates for women of 1.2 or so, way below replacement. Why should this be so? The demographic transition and the tradeoff of fecundity for survival is undoubtedly part of it. But a new factor in our egalitarian society may be precisely the lack of time. Women in modern society choose to or must work. The pressures of work, be it a full career or just a job, take precedence over the time-requiring needs of children. One tale of our times is the ambitious career woman who has no time for child bearing and raising during her 20s and 30s simply because that is the same period

15.3 Anti-Aging Interventions

of time during which her work perfomance will dictate the success of her career. A large portion of the in vitro fertilization clinic’s business is catering to older career women who now want children, have time for them, and will pay the price to have them. Such women and their spouses testify to the desire for children even after the demographic transition. I submit that the failure to maintain a 2.1 replacement birth rate is a consequence of the lack of time. Longevity extension such as proposed here will give people the time to raise children in their youth, knowing that they have sufficient time ahead to have a full career or two. Kass has it backward: it is the lack of time that encourages men and women to think on themselves, and it is the abundance of time that will enable them to be both nurturing parents and successful adults who have time enough to learn the art of living well and beyond themselves. And if we as parents will take the time and effort to raise children lovingly and well, then I agree with Kass that they will learn and grow and prosper and “hand down and perpetuate this pursuit of what is humanly finest to succeeding generations for all time to come” (Kass 2004, p. 320). Children will still be our message to the future, but perhaps more lovingly crafted then than now. In these and other writings, Kass lets us know that he is not against science but does oppose the coarsening and perhaps unforeseen effects of technology on human life. Who doesn’t? But an opposition which is not rigorously thought through cannot, no matter how graceful its words, be an effective shield against the future (see, e.g., O’Nell 2002). Finally, there is no reason to prefer the “natural” rather than the “technically enhanced” as a matter of principle, for if we did then we would have to abjure the practice of medicine altogether (Kennedy 2004; Testa and Harris 2004). What Are the Limits of Medicine? Other objections have to do with the limits of medicine and what should or should not be done in the name of health. At least three basic objections have been put forth in one form or another by various authors: First, aging is part of the life cycle that defines human beings and tampering with that

519

cycle could literally be dehumanizing. Second, by extending the longevity of individuals, we will play havoc with the web of relationships that define human life and values. Third, longevity extension will not improve and may exacerbate social justice. Let us examine these claims. The first argument is part of Kass’s philosophy, and I dealt with its details above. The comments of Juengst (2004) on this issue offer a more global criticism and deserve to be considered in this context: Of course, arguing that the traditional human life cycle is normative for human beings requires a good bit of philosophical work if it is not to be accused of making a virtue of necessity. Just because human beings have always lived within a particular pattern of life experiences is not necessarily a reason to continue doing so. In fact, the social, technological, and biological dimensions of the typical human life story have been rewritten continuously over our species’ history without diminishing the moral status of those people whose lives have been made possible by that evolution. . . . Until it is clear why, in the light of all its other intrinsic values, it is important for medicine to conserve the human species in its current form, a commitment to life cycle traditionalism in medicine can only count as an idiosyncratic ideology, which autonomous physicians (and their patients) in a free society should have the right to assess, adopt, or reject as they will. (pp. 330–331) A related form of this rebuttal deals with the need of some individuals to not exceed what they consider to be the natural bounds of human life. They have the right to make that decision for thermselves. But why should they also make that decision for me? Is it perhaps the limit to my life that gives meaning to theirs? If so, then when did I cease to be an individual and become a symbolic object? And who exactly made that decision for me? And by what authority? These are important questions when asked in the context of a society dedicated to the equality of its members. There is a difference in the case of a physician who fails

520 Chapter 15 Aging-related Research and Its Impact on Society to help a patient because there is no known cure for that illness, and the case of a physician who fails to help a patient, even though an effective cure exists, but who fails to administer it either on his or her own behalf or on behalf of society (Fossel 2003a,b). The former case is sad but ethical, for the physician did all he or she could and had no further options. The latter case is sad and unethical, for the physician did have an effective alternative and arbitrarily decided not to use it. By devaluing the patient’s interests to those of society, the physician did harm and cooperated with what may be classified as an unethical society.

family relationship overcome the free and informed consent of the patient? And what philosophical necessity allows the idiosyncracies of one person (the physician) to overrule the autonomy of another? The history of the 20th century is replete with examples of how the elevation of certain social relationships over others led to the wholesale violation of individual’s human rights. Based on this empirical evidence, there is no reason to believe that focusing on social relationships would do more to elevate human freedom and happiness than would focusing on the individual autonomy of consenting adults.

Things will never be as they were; there is no avoiding the ethical responsibility by pretending that we live in the past, powerless to help those around us. And if we enact legal restrictions on treatments that can profoundly alleviate human suffering, they will likely be circumvented to save lives, and in that very act will cost us all the more dearly. Laws that encourage lawlessness serve only to further destroy the social fabric. If we restrict a personal medical treatment, even for the best of social reasons, we play a dangerous game, one that cannot be whitewashed as reflecting social virtues of a simpler past. Rather, it not only epitomizes an apparent lack of care, but encourages further social carelessness in others. (Fossel 2003a, p. 1507)

Social Justice. Third are objections based on the claims that social justice will be best served by not using intervention that cannot be distributed equally and fairly, which will serve to continue and to reinforce various prejudices, and which might possibly serve to continue the subordination of certain groups (Chapman 2004). Science is a social activity and does take place in a social context. What types of social concerns merit the public regulation of science? All will agree that honesty is the prime directive of science (Bronowski 1956; chapter 3), and we must be on guard against plagiarism and other forms of dishonesty. Almost all will agree that safety is a prime concern, and we accept regulations specifying just what may and may not be done in a laboratory so as not to harm scientific personnel. We also accept regulations requiring extensive testing to ensure the safety and efficacy of prescription medicines. The society of scientists has called a temporary halt to certain scientific investigations until they could come together and formulate appropriate safety protocols, the best example of this being the Asilomar Conference convened by scientists in 1972 to deal with the then-new topic of genetic engineering. This conference is perhaps the best example of the precautionary principle being applied to molecular biology. There is much less agreement when it comes to the question of the relative social worth of different goals or the question of distributive justice. The United States currently has some 40 million people without basic health insurance.

All the words in the world will not obscure the fact that we have eaten of the tree of knowledge and are now responsible for our actions and inactions. We are our brother’s keepers. The second argument also constitutes part of Kass’s objections and is based on the idea of an ideal Platonic society. I raised in the above discussion some objections based on the likelihood that extended longevity might enhance, not diminish, the value of family relationships. Indeed, the increased longevity of the past century, based on both a decrease in premature mortality and an increase in the senescent span, has led to an increase in cross-generational kinship relations (Riley 1983). A more global objection can be raised: When does the physician’s opinion of a

15.3 Anti-Aging Interventions

It is a reasonable point of view that monies should not be spent on some future medical technology for those who already are assured of good health care while so many are unfairly deprived of even minimal health care. But the political decision that led to their deprivation will not be reversed by not spending money on anti-aging research. It will be reversed only when those 40 million deprived individuals make their political weight felt in Congress. Social justice requires political muscle for its implementation. Even if those 40 million were somehow miraculously provided with heath care, the social argument would simply expand so that the existence of unfairly deprived people anywhere in the world would constitute a reason not to proceed with health span extension in this country. Despite their logical imperfections, these are powerful arguments that illustrate our ethical lapses as a society, and we as individuals are uncomfortable when forced to face the unfairness that we and our representatives have allowed to happen. All that can be said in this context is that the perfect is the enemy of the good. Just because some countries have abysmally short mean life spans does not mean that the Japanese and Scandanavians must also die young. Just because many people in this world do not have access to clean water, for example, does not mean that those with access to potable water cannot improve their system until everyone has a rudimentary system. But these objections do not mean that we should not try to better the situation of all. We really are our brother’s keeper, if only because all of us—rich and poor alike—are in the same boat. Perhaps health span extension in developed societies should be somehow legally tied to advances in basic health care in all societies. These legal ties might include both a “tithe” of sorts on the cost of the therapy and the targeting of those monies for specific ongoing projects benefitting others. And the tithe might also take the form of an obligatory Peace Corps type of personal involvement in healing the world’s ills. Why should not one of the chapters of an extended life revolve about doing something for others? And perhaps such an obligation might be one of the conditions imposed on the deployment of prolongevity

521

interventions by some future Asilomar-type biogerontological conference (see below). We are obliged to repair the world, but we are not required to repair it all at once. Incremental repair is doable and allows us to escape from the status quo. Insisting on perfection will condemn us to the status quo. 15.3.4.2 The Arguments in Favor of Intervening in the Aging Process

While reviewing the objections of those opposed to intervening in the aging process, I also examined the arguments of those people who are in favor of intervention. Their objections to the antiintervention arguments undermine the case against intervention and to that extent are used as part of a prolongevity stance. But are there no positive arguments for intervening in the aging process? It might not be worthwhile going ahead with this process if the best you can say is that much of what the opponents say is not right. And what are we to make of those anti-intervention arguments that are correct? The simplest and most direct argument is that longevity is a positive good, that efforts to increase its quantity or improve its quality are not only beneficial but are generally applauded by Western tradition (Caplan 2004). It is certainly a moral act to alleviate the losses of old age by seeking cures or treatments for an age-related disease. The increased longevity which flows out of this treatment is not morally suspect. The enhancement of longevity by decelerating aging in some way is justified not only because of the inherent positive good of alleviating the losses of old age, but because aging has no biological function (see chapter 4), and so the alleviation of its losses does not make us incomplete in some way. There is no intrinsic virtue in death: “Biology seems to resolutely reject the idea of the utility of natural death, a utility that would allow us to find virtue in death and thus provide some consolation” (Klarsfeld and Revah 2004, p. 193). We ascribe to aging various religious and metaphysical attributes (Post 2004). Altering the timing of senescence by doubling the health span increases healthy life (a positive good) and may

522 Chapter 15 Aging-related Research and Its Impact on Society ameliorate but will not obliterate senescence, thus retaining the existential attributes important to many. The fact that this longevity extension will likely be brought about by pharmaceutical means indicates that our genetic nature will not be altered or violated. The gift of time will give us the opportunity to live a more varied and intensive life, more open and responsive to the world and to our obligations. Writing from a feminist and philosophical perspective, Overall (2003) examined the arguments for and against intentional longevity extension and reviewed the social implications of such a policy and concluded that, “Other things being equal, a long life is a better life, and a social policy that promotes the extension of human life is amply justified” (p. 218). She finds the view of those (e.g., Callahan, 1999) who claim it a modern failure that we struggle against death and for life, to be reminescent of “medieval European attitudes toward death [which] do not provide a model for our own era” (Overall 2003, p.32). The practical social outcome of the view that human life should not be extended is that, at some point, it can lead to the idea that it is better for a whole class of people (e.g., aged poor females) to die, and this she finds to be morally untenable. Moreover, since life is the basis of all other philosophical goods, then “given a minimal level of health and well-being, a much longer life would almost certainly be desirable for large numbers of people” (Overall 2003, p. 182).We may use that time wisely, as some do today. We may use that time foolishly, as some do today. We should use more of that time to benefit others than we do today. At the end of that time, we will still have to assess and justify that life if only to ourselves. Health span extension simply gives us a greater opportunity to live a fulfilling life without making us immortal, nonhuman, or unnatural. In the environmental sciences, much use is made of the precautionary principle, which deals with how we should judge research that has uncertain outcomes. This principle likely was the underlying motivation for molecular geneticists in 1972, faced with the then-new techniques of directed gene transfer from one organism to another, to call for a voluntary moratorium on their

activities while they debated them at the Asilomar Conference and devised a consensus set of guidelines. Perhaps the biogerontologists and other interested parties should do the same when the technology improves to the point that it has some initial level of clinical efficacy. In the meantime, we should continue these discussions so that, when the time comes for such a conference, we will have thought through our positions as well as those who disagree with us and will know exactly what it is that we are talking about. We should neither approve nor disapprove of a prolongevity policy on the grounds of financial implications only, but it is entirely valid to ask just how such a policy would affect our financial future (Post 2004). Our present course is almost certain to lead us into a financial disaster (Peterson 2004); if the prolongevity policy simply adds to those deficits, then it might be prudent to conclude that we cannot afford to implement such a policy, desirable though it might be. However, if people are offered the opportunity to adopt a prolongevity therapy and the healthy lifestyle that necessarily goes with it, I suspect that most individuals would not turn it down simply because the government might require that people on life extension therapy could not retire at age 65 but would have to wait until age 90 or so. As pointed out above, such 90 year olds should be functionally equivalent to 65 year olds today. The 25-year delay in payouts under Social Security and Medicare should be sufficient to convert the doleful future that Peterson (2004) believes awaits us to one that would better allocate our resources between the different generations. If we decide to adopt the prolongevity strategy for other reasons, then the financial benefits it might bestow are simply a happy consequence of adopting a wise policy.

15.3.5 Social Effects of Anti-Aging Interventions Our future likely will catch us unawares. In March 1995, Science magazine asked 60 prominent scientists what they saw in the future for science. All sorts of interesting scenarios ranging

15.3 Anti-Aging Interventions

from physics to economics were set forth—some plausible, some fanciful, all imaginative. But not one of these scientists speculated about the effect of our increasing knowledge of aging processes on our future. In 2001, a similar survey of 25 other prominent scientists again failed to mention the manipulation of aging and senescence (Brockman 2002). Only in 2005 was the topic of human life span extension raised in Science magazine as one of the important unsolved scientific questions (Couzin, 2005). It appears as if scientists are as skilled in predicting the future as are ordinary people, in that we cannot recognize it until it has taken place. But some pioneers have envisioned this future when others did not (e.g., Boulding 1965), and so it has not been totally unforeseen. The outcome is still worth discussing, even if this author is as fallible as the others, and even if these predictions should be read with a skeptical eye. And so the question is, given the two scenarios developed above, what might be the effects of each on our society? There are different predictions, ranging from the serenely optimistic to the darkly pessimistic, as to the effects of prolongevity intervention on the age structure, mores, and daily rhythm of our society. Because of space limitations, my predictions and related supplementary material are not printed here but are posted on the Web at http://bio.wayne.edu/profhtml/ arking/textbook/supplement.html and the interested reader is referred there for further information. More important than the predictions, almost all of which are going to be wrong in some way, is how we as a society choose between different scenarios. 15.3.5.1 How to Choose between Different Scenarios?

How do we make a rational decision between different scenarios when there is no evidence one way or the other? The Pentagon was not conceptually wrong when they considered that perhaps the most accurate estimates of the future might be obtained by the summed opinions of wellinformed individuals. If the opinions of experimental biogerontologists on pertinent aspects of longevity extension could be periodically re-

523

cruited and analyzed, then these informed expert opinions could be used to set the values of important biological inputs into existing demographic and econometric computer models. A “Sim City” writ large, as it were. This would not address the ethical issues of course, but it would address the social and economic ones. Will the population stratify into long-lived and short-lived cohorts? Will increased wealth follow on increased longevity, as the phenomenon of compound interest suggests? Would that make our society more egalitarian? If delayed retirement is a requirement of the intervention, then what does this do to retirement plans and work place structure? What sorts of social changes might arise in response to these changes? And so forth. As our laboratory expertise in longevity extension grows and the original estimates become more reliable, then so might the accuracy of our perceptions of that perceived future and the problems that will need to be dealt with. It would certainly help if those with expertise in economic and psychological and societal aspects of gerontology would lend their talents to this quest. Such best estimates will not be the outcome of personal idiosyncracies and so might carry some persuasive weight with decision makers, particularly if the simulations show us a way out of the age-based financial difficulties and the demographic shortage of working-age people facing all developed societies. We should perhaps remind ourselves that such a problematic future is in fact the default option if we and society do nothing. Our choice is not between a nice present and a scary future, but between two different scary futures. Such a survey of biogerontologists was carried out in which they were asked: What will be the life expectancy at birth (e0) of a child born in the year 2100? At the present time, the e0 in the developed nations is 75.9 years. The 60 biogerontologists surveyed offered predictions to the foregoing question based on their professional assessment of the data. The responses ranged from 75 years to 5000 years, with a median value of 100 years and a mean value of 258 years (Richel 2003). All but one of the respondents predicted an increase in the e0 value by 2100. Most of those surveyed predict that the e0 of

524 Chapter 15 Aging-related Research and Its Impact on Society developed nations in 2100 will be somewhere between 90 and 130 years, and so the median value is not some unrepresentative number. It is, however, a 33% increase in the current value. What would happen to our various socioeconomic projections if we entered a median e0 value of 100 into the computer models, that value now being the current best guess opinion of the leading researchers in the field? Would the transformed societies likely predicted by such econometric models be as ignored by the political leadership as are the scientific models of global warming, or would they be taken as seriously as are the projections of Social Security deficits? If the e0 prediction and the expected demographic, social, and economic outcomes were to be updated each year, and if the predictions were to take the reports of anti-aging interventions into consideration, then societal leaders might start paying attention after a while. At the very least, the data obtained from such expert opinion surveys might suggest when the time is ripe for an Asilomar-type biogerontology conference. I think history shows us that the major problems and benefits associated with a new invention are not the ones we can easily see. All our forecasting will not make us fully aware of the texture of that future society. Yet forecast we must, if only to demonstrate that many of the feared outcomes are unlikely imaginings. Since many of the ethical urgings for and against any anti-aging therapy are based in part on their perceived social effects (e.g., Callahan 1987; Overall 2003), it might be useful to test the social scientists’ expectations against those of the philosophers. And we will have to insist that the ethics of not acting be accorded equal status with the ethics of acting. Refusal to alleviate another’s pain and suffering in the name of some higher ethic seems to violate that compassion “which is the highest of human motivations, allowing us first to understand, then to prevent, the suffering, fear, and tragedy of others” (Fossel 2003b, p. 71). Once we have the conceptual tools in hand with which to alleviate the onset of agerelated diseases, then we need extraordinary reasons to not act, to deny our compassion, and still be ethical people in an ethical society.

We may also set an example against the increased cautiousness of a society that can mostly see harm in the unknown. Half a millenium ago, explorers set out to discover the unknown shores that lay beyond their knowledge. The Chinese Empire proved timid and pulled back. Europe did not, and that has made all the difference.

15.4 Conclusion These scenarios may sound fanciful and unreal, and the philosophical arguments on either side may be considered as irrelevant to good science or ordinary life. But we should remember that the current decline in old age mortality and the increase of life expectancy by 2 years per decade may, by itself, ensure that 10 years will be added to the life expectancy of the elderly by the year 2040, and a full 20 years by the end of the 21st century. We are talking about the probable life expectancies of the children being born today. These estimates may or may not come true, but they cannot be regarded as figments of an overheated imagination. The extended senescent scenario will happen by default regardless of our wishes. So why not anticipate it and adopt policies that will extend the health span of all rather than the senescent span of a relative few? If in the process of working out some sort of fair and just prolongevity policy for the country we find that one outcome is that the financial problems associated with senescence and aging are alleviated, then that would be a secondary but useful result of adopting wise policies in the first place. The extended health span scenario is currently a laboratory effect but one which may well move out into the larger society within the near future, and probably sooner rather than later. If this sort of possible future strikes a responsive chord in enough people, it might constitute an acceptable societal goal. There are strong opinions on all sides of the question. I have reviewed these positive and negative arguments, both empirical and philosophical. Perhaps it will not be facts but value judgments that will play a large role in our decision to proceed or not. The po-

15.4 Conclusion

litical discussion is unlikely to be rational. And yet our own history strongly suggests that even if the development and implementation of the interventions might be delayed by political action, they will not be permanently (or even long) stopped. The technology will be used somewhere, somehow. Sooner or later, we will have to deal with the technology and its after effects. Perhaps the worst approach would be if the philosophical arguments against health span extension were successful enough to prohibit any public funding or oversight over the development and implementation of the technology but did not inhibit the private sector from doing so. This is, of course, the current situation in the U.S. regarding stem cells. All this policy accomplished was to allow our politicians to feel good while ceding to people and corporations in other countries the responsibility for the development of stem cell technology and policy. The private marketplace, although powerful, is a limited tool. Its vision does not effectively extend to all societal needs. Forbidding public oversight and funding would be a Pyrrhic victory for the antiinterventionists because it would likely allow the private sector to make true their worst fears. They might well win the argument and lose the contest. Intelligently focused public funds and policies can increase the responsiveness and return of private companies that follow their prescriptions, and thus deal with some of the other real problems and opportunities facing our society today. For example, the implementation of such health span extensions should be tightly linked to the implementation of intergenerational resource transfers that fully engage and enhance the early development and education of children. This is consistent with our evolutionary history (figure 14.9), it is consistent with what we know about the development of the human brain and emotions, and it is the best way with which to ensure both a broad advance against existing social injustices as well as ensuring that those individuals who choose to take the interventions will have minds fully developed to

525

understand their prospective benefits and obligations. Private market devotees might not willingly agree to such compromises outside the framework of a political settlement. Refusal to consider public funding may be a principled one, but it is also a short-sighted one that cedes the argument in advance to the most narrowly focused sector of society. To define the successes of the past halfcentury of biomedical and social advances as a problem, as is now the current craze, is extraordinarily shortsighted. It causes us to overlook the possibility that one consequence of our present biogerontological studies might be the gift of an additional 35+ years of healthy life. If it comes about, that gift will transform society largely, I think, for the better. It will offer us the potential to live our lives fully. We should plan for it. If we are to take responsibility for our own lives, then we must also have the knowledge with which to judge our progress and make the adjustments we deem necessary. A better understanding of the biology of aging has been the goal of my quest through this book. Our understanding is as imperfect as are the data. Yet we have seen that the aging process can be defined, described, measured, and modified. We can detect fundamental similarities in the biological processes involved in aging. We can begin to view aging as the playing out of a fundamental biological process susceptible to intervention. We can appreciate that genes do not work in a vacuum but rather in the context of specific environments within and external to the individual. This knowledge should allow us now to view aging in a somewhat different light than we did when we first opened this book. Perhaps we can now better appreciate that the study of longevity determination and senescence leads us inevitably to a full integration of all the biological disciplines with the environmental factors that influence them. And perhaps we have now reached the final destination in our journey through the fact and theory of the biology of aging—to return to where we started and see it with a new vision.

This page intentionally left blank

References

Abbot, M. H., E. A. Murphy, D. R. Bolling and H. Abbey. 1974. The familial component in longevity. A study of offspring of nonagenarians. II. Preliminary analysis of the completed study. Johns Hopkins Med. J. 134: 1–16. Abdenur, J. E., W. T. Brown, S. Friedman, M. Smith and F. Lifshitz. 1997. Response to nutritional and growth hormone treatment in progeria. Metab. Clin. Exp. 46: 851–856. Abraham, C. R., D. J. Selkoe and H. Potter. 1988. Immunochemical identification of the serine protein inhibitor alpha 1-antichymotrypsin in the brain amyloid deposits of Alzheimer’s disease. Cell 52: 487–501. Ackermann, M., S. C. Stearns and U. Jenal. 2003. Senescence in a bacterium with asymmetric division. Science 300: 1920. Adams, A. 2004. Mitochondria at the crossroads of life and death. The Scientist 18 (19): 25–29. Adams, M. B. 2004. The quest for immortality: Visions and presentiments in science and literature. In The Fountain of Youth: Cultural, Scientific, and Ethical Perspectives on a Biomedical Goal, S. G. Post and R. H. Binstock, eds. Oxford University Press, New York. Agarwal, S. and R. S. Sohal. 1995. Differential oxidative damage to mitochondrial proteins during aging. Mech. Ageing Devel. 85: 55–63. Agarwal, S. and R. S. Sohal. 1996. Relationship between susceptibility to protein oxidation, aging and maximum life span potential of different species. Exp. Gerontol. 31: 365–372. Aguilaniu, H., L. Gustafsson, M. Rigoulet and T. Nystrom. 2003. Asymmetric inheritance of oxidatively damage proteins during cytokinesis. Science 299: 1751–1753. Aikagi et al. 2003. Albert, R., H. Heong and A. L. Barabasi. 2000. Error and attack tolerance of complex networks. Nature 406: 378–382. Alberts, B., D. Bray, J. Lewis, M. Raff, K. Roberts and J. D. Watson. 1983. Molecular Biology of the Cell. Garland, New York. Allsopp, R. C. and C. B. Harley. 1995. Evidence for a critical telomere length in senescent human fibroblasts. Exp. Cell Res. 219: 130–136. Allsopp, R. C., H. Vaziri, C. Patterson, S. Goldstein,

E. V. Younglai, A. B. Futcher, C. W. Greider and C. B. Harley. 1992. Telomere length predicts replicative capacity of human fibroblasts. Proc. Natl. Acad. Sci. USA 89: 10114–10118. Alpha-Tocopherol, Beta Carotene Cancer Prevention Study Group. 1994. The effect of vitamin E and beta carotene on the incidence of lung cancer and other cancers in male smokers. N. Engl. J. Med. 330: 1029–1035. Amador, M. A., M. L. Abler, E. J. DeRocher, D. M. Thompson, A. van Hoof, N. D. LeBrasseur, A. Lers and P. J. Green. 2000. Identification of BFN1, a bifunctional nuclease induced during leaf and stem senescence in Arabidopsis. Plant Physiol. 122: 169–180. Amdam, G. V., Z. L. P. Simoes, A. Hagen, K. Norberg, K. Schroder, O. Mikkelsen, T. B. L. Kirkwood and S. W. Omholt. 2004. Hormonal control of the yolk precursor vitellogenin regulates immune function and longevity in honeybees. Exp. Gerontol. 39: 767–773. Ames, B. 1988. Endogenous genetic damage as related to cancer and aging. Paper presented at 41st Annual Meeting of the Gerontological Society of America, November 19, 1988. Ames, B. N. 1994. The assay of endogenous oxidative DNA damage as related to aging. Pp. 397–405 in A. K. Balin (ed.), Practical Handbook of Human Biologic Age Determination. CRC Press, Boca Raton, FL. Ames, B. N. and J. Liu. 2004. Delaying the mitochondrial decay of aging with acetylcarnitine. Ann. N.Y. Acad. Sci. 1033: 108–116. Ames, B. N., M. K. Shigenaga and T. M. Hagen. 1993. Oxidants, antioxidants and the degenerative diseases of aging. Proc. Natl. Acad. Sci. USA 90: 7915–7922. Aminoff, D. 1988. The role of sialoglycoconjugates in the aging and sequestration of red cells from circulation. Blood Cells 14: 229–247. Anderson, R. M., K. J. Bitterman, J. G. Wood, O. Medvedik and D. A. Sinclair. 2003. Nicotinamide and PNC1 govern lifespan extension by calorie restriction in Saccharomyces cerevisiae. Nature 423: 181–185. Anderton, B. H. 1987. Progress in molecular pathology. Nature 325: 658–659.

527

528

References

Andres, R. 1984. Mortality and obesity: The rationale for age-specific height-weight tables. Pp. 759–766 in R. A. Andres, E. L. Bierman and W. R. Hazzard (eds.), Principles of Geriatric Medicine. McGrawHill, New York. Apfeld, J. and C. Kenyon. 1998. Cell nonautonomy of C. elegans daf-2 function in the regulation of diapause and life-span. Cell 95: 199–210. Appel, L. J., T. J. Moore, E. Obarzanek, W. M. Vollmer, L. P. Svetkey, et al. 1997. A clinical trial of the effects of dietary patterns on blood pressure. DASH Collaborative Research Group. N. Engl. J. Med. 336: 1117–1124. Araki, T., H. Kato, Y. Kanai and K. Kogure. 1993. Selective changes of neurotransmitter receptors in middle-aged gerbil brain. Neurochem. Intl. 23: 541–548. Araki, T., Y. Sasaki and J. Milbrandt. 2004. Increased nuclear NAD biosynthesis and SIRT1 activation prevent axonal degeneration. Science 305: 1010–1013. Arantes-Oliveira, N., J. Apfeld, A. Dillin and C. Kenyon. 2002. Regulation of lifespan by germline stem cells in Caenorhabditis elegans. Science 295: 502–505. Arbeitman, M. N., E. E. M. Furlong, F. Imam, E. Johnson, B. H. Hull, B. S. Baker, M. A. Krasnow, M. P. Scott, R. W. Davis and K. P. White. 2002. Gene expression during the life cyle of Drosophila melanogaster. Science 297: 2270–2275. Archer, J. R. and D. E. Harrison. 1996. L-deprenyl treatment in aged mice slightly increases life spans, and greatly reduces fecundity by aged males. J. Gerontol. Biol. Sci. 51A: B448–B453. Arias, E. B., L. E. Gosselin and G. D. Cartee. 2001. Exercise training eliminates age-related differences in skeletal muscle insulin receptor an dIRS-1 abundance in rats. J. Gerontol. Biol. Sci. 56A: B449– B455. Arking, R. 1987a. Genetic and environmental determinants of longevity in Drosophila. Pp. 1–22 in A. D. Woodhead and K. H. Thompson (eds.), Evolution of Longevity in Animals: A Comparative Approach. Plenum Press, New York. Arking, R. 1987b. Successful selection for increased longevity in Drosophila: Analysis of the survival data and presentation of a hypothesis on the genetic regulation of longevity. Exp. Gerontol. 22: 199–220. Arking, R. 1988. Genetic analyses of aging processes in Drosophila. Exp. Aging Res. 14: 125–135. Arking R 2004. Extending human longevity: A biological probability. Pp. x–x in S. G. Post and R. H. Binstock (eds.), The Fountain of Youth: Cultural, Scientific, and Ethical Perspectives on a Biomedical Goal. Oxford University Press, New York.

Arking, R., S. Buck, A. Berrios, S. Dwyer and G. T. Baker III. 1991. Elevated paraquat resistance can be used as a bioassay for longevity in a genetically based long-lived strain of Drosophila. Dev. Genet. 12: 362–370. Arking, R., S. Buck, D-S. Hwangbo and M. Lane. 2002a. Metabolic alterations and shifts in energy allocations are corequisites for the expression of extended longevity genes in Drosophila. Ann. N.Y. Acad. Sci. 959: 251–262. Arking, R., S. Buck, V. N. Novoseltsev, S-D. Hwangbo and M. Lane. 2002b. Genomic plasticity, energy allocations, and the extended longevity phenotypes of Drosophila. Aging Res. Rev. 1: 209–228. Arking, R., J. Novoseltsev, D. S. Hwangbo, V. Novoseltsev and M. Lane. 2002c. Different age-specific demographic profiles are generated in the same normal-lived Drosophila strain by different longevity stimuli. J. Gerontol.: Biol. Sci. 57A: B390–B399. Arking, R., S. Buck, R. A. Wells and R. Pretzlaff. 1988. Metabolic rates in genetically based long lived strains of Drosophila. Exp. Gerontol. 23: 59–76. Arking, R., V. Burde, K. Graves, S. Buck, R. Hari, S. Soliman, A. Saraiya, K. Sathrasala, N. Wehr and R. Levine. 1997. Extended longevity is due to altered antioxidant gene expression. Gerontologist 37: 323 (abstract). Arking, R., V. Burde, K. Graves, R. Hari, E. Feldman, A. Zeevi, S. Soliman, A. Saraiya, S. Buck, J. Vettraino, K. Sathrasala, N. Wehr and R.L. Levine. 2000a. Forward and reverse selection for longevity in Drosophila is characterized by alteration of antioxidant gene expression and oxidative damage patterns. Exp. Gerontol. 35: 167–85. Arking, R. and S. P. Dudas. 1989. A review of genetic investigations into aging processes of Drosophila. J. Am. Geriatr. Soc. 37: 757–773. Arking, R., A. G. Force, S. P. Dudas, S. Buck and G. T. Baker III. 1996. Factors contributing to the plasticity of the extended longevity phenotype of Drosophila. Exp. Gerontol. 31: 623–643. Arking, R. and C. Giroux. 2001. Antioxidant genes, hormesis, and demographic longevity. J. AntiAging Med. 4: 125–136. Arking, R. and R. A. Wells. 1990. Genetic alteration of normal aging processes is responsible for extended longevity in Drosophila. Dev. Genet. 11: 141–148. Armeni, T., G. Principato, J. L. Quiles, C. Pieri, S. Bompadre and M. Battino. 2003. Mitochondria dysfunctions during aging: Vitamin E deficiency or caloric restriction—two different ways of modulating stress. J. Bioenerg. Biomembr. 35: 181–191. Armstrong, D. 1984. Free radical involvement in the formation of lipopigments. Pp. 129–141 in D. Armstrong, R. S. Sohal, R. G. Cutler and T. F. Slater (eds.), Free Radicals in Molecular Biology, Aging and Disease. Raven Press, New York.

References

Arnheim, N. and G. Cortopassi. 1992. Deleterious mitochondrial DNA mutations accumulate in aging human tissues. Mutat. Res. 275: 157–167. Asaumi, S., H. Kuroyanagi, N. Seki and T. Shirasawa. 1999. Orthologues of the Caenorhabditis elegans longevity gene clk-1 in mouse and human. Genomics 58: 293–301. Ascencio, C., J. C. Rodriguez-Aguilera, M. Ruiz-Ferrer, J. Vela and P. Navas. 2003. Silencing of ubiquinone biosynthesis genes extends life span in Caenorhabiditis elegans. FASEB J. 17: 1135–1137. Ashburner, M. and J. J. Bonner. 1979. The induction of gene activity in Drosophilia by heat shock. Cell 17: 241–254. Ashcroft, G. S., S. J. Mills and J. J. Ashworth. 2002. Ageing and wound healing. Biogerontology. 3: 337–345. Asif, M., J. Egan, S. Vasan, G. N. Jyothirmayi, M. R. Masurekar, S. Lopez, C. Williams, R. L. Torres, D. Wagle, P. Ulrich, A. Cerami, M. Brines and T. J. Regan. 2000. An advanced glycation endproduct cross-link breaker can reverse age-related increases in myocardial stiffness. Proc. Natl. Acad. Sci. USA 97: 2809–2813. Atchley, W. R. and W. M. Fitch. 1991. Gene trees and the origins of inbred strains of mice. Science. 254: 554–558. Atkinson, J., I. Lartaud and C. Capdeville-Atkinson. 1992. Debit sanguin cerebral: Evolution de la regulation avec l’age. Presse Med. 21: 1227–1230. Atkinson, R. L. 1997. Use of drugs in the treatment of obesity. Annu. Rev. Nutr. 17: 383–403. Aufderheide, K. J. 1987. Clonal aging in Paramecium aurelia. II. Evidence of functional changes in the macronucleus with age. Mech. Ageing Dev. 37: 265–279. Austad, S. N. 1989. Life extension by dietary restriction in the bowl and doily spider, Frontinella pyramitela. Exp. Gerontol. 24: 83–92. Austad, S. N. 1993. Retarded senescence in insular populations of Virginia opossums (Didelphis virginiana). J. Zool. 229: 695–708. Austad, S. N. 1997. Comparative aging and life histories in mammals. Exp. Gerontol. 32: 23–38. Austad, S. N. 2001. Does caloric restriction in the laboratory simply prevent overfeeding and return house mice to their natural level of food intake? SAGE document. http://sageke.sciencemag.org/ cgi/content/full/sageke;2001/6/pe3. Austad, S. N. and K. E. Fischer. 1991. Mammalian aging, metabolism and ecology: Evidence from the bats and marsupials. J. Gerontol. Biol. Sci. 46: B47–B53. Aviram, M. 2000. Review of human studies on oxidative damage and antioxidant protection related to cardiovascular diseases. Free Radic. Res. 33 Suppl: S85–97.

529

Avise, J. C. 1993. The evolutionary biology of aging, sexual reproduction and DNA repair. Evolution 47: 1293–1301. Aviv, A., J. Shay, K. Christensen and W. Wright. 2005. The longevity gender gap: are telomeres the explanation? http://sageke.sciencemag.org/cgi/content/full/2005/23/pe16 Axton, M (ed.). 2002. The Chipping Forecast II. Nat. Gen. 32 (4 suppl. 2): 461–552. Azzi, A., R. Gysin, P. Kempna, R. Ricciarelli, L. Villacorta, T. Visarius and J. M. Zingg. 2003. The role of alpha-tocopherol in preventing disease: from epidemiology to molecular events. Mol. Aspects Med. 24: 325–336. Bailey A. J. 2001. Molecular mechanisms of ageing in connective tissues. Mech. Ageing Dev. 122: 735–755. Baker, D. J., K. B. Jeganathan, J. D. Cameron, M. Thompson, S. Juneja, et al. 2004. BubR1 insufficiency causes early onset of aging-associated phenotypes and infertility in mice. Nature Genet. 36: 744–749. Baker, G. T., M. Jacobson and G. Mokrynski. 1985. Aging in Drosophila. Pp. 511–578 in V. Cristofalo (ed.), Cell Biology Handbook in Aging. CRC Press, Boca Raton, FL. Baker, G. T. III and R. Sprott. 1988. Biomarkers of aging. Exp. Gerontol. 23: 223–239. Baker, G. T. III, R. E. Zschunve and E. M. Podgorski Jr. 1979. Alteration in thermal stability of ribosomes from Drosophila melanogaster with age. Experientia 35: 1053–1054. Balazs, E. A. 1977. Intercellular matrix of connective tissue. Pp. 227–240 in C. E. Finch and L. Hayflick (eds.), Handbook of the Biology of Aging. Van Nostrand Reinhold, New York. Balin, A. K. 1983. Testing the free radical theory of aging. Pp. 137–182 in R. C. Adelman and G. S. Roth (eds.), Testing the Theories of Aging. CRC Press, Boca Raton, FL. Balin, A. K. (ed.) 1994a. Practical Handbook of Human Biologic Age Determination. CRC Press, Boca Raton, FL. Balin, A. K. 1994b. Skin changes as a reflection of biologic age. Pp. 343–373 in A. K. Balin (ed.), Practical Handbook of Human Biologic Age Determination. CRC Press, Boca Raton, FL. Balin, A. K., A. J. Fisher, M. Anzelone, I. Leong and R. G. Allen. 2002. Effects of establishing cell cultures and cell culture conditions on the proliferative life span of human fibroblasts isolated from different tissues and donors of different ages. Exp. Cell Res. 274: 275–287. Baltes, P. B. and J. Smith. 2003. New frontiers in the future of aging: From successful aging of the young old to the dilemmas of the fourth age. Gerontology 49: 123–135.

530

References

Bank, L. and L. F. Jarvik. 1979. A longitudinal study of aging human twins. Pp. 303–333 in E. L. Schneider (ed.), The Genetics of Aging. Plenum Press, New York. Bannerjee, D. and F. Slack. 2002. Control of developmental timing by small temporal RNAs: A paradigm for RNA-mediated regulation of gene expression. BioEssays 24: 119–129. Barbieri, M., M. R. Rizzo, D. Manzella, R. Grella, E. Ragno, M. Carbonella, A. M. Abbatecola and G. Paolisso. 2003. Glucose regulation and oxidative stress in healthy centernarians. Exp. Gerontol. 38: 137–143. Barger, J. L., R. L. Walford and R. Weindruch. 2003. The retardation of aging by caloric restriction: Its significance in the transgenic era. Exp. Gerontol. 38: 1343–1351. Barinaga, M. 1995. Obese protein slims mice. Science 269: 475–476. Barja, G. 1999. Mitochondrial oxygen radical generation and leak: Sites of production in states 4 and 3, organ specificity, and relation to aging and longevity. J. Bioenerg. Biomembr. 31: 347–366. Barja, G. 2000. The flux of free radical attack through mitochondrial DNA is related to aging rate. Aging Clin. Exp. Res. 12: 342–355. Barja, G. 2002. Endogenous oxidative stress: relationship to aging, longevity, and caloric restriction. Ageing Res. Rev. 1: 397–411. Barja, G., S. Cadenas, C. Rojas, R. Perez-Campo and M. Lopez-Torres. 1994. Low mitochondrial free radical production per unit O2 consumption can explain the simultaneous presence of high longevity and high aerobic metabolic rate in birds. Free Rad. Res. 21: 317–328. Barja, G. and A. Herrero. 2000. Oxidative damage to mitochondrial DNA is inversely related to maximum life span in the heart and brain of mammals. FASEB J. 14: 312–318. Barker, D. J. P. 1995. Intrauterine programming of adult disease. Mol. Med. Today 1: 418–423. Baron, J. C. and G. Marchal. 1992. Viellissement cerebral et cardiovasculaire et metabolisme energetique cerebral: Etudes chez l’homme par la tomographie a positons. Presse Med. 21: 1231–1237. Barrett, J. et al. 1986. Biology. Prentice-Hall, Englewood Cliffs, NJ. Barrows, C. H. and G. C. Kokkonen. 1982. Dietary restriction and life extension—biological mechanisms. Pp. 219–243 in G. Moment (ed.), Nutritional Approaches to Aging Research. CRC Press, Boca Raton, FL. Bartke, A. 2000. Delayed aging in Ames dwarf mice. Relationships to endocrine function and body size. Results Problems Cell Differen. 29: 181–198. Bartke, A. 2001. A word of caution: Can growth hormone accelerate aging. J. Anti-Aging Med. 4: 301–309.

Bartke, A., J. C. Wright, J. A. Mattison, D. K. Ingram, R. A. Miller and G. S. Roth. 2001. Longevity: extending the lifespan of long-lived mice. Nature 414: 412. Bartlett, J. 1919. Familiar Quotations, 10th ed, rev. and enl. by Nathan Haskell Dole. Boston: Little, Brown, 1919; reprinted online by Bartleby.com, 2000. Bartus, R.T., R. L. Dean III, B. Beer and A. S. Lippa. 1982. The cholinergic hypothesis of geriatric memory dysfunction. Science 217: 408–417. Barz, W. P. and P. Walter. 1999. Two endoplasmic reticulum (ER) membrane proteins that facilitate ERto-Golgi transport of glycosylphosphatidylinositolanchored proteins. Mol. Biol. Cell 10: 1043–1059. Baserga, R. 1985. The Biology of Cell Reproduction. Harvard University Press, Cambridge, MA. Bates, S. R. and E. C. Gangloff (eds.). 1986. Atherogenesis and Aging. Springer-Verlag, New York. Bayreuther, K. and J. Gogol. 1993. Terminal differentiation, aging, apoptosis, or transformation of the WI-38 fibroblasts in the fibroblast stem cell system in vitro. Pp. 264–271 in A. Bernd, J. BereiterHahn, F. H Hevert and H. Holzmann (eds.), Cell and Tissue Culture Models in Dermatological Research. Springer-Verlag, Berlin. Bayreuther, K., P. I. Francz and H. P. Rodemann. 1992b. Fibroblasts in normal and pathological terminal differentiation, aging, apoptosis and transformation. Arch. Gerontol. Geriatr., Suppl. 3: 47– 74. Bayreuther, K., P. I. Francz, J. Gogol and K. Kontermann. 1992a. Terminal differentiation, aging, apoptosis and spontaneous transformation in fibroblast stem cell systems in vivo and in vitro. Ann. N.Y. Acad. Sci. 663: 167–179. Bayreuther, K., H. P. Rodemann, R. Hommel, K. Dittman, M. Albiez and P. I. Francz. 1988. Human skin fibroblasts in vitro differentiate along a terminal cell lineage. Proc. Natl. Acad. Sci. USA 85: 5112–5116. Beausejour, C. M., A. Krtolica, F. Galimi, M. Narita, S. W. Lowe, P. Yaswen and J. Campisi. 2003. Reversal of human cellular senescence: roles of the p53 and p16 pathways. EMBO J. 22: 4212–4222. Beaverton, R. J. H. 1987. Longevity in fish: Some ecological and evolutionary considerations. Pp. 161– 186 in A. D. Woodhead and K. H. Thompson (eds.), Evolution of Longevity in Animals: A Comparative Approach. Plenum Press, New York. Becker, J., V. Metzger, A.-M. Courgeon and M. BestBelpomme. 1990. Hydrogen peroxide activates immediate binding of a Drosophila factor to DNA heat-shock regulatory element in vivo and in vitro. Eur. J. Biochem. 189: 553–558. Beizer, J. L. and M. L. Timiras. 1994. Pharmacology and drug management in the elderly. Pp. 279–284

References

in P. S. Timiras (ed.), Physiological Basis of Aging and Geriatrics, 2nd ed. CRC Press, Boca Raton, FL. Bell, A. G. 1918. The Duration of Life and Conditions Associated with Longevity. A Study of the Hyde Genealogy. Genealogical Record Office, Washington, DC. Bell, E., L. F. Mareck, D. S. Levinstone, C. Merrioll, S. Sher, I. T. Yong and M. Eden. 1978. Loss of division potential in vitro: Aging or differentiation. Science 202: 1158–1162. Bell, G. 1985. Evolutionary and nonevolutionary theories of senescence. Am. Nat. 124: 600–603. Bell, G. 1988. Uniformity and diversity in the evolution of sex. Pp. 126–138 in R. E. Michod and B. R. Levin (eds.), The Evolution of Sex. Sinauer Associates, Sunderland, MA. Bellamy, D. 1988. Degenerative diseases of ageing as problems of natural selection: A commentary upon the concept of “normal ageing.” Gerontology 34: 315–326. Benard, C., et al. 2002. The C. elegans maternal-effect gene clk-2 is essential for embryonic development, encodes a protein homologous to yeast Tel2p, and is required for the regulation of telomere length. Development xx:ssss. Benne, R. and H. F. Tabak. 1986. Senescence comes of age. Trends Genet. 2: 147–148. Ben-Porath, I. and R. A. Weinberg. 2004. When cells get stressed: An integrative view of cellular senescence. J. Clin. Invest. 113: 8–13. Ben-Yehuda, A., P. Szabo, R. Dyall and M. E. Weksler. 1994. Bone marrow declines as a site of B-cell precursor differentiation with age: Relationship to thymus involution. Proc. Natl. Acad. Sci. USA 91: 11988–11992. Benzi, G., F. Marzatico, O. Pastoris and R. F. Villa. 1989. Relationship between aging, drug treatment and the cerebral enzymatic antioxidant system. Exp. Gerontol. 24: 137–148. Berlett, B. S. and E. R. Stadtman. 1997. Protein oxidation in aging, disease, and oxidative stress. J. Biol. Chem. 272: 20313–20316. Bernal, Y. N. and F. Stern. 2004. Energy restriction controls aging through neuroendocrine signal transduction. Ageing Res. Rev. 3: 189–198. Bernet, J. 1992. In Podospora anserina, protoplasmic incompatibility genes are involved in cell death control via multiple gene interactions. Heredity 68: 79–87. Bernier, L. and E. Wang. 1996. A prospective view on phosphatases and replicative senescence. Exp. Gerontol. 31: 13–19. Bernstein, C. and H. Bernstein. 1991. Aging, Sex and DNA Repair. Academic Press, San Diego, CA. Berr, C., B. Balansard, J. Arnaud, A.-M. Roussel and A. Alperovitch. 2000. Cognitive decline is associ-

531

ated wth systemic oxidative stress: The EVA study. J. Am. Geriatr. Soc. 48: 1285–1291. Bewley, G. and C. C. Laurie-Ahlberg. 1984. Genetic variation affecting the expression of catalase in Drosophila melanogaster: Correlations with rates of enzyme synthesis and degradation. Genetics 106: 435–448. Bewley, G. C. and W. J. Mackay. 1989. Development of a genetic model for acatalasemia: Testing the oxygen free radical theory of aging. Pp. 359–378 in D. E. Harrison (ed.), Genetic Effects of Aging, vol. 2. Growth Publishing, Bar Harbor, ME. Bhole, D., M. J. Akkikian and J. Tower. 2004. Doxcycline-regulated over-expression of hsp22 has negative effects on stress resistance and life span in adult Drosophila melanogaster. Mech. Ageing Dev. 125: 651–663. Bidder, G. P 1925. The mortality of plaice. Nature 115: 495–496. Bidder, G. P. 1932. Senescence. Br. Med. J. 115: 5831. Binstock, R. H. 2004a. The prolonged old, the longlived society, and the politics of age. Pp. xx–xx in S. G. Post and R. H. Binstock (eds.), The Fountain of Youth: Cultural, Scientific, and Ethical Perspectives on a Biomedical Goal. Oxford University Press, New York. Binstock, R. H. 2004b. The search for prolongevity: A contentious pursuit. Pp. xx–xx in S. G. Post and R. H. Binstock (eds.), The Fountain of Youth: Cultural, Scientific, and Ethical Perspectives on a Biomedical Goal. Oxford University Press, New York. Bird, J., E. L. Ostler and R. G. A. Faragher. 2003. Can we say that senescent cells cause ageing? Exp. Gerontol. 38: 1319–1326. Biron, P., J. G. Mongeau and D. Bertrand. 1976. Familial aggregation of blood pressure in 568 adopted children. Can. Med. Assoc. J. 115: 773–774. Bittles, A. H., B. A. Peterson, S. G. Sullivan, R. Hussain, E. J. Glasson and P. D. Montgomery. 2002. The influence of intellectual disability on life expectancy. J. Gerontol. Med. Sci. 57A: M470–M472. Björksten, J. 1968. The cross linkage theory of aging. J. Am. Geriatr. Soc. 16: 408–427. Björksten, J. and W. J. Champion. 1942. Mechanical influence upon tanning. J. Am. Chem. Soc. 64: 868–869. Blackett, A. D. and D. A. Hall. 1981a. The effect of vitamin E on mouse fitness and survival. Gerontology 27: 133–139. Blackett, A. D. and D. A. Hall. 1981b. Tissue vitamin E levels and lipofuscin accumulation with age in the mouse. J. Gerontol. 36: 529–533. Blalock, E. M., K-C Chen, K. Sharrow, J. P. Herman, N. M. Porter, T. C. Foster and P. W. Landfield. 2003. Gene microarrays in hippocampal aging: Statistical profiling identifies novel processes

532

References

correlated with cognitive impairment. J. Neurosci. 23: 3807–3819. Block, G. 1991. Vitamin C and cancer prevention: The epidemiologic evidence. Am. J. Clin. Nutr. 53(suppl. 1): 270S–282S. Bloom, F. E., A. Lazerson and L. Hofstadter. 1985. Brain, Mind and Behavior. Freeman, New York. Bloom, W. and D. W. Fawcett. 1968. A Textbook of Histology, 9th edition. Saunders, Philadelphia. Bluher, M., B. B. Kahn and C. R. Khan. 2003. Extended longevity in mice lacking the insulin receptor in adipose tissue. Science 299: 572–574. Bluher, M., M. D. Michael, O. D. Peroni, K. Ueki, N. Carter, B. B. Kahn, et al. 2002. Adipose tissue selective insulin receptor knockout protects against obesity and obesity-related glucose intolerance. Dev. Cell 3: 25–38. Blumenthal, H. T. 2002. The aging-disease dichotomy: True or false? J. Gerontol. Biol. Sci. 58A: 138–145. Bodkin, N. L., T. M. Alexander, H. K. Ortmeyer, E. Johnson and B. C. Hansen. 2003. Mortality and morbidity in laboratory-maintained rhesus monkeys and effects of long-term dietary restriction. J. Gerontol. Biol. Sci. 58A: 212–219. Bodnar, A. G., M. Ouellette, M. Frolkis, S. E. Holt, C. P. Chiu, G. B. Morin, C. B. Harley, J. W. Shay, S. Lichteiner and W. E. Wright. 1998. Extension of life-span by introduction of telomerase into normal human cells. Science 279: 349–352. Bohr, V. A. and R. M. Anson. 1995. DNA damage, mutation and fine structure DNA repair in aging. Mutat. Res. 338: 25–34. Borkan, G. A. and A. H. Norris. 1980. Assessment of biological age using a profile of physical parameters. J. Gerontol. 35: 177–184. Borras, C., J. Sastre, D. Garcia-Sala, A. Lloret, F. V. Pallardo and J. Vina. 2003. Mitochondria from females exhibit higher antioxidant gene expression and lower oxidative damage than males. Free Rad. Biol. Med. 34: 546–552. Bortoluzzi, S., C. Romualdi, A. Bisognin and G. A. Danieli. 2003. Disease genes and intracellular protein networks. Physiol. Genomics 15: 223–227. Boukamp, P. 2003. Biological clocks in the aging cell. Pp 107–119 in T. von Zglincki (ed.), Aging at the Molecular Level. Kluwer Academic Publishers, Dordrecht, The Netherlands. Boulding, K. E. 1965. The menace of Methuselah: Possible consequences of increased life expectancy. J. Washington Acad. Sci. 55: 171–179. [Reprinted in Population and Development Review, 29: 493– 504.] Bouliere, F. and S. Parot. 1962. Le veillissement de deux populations blanches vivant dans des conditions ecologiques tres differentes, stude comparative. Rev. France Etudes Clin. Biol. 7: 629–635. Boulton, S. J., A. Gartner, J. Reboul, P. Vaglio, N.

Dyson D. E. Hill and M. Vidal. 2002. Combined functional genomic maps of the C. elegans DNA damage response. Science 295: 127–131. Bowden, D. M. 1990. Primate aging research in the USA and the challenge of measuring the rate of aging in long-lived species. Jikken Dobutsu 39: 183–184. Bowden, D. M., R. Short, D. D. Williams and E. A. Clark. 1994. Immunologic markers in a longitudinal study of aging in pigtailed macaques (Macaca nemestrina). J. Gerontol. Biol. Sci. 49: B93–B103. Boyle, W. J., W. S. Simonet and D. L. Lacey. 2003. Osteoclast differentiation and activation. Nature 423: 337–342. Braeckman B. P., K. Houthoofd, K. Brys, I. Lenaerts, A. De Vreese, S. Van Eygen, H. Raes and J. R. Vanfleteren. 2002. No reduction of energy metabolism in Clk mutants. Mech. Ageing Dev. 123: 1447–1456. Braeckman, B. P., K. Houthoofd and J. R. Vanfleteren. 2001. Insulin-like signaling, metabolism, stress resistance and aging in Caenorhabditis elegans. Mech. Aging Dev. 122: 673–693. Brandfonbrener, M., M. Landowne and N. W. Shock. 1955. Changes in cardiac output with age. Circulation 12: 557–566. Branicky, R., C. Benard and S. Hekimi 2000. clk-1, mitochondria, and physiological rates. BioEssays 22(1): 48–56. Brant, L. J., J. L. Fozard and E. J. Metter. 1994. Age differences in biological markers of mortality. Pp. 458–470 in A. K. Balin (ed.), Practical Handbook of Human Biologic Age Determination. CRC Press, Boca Raton, FL. Bray, D. 2003. Molecular networks: The top-down view. Science 301: 1864–1865. Bremner, W. J., A. M. Matsumoto, R. A. Steiner, D. K. Clifton and D. M. Dursa. 1986. Neuroendocrine correlates of aging in the male. Pp. 47–57 in L. Mastroianni Jr. and C. A. Paulsen (eds.), Aging, Reproduction and the Climacteric. Plenum Press, New York. Bremner, W. J., M. V. Vitiello and P. N. Prinz. 1983. Loss of circadian rhythmicity in blood testosterone levels with aging in normal men. J. Clin. Endocrinol. Metab. 56: 1278–1281. Brenneisen, P., J. Gogol and K. Bayreuther. 1994. DNA synthesis and fos and jun protein expression in mitotic and post mitotic WIO38 fibroblasts in vitro. Exp. Cell Res. 211: 219–230. Brenner, D. A., M. O’Hara, P. Angel, M. Chojkrer and M. Karin. 1989. Prolonged activation of jun and collagenase genes by tumour necrosis factor. Nature 337: 661–663. Brenner, S. 1974. The genetics of Caenorhabditis elegans. Genetics 77: 87–94. Broccoli, D., J. W. Young and T. deLange. 1995.

References

Telomerase activity in normal and malignant hematopoietic cells. Proc. Natl. Acad. Sci. USA 92: 9082–9086. Brockman, J. (ed.) 2002. The Next Fifty Years: Science in the First Half of the Twenty-First Century. Vintage Books, Random House. New York. Brody, H. and N. Vijayashankar. 1977. Anatomical changes in the nervous system. Pp. 241–261 in C. E. Finch and E. L. Schneider (eds.), Handbook of the Biology of Aging. Van Nostrand Reinhold, New York. Brody, J. A. and T. P. Miles. 1990. Mortality postponed and the unmasking of age-dependent non-fatal conditions. Aging 2: 283–289. Bronikowski, A. M., S. C. Alberts, J. Altmann, C. Packer, K. D. Carey and M. Tatar. 2002. The aging baboon: Comparative demography in a nonhuman primate. Proc. Natl. Acad. Sci. USA 99: 9591–9595. Bronikowski, A. M., P. A. Carter, T. J. Morgan, T. Garland Jr., N. Ung, T. D. Pugh, R. Weindruch and T. A. Prolla. 2003. Lifelong voluntary exercise in the mouse prevents age-related alterations in gene expression in the heart. Physiol. Genomics 12: 129–138. Bronowski, J. 1956. Science and Human Values [revised ed. 1965]. Harper and Row, New York. Bronson, R., D. Birt and S. N. Meydani, 1999. Biomarkers as early predictors of long-term health status and human immune function. Nutr. Rev. 57: S7–S12. Brooks, A., G. J. Lithgow and T. J. Johnson. 1994. Mortality rates in a genetically heterogeneous population of Caenorhabditis elegans. Science 263: 668–671. Broome, C. S., A. Vasilaki and A. McArdle. 2003. Skeletal muscle aging. Pp. 73–99 in R. Aspinall (ed.), Aging of Organs and Systems, Kluwer Academic Publishers, Dordrecht, the Netherlands. Brouilette, S., R. K. Singh, J. R. Thompson, A. H. Goodall and N. J. Samani. 2003. White cell telomere length and risk of premature myocardial infarction. Arterioscler. Thromb. Vasc. Biol. http://www.atvbaha.org, DOI: 10.1161/01.ATV .000000674626.96344.32. Brown, G. C. 1992. The leaks and slips of bioenergetic membranes. FASEB J. 6: 2961–2965. Brown, G. W. and M. M. Flood. 1947. Tumbler mortality. J. Am. Stat. Assoc. 42: 562–574. Brown, M. 1987. Changes in fibre size, not number, in ageing skeletal muscle. Age Ageing 16: 244–248. Brown, M. S. and J. L. Goldstein. 1996. Heart attacks: Gone with the century? (editorial). Science 272: 629. Brown, W. T. 1985. Genetics of human aging. Pp. 105– 114 in M. Rothstein (ed.), Review of Biological Research in Aging, vol. 2. Alan R. Liss, New York.

533

Brown, W. T. 1992. Progeria: A human-disease model of accelerated aging. Am. J. Clin. Nutr. 55(suppl. 6): 1222S–1224S. Brown-Borg, H. M., K. E. Borg, C. J. Meliska and A. Bartke. 1996. Dwarf mice and the ageing process. Nature 384: 33. Browner, W. S., J. Westenhouse and J. A. Tice. 1991. What if Americans ate less fat? A quantitative estimate of the effect on mortality. J. Am. Med. Assoc. 265: 3285–3291. Browning, R. 1864. Rabbbi Ben Ezra. P. 306 in T. R. Cole and M. G. Winkler (eds.), The Oxford Book of Aging: Reflections on the Journey of Life. Oxford University Press, Oxford, 1994. Brown-Sequard, C. E. 1889. Des effects produits chez l’homme par des injections sous-cutanees d’un liquide retire des testicules frais de cobayes et de chiens. Comptes Rend. Soc. Biol. 41: 415–422. Brunk, C. F. 1986. Genome reorganization in Tetrahymena. Intl. Rev. Cytol. 99: 49–83. Bryan, T. M., A. Englezou, J. Gupta, S. Bacchetti and R. R. Reddel. 1995. Telomere elongation in immortal human cells without detectable telomerase activity. EMBO J. 14: 4240–4248. Bucala, R., R. Model and A. Cerami. 1984. Modification of DNA by reducing sugars: A possible mechanism for nucleic acid aging and age-related dysfunctions in gene expression. Proc. Natl. Acad. Sci. USA 81: 105–109. Buck, S. and R. Arking. 2001. Metabolic alterations in genetically selected Drosophila strains with different longevities. J. Am. Aging Assoc. 24: 151– 162. Buck, S., M. Nicholson, S. P. Dudas, G. T. Baker III and R. Arking. 1993a. Larval regulation of adult longevity in a genetically selected long lived strain of Drosophila melanogaster. Heredity 71: 23–32. Buck, S., Vettriano, J. and Arking, R. 2000. Extended longevity in Drosophila is consistently associated with a decrease in developmental viability. J. Gerontol.: Biol. Sci. 55A: B292–B301. Buck, S., R. A. Wells, S. P. Dudas, G. T. Baker III and R. Arking. 1993b. Chromosomal localization and regulation of the longevity determinant genes in a selected strain of Drosophila melanogaster. Heredity 71: 11–22. Buffon, G. L. Leclerc, comte de. 1749–1804. Histoire naturelle, générale et particulière. 44 vols. Imprimerie Royale, Paris. Burks, D. J., J. F. de Mora, M. Schubert, D. J. Withers, M. G. Myers, H. H. Towery, S. L. Altamuro, C. L. Flint and M. F. White. 2000. IRS-2 pathways integrate female reproduction and energy homeostasis. Nature 407: 377–382. Burtless, G. and J. F. Quinn. 2003. Living longer: living better: The policy challenge of an aging work force. Public Policy Aging Rept. 11(3): 5–11.

534

References

Buskirk, E. R. 1985. Health maintenance and longevity: Exercise. Pp. 894–931 in C. E. Finch and E. L. Schneider (eds.), Handbook of the Biology of Aging, 2nd ed. Van Nostrand Reinhold, New York. Busuttil, R. A., M. Rubio, M. E. T. Dolle, J. Campisi and J. Vijg. 2003. Oxidative damage, somatic mutations and cellular Aging. Pp 79–90 in T. von Zglincki (ed.), Aging at the Molecular Level. Kluwer Academic Publishers, Dordrecht, the Netherlands. Butler, R. 1995. Human aging study. Pp. 481–482 in G. L. Maddox (ed.), The Encyclopedia of Aging, 2nd ed. Springer, New York. Butlin, R. K. and H. I. Griffiths. 1993. Ageing without sex. Nature 364: 680. Butterfield, D. A., B. J. Howard, S. Yatin, K. L. Allen and J. M. Carney. 1997. Free radical oxidation of brain proteins in accelerated senescence and its modulation by N-tert-butyl-alpha-phenylnitrone. Proc. Natl. Acad. Sci. USA 94: 674–678. Cadenas, E. and K. J. A. Davies. 2000. Mitochondrial free radical generation, oxidative stress, and aging. Free Rad. Biol. Med. 29: 222–230. Calabrese, E. J. and L. A. Baldwin. 1999. Chemical hormesis: Its historical foundations as a biological hypothesis. Toxicol. Pathol. 27: 195–216. Callahan, D. 1987. Setting Limits: Medical Goals in an Aging Society. Simon & Schuster, New York. Callahan, D. 1999. Age, sex, & resource allocation. Pp. xx–xx in M. U. Walker (ed.), Mother Time: Women, Aging and Ethics. Rowman & Littlefield, Lanham, MD. Calleja, M., P. Pena, C. Ugalde, C. Ferreiro, R. Marco and R. Garesse. 1993. Mitochondrial DNA remains intact during Drosophila aging, but the levels of mitochondrial transcripts are significantly reduced. J. Biol. Chem. 268: 18891–18897. Camougrand, N. and M. Rigoulet. 2001. Aging and oxidative stress: studies of some genes involved both in aging and in response to oxidative stress. Respir. Physiol. 128: 393–401. Campbell, S. D., A. J. Hilliker and J. P. Phillips. 1986. Cytogenetic analysis of the cSOD microregion in Drosophila melanogaster. Genetics 112: 205–215. Campfield, L. A., F. J. Smith, Y. Glusez, R. Devos and P. Burn. 1995. Recombinant mouse OB protein: Evidence for a peripheral signal linking adiposity and central neural networks. Science 269: 546–548. Campisi, J. 1996. Replicative senescence: An old lives’ tale? Cell 84: 497–500. Camus, J. P. 1966.[Gout, diabetes, hyperlipemia: a metabolic trisyndrome]. Rev. Rhum. Mal. Osteoartic. 33: 10–14. (In French). Cantin, M. and J. Genest. 1986. The heart as an endocrine organ. Sci. Am. 254(2): 76–81. Cao, S. X., J. M. Dhahbi, P. L. Mote and S. R. Spindler. 2001. Genomic profiling of short-and long-term

caloric restriction in the liver of aging mice. Proc. Natl. Acad. Sci. USA 98: 10630–10635. Caplan, A. L. 2004. An unnatural process: Why it is not inherently wrong to seek a cure for aging. Pp. x–x in S. G. Post and R. H. Binstock (eds.), The Fountain of Youth: Cultural, Scientific, and Ethical Perspectives on a Biomedical Goal. Oxford University Press, New York. Cardozo-Pelaez, F., S. Song, A. Parthasaathy, C. J. Epstein and J. Sanchez-Ramos. 1998. Attenuation of age-dependent oxidative damage to DNA and protein in brainstem of tg CuZnSOD mice. Neurobiol. Aging 19: 311–316. Carey, J. R. 2001. Demographic mechanisms for the evolution of long life in social insects. Exp. Gerontol. 36: 713–722. Carey, J. R. 2003. Longevity: The Biology and Demography of Life Span. Princeton University Press, Princeton, NJ. Carey, J. R., P. Liedo, L. Harshman, Y. Zhang, H. G. Muller, L. Partridge, J. L. Wang. 2002. Life history response of Mediterranean fruit flies to dietary restriction. Aging Cell 1: 140–148. Carey, J. R., P. Liedo, H. G. Muller, J. L. Wang and J. M. Chiou. 1999. Mortality oscillations induced by periodic starvation alter sex-mortality differentials in Mediterranean fruit flies. J Gerontol: Biol. Sci. 54: B424–B431. Carey, J. R., P. Liedo, H.-G. Muller, L-L. Wang, Y. Shang and L. Harshman. 2004. Stochastic dietary restriction using a Markov-chairn feeding protocol elicits complex life history response in medflies. Aging Cell, 10 December 2004; DOI: 10.1111/j.1474-9728.2004.00140.x. Carey, J. R., P. Liedo, D. Orozco and J. W. Vaupel. 1992. Slowing of mortality rates at older ages in large medfly cohorts. Science 258: 457–461. Carey, J. R. and S. Tuljapurkar (eds.). 2003. Life Span: Evolutionary, Ecological, and Demographic Perspectives. Population Council, New York. Cargill, S. L., J. R. Carey, H.-G. Muller and G. Anderson. 2003. Age of ovary determines remaining life expectancy in old ovariectomized mice. Aging Cell 2: 185–190. Carlen, M., R. M. Cassidy, H. Brismar, G. A. Smith, L. W. Enquist and J. Frisen. 2002. Functional integration of adult-born neurons. Curr Biol. 12: 606–608. Carlson, A. 1987. Brain neurotransmitters in aging and dementia: Similar changes across diagnostic dementia groups. Gerontology 33: 159–167. Carnes, B. A., S. J. Olshansky and D. Grahn. 2003. Biological evidence for limits to the duration of life. Biogerontology 4: 31–45. Carrel, A. 1912. On the permanent life of tissues outside the organism. J. Exp. Med. 15: 516–528. Carter, B. D., C. Kaltschmidt, B. Kaltschmidt, N.

References

Offenhauser, R. Bohm-Matthaei, P. A. Baeuerle and Y-A. Barde. 1996. Selective activation of NF-kB by nerve growth factor through the neurotrophin receptor p75. Science 272: 542–545. Carter, C. S., M. M. Ramsey and W. E. Sonntag. 2002. A critical analysis of the role of growth hormone and IGF-1 in aging and life span. Trends Genet. 18: 295–301. Caspari, R. and S-H. Lee. 2004. Older age becomes common late in human evolution. Proc. Natl. Acad. Sci. USA. www.pnas.org/cgi/doi/10.1073. pnas.0402857101. Casper, R. C. 1995. Nutrition and its relationship to aging. Exp. Gerontol. 30: 299–314. Castrillon, D. H., L. Miao, R. Kollipara, J. W. Horner and R. A. DePhenho. 2003. Suppression of ovarian follicle activation in mice by the transcription factor foxo3a. Science 301: 215–218. Cawthon, R. M., K. R. Smith, E. O’Brien and R. A. Kerber. 2003. Association between telomere length in blood and mortality in people aged 60 years or older. The Lancet 361: 393–395. Cefalu, W. T., A. D. Bell-Farrow, Z. Q. Wang, W. E. Sonntag, M. X. Fu, J. W. Baynes and S. R. Thorpe. 1995. Caloric restriction decreases age-dependent accumulation of the glycoxidation products, N epsilon-(carboxymethyl)lysine and pentosidine, in rat skin collagen. J. Gerontol. Biol. Sci. 50A: B337–B341. Cerami, A. 1985. Glucose as a mediator of aging. J. Am. Geriatr. Soc. 33: 626–634. Cerami, A., H. Vlassara and M. Brownlee. 1987. Glucose and aging. Sci. Am. 256(5): 90–97. Champagne, R and MJ Meaney. 2001. Like mother, like daughter: Evidence for non-genomic transmission of parental behavior and stress responsivity. Prof. Brain Res. 133: 287–302. Chang, K.T. and K-T. Min. 2002. Regulation of lifespan by histone deacetylase. Ageing Res. Rev. 1: 313–326. Chang-Claude, J., R. Frentzel-Beyme and U. Eilber. 1992. Mortality pattern of German vegetarians after 11 years of follow-up. Epidemiology 3: 395– 401. Chapman, A. R. 2004. The social and justice implications of extending the human life span. In S. G. Post and R. H. Binstock (eds.), The Fountain of Youth: Cultural, Scientific, and Ethical Perspectives on a Biomedical Goal. Oxford University Press, New York. Chapman T. and L. Partridge. 1996. Female fitness in Drosophila melanogaster: an interaction between the effect of nutrition and of encounter rate with males. Proc. R. Soc. Lond. B 263: 755–759. Charlesworth, B. 1994a. Evolution in Age-Structured Populations, 2nd ed. Cambridge University Press, Cambridge.

535

Charlesworth, B. 1994b. Evolutionary mechanisms of senescence. Pp. 13–21 in M. R. Rose and C. E. Finch (eds.), Genetics and Evolution of Aging. Kluwer Academic Publishers, Dordrecht, the Netherlands. Charlesworth, B. 1996. Evolution of senescence: Alzheimer’s disease and evolution. Curr Biol. 6: 20–22. Chauhan, J., Z. J. Hawrysh, M. Gee, E. A. Donald and T. K. Basu. 1987. Age-related olfactory and taste changes and interrelationships between taste and nutrition. J. Am. Dietetic Assoc. 87: 1543–1550. Cheal, M. L. 1986. The gerbil, a unique model for research on aging. Exp. Aging Res. 12: 3–21. Chen, J. B., J. Sun and S. M. Jazwinski. 1990. Prolongation of the yeast life span by the v-Ha-RAS oncogene. Mol. Microbiol. 4: 2081–2086. Chen, J. Q., M. Eshete, W. L. Alwoprth and J. D. Yager. 2004. Binding of MCF-7 cell mitochondrial proteins and recombinant human estrogen receptors alpha and beta to human mitochondrial DNA estrogen response elements. J. Cell Biochem. 93: 358–373. Chen, M., J. B. Halter and D. Porte, Jr. 1987. The role of dietary carbohydrate in the decreased glucose tolerance of the elderly. J. Am. Geriatr. Soc. 35: 417–424. Cherif, H., J. L. Tarry, S. E. Ozanne and C. N. Hales. 2003. Ageing and telomeres: A study into organand gender-specific telomere shortening. Nuclei Acids Res. 31: 1576–1583. Chien, K. R. and G. Karsenty. 2005. Longevity and lineages: toward the integrative biology of degenerative diseases in heart, muscle, and bone. Cell 120: 533–544. Chien, K. R., A. Moretti and K.-L. Laugwitz. 2004. ES cells to the rescue. Science 306: 239–240. Chimenti, C., J. Kajstura, D. Torella, K. Urbanek, H. Heleniak, C. Colussi, F De Meglio, B. NadalGinard, A. Frustaci, A. Leri, A. Maweri and P. Anversa. 2003. Circulat. Res. 93: 604–613. Chippindale, A. K., A. M. Leroi, S. B. Kim and M. R. Rose. 1993. Phenotypic plasticity and selection in Drosophila life-history evolution. I. Nutrition and the cost of reproduction. J. Evol. Biol. 6: 171–193. Chisholm, G. M. and D. Steinberg. 2000. The oxidative modification hypothesis of atherogenesis: An overview. Free Rad. Biol. Med. 28: 1815–1826. Chou, M. W., R. A. Pegram, P. Gao, S. R. Hansard, J. G. Shaddock and D. A. Casciano. 1991. The effects of dietary restriction and aging on in vivo and in vitro binding of aflatoxin B1 to cellular DNA. Biomed. Environ. Sci. 4: 134–143. Chou, M. W., J. Kong, K. T. Chung and R. W. Hart. 1993. Effect of caloric restriction on the metabolic activation of xenobiotics. Mutat Res. 295: 223– 235.

536

References

Chow, D. K., W. Ibrahim, Z. Wei and A. C. Chan. 1999. Vitamin E regulates mitochondrial hydrogen peroxide generation. Free Rad. Biol. Med. 27: 580–587. Christenson, B. A. and N. E. Johnson. 1995. Educational inequality in adult mortality: An assessment with death certificate data from Michigan. Demography 32: 215–229. Chung, H. Y., H. J. Kim, K. J. Jung, J. S. Moon, M. A. Yoo, K. W. Kim and B. P Yu. 2000a. The inflammatory process in aging. Rev. Clin. Gerontol. 10: 207–222. Chung, H. Y., H. J. Kim, J. W. Kim and B. P. Yu. 2000b. The inflammation hypothesis of aging: molecular modulation by caloric restriction. Ann. N.Y. Acad. Sci. 928: 327–335. Chung, H. Y., H. J. Kim, K. W. Kim, J. S. Choi and B. P. Yu. 2002. Molecular inflammation hypothesis of aging based on the anti-aging mechanism of calorie restriction. Microsc Res Tech. 59: 264–272. Chung, H. Y., H. J. Kim, J. W. Kim and B. P. Yu. 2001. The inflammation hypothesis of aging: molecular modulation by calorie restriction. Ann. N.Y. Acad. Sci. 928: 327–335. Clancy, D. J., D. Gems, E. Hafen, S. J. Leevers and L. Partridge. 2002. Dietary restriction in long-lived dwarf flies. Science 296: 319. Clancy, D. J., D. Gems, L. G. Harshman, S. Oldham, H. Stocker, E. Hafen, S. J. Leevers and L. Partridge. 2001. Extension of life-span by loss of CHICO, a Drosophila insulin receptor substrate protein. Science 292: 104–107. Clark, A. G. 1994. Mutation-selection balance and the evolution of senescence. Pp. 22–28 in M. R. Rose and C. E. Finch (eds.), Genetics and Evolution of Aging. Kluwer Academic Publishers, Dordrecht, the Netherlands. Clark, A. M., H. A. Bertrand and R. E. Smith. 1963. Life-span differences between haploid and diploid males of Habrobracon serinopae after exposure as adults to X-rays. Am. Nat. 97: 203–208. Clark, D. H. 1975. Exercise Physiology. Prentice-Hall, Englewood Cliffs, NJ. Clark, M. A. and A. S. Weiss. 1993. Elevated levels of glycoprotein gp200 in progeria fibroblasts. Mol. Cell. Biochem. 120: 51–60. Clarke, A. C. 1963. Profiles of the Future: An Inquiry Into the Limits of the Possible. Harper and Row, New York. Clarke, J. M. and J. Maynard Smith. 1955. The genetics and cytology of Drosophila subobscura. XI. Hybrid vigor and longevity. J. Genet. 53: 172– 180. Clarkson, T. B., M. R. Adams, K. W. Weingard, L. C. Miller and S. Hendrick. 1987. Effect of age on atherosclerosis progression in non-human primates. Pp. 57–71 in S. R. Bates and E. C. Gangloff

(eds.), Atherogenesis and Aging. Springer-Verlag, New York. Cohen, B. H. 1964. Family patterns of mortality and life span. Q. Rev. Biol. 39: 130–181. Cohen, H. Y., C. Miller, K. J. Bitterman, N. R. Walt, B. Hekking, B. Kessler, K. T. Howitz, M. Goroope, R. deCabo and D. Sinclair. 2004. Caloric restriction promotes mammalian cell survival by inducing the SIRT1 deacetylase. Science 305: 390–392. Cohen, M., L. Cheng and H. N. Bhagavan. 1994. Vitamin C and the elderly—an update. Pp. 203–262 in R. Watson (ed.), Handbook of Nutrition in the Aged, 2nd ed. CRC Press, Boca Raton, FL. Cohn, L., A. G. Feller, M. W. Draper, I. W. Rudman and D. Rudman. 1993. Carpal tunnel syndrome and gynecomastia during growth hormone treatment of elderly men with low circulating IGF-1 concentrations. Clin. Endocrinol. 39: 417–429. Colcombe, S. J., K. I. Erikson, N. Raz, A. G. Webb, N. J. Cohen, E. McAuley and A. F. Kramer. 2003. Aerobic fitness reduces brain tissue loss in aging humans. J. Gerontol. Med. Sci. 58A: 176–180. Coleman, D. L., K. W. Schwizer and E. Letier. 1984. Effect of genetic background on the therapeutic effects of dehydroepiandrosterone (DHEA) in diabetes-obesity mutants used in aged mice. Diabetes 33: 26–32. Coleman, R., M. Silbermann, D. Gershon and A. Z. Reznick. 1987a. Effect of long term stress on the ultrastructure of the aging mouse heart. Gerontology 33: 19–33. Coleman, R., M. Silbermann, D. Gershon and A. Z. Reznick. 1987b. Giant mitochondria in the myocardium of aging and endurance-trained mice. Gerontology 33: 34–39. Colige, A., B. Nusgens and C. M. Lapiere. 1991a. Altered response of progeria fibroblasts to epidermal growth factor. J. Cell Sci. 100: 649–655. Colige, A., J. C. Roujeau, F. De la Rocque, B. Nusgens and C. M. Lapiere. 1991b. Abnormal gene expression in skin fibroblasts from a Hutchinson-Gilford patient. Lab. Invest. 64: 799–806. Comfort, A. 1956. The Biology of Senescence. Holt, Rinehart and Winston, New York. Comfort, A. 1960. Discussion session. I. Definition and universality of aging. Pp. 3–13 in B. L. Strehler (ed.), The Biology of Aging. AIBS, Washington, DC. Comfort, A. 1964. Ageing: The Biology of Senescence, 2nd ed. Holt, Rinehart and Winston, New York. Comfort, A. 1979. The Biology of Senescence, 3rd edition. Elsevier, New York. Consumers Union 1992. Special report on how to live longer. January issue of Consumer Reports magazine. Conter-Audonneau, J. L., C. Jeanmaire and G. Pauly. 1999. A histological study of human wrinkle struc-

References

tures: comparison between sun-exposed areas of the face, with or without wrinkles, and sunprotected areas. B. J. Dermatol. 140: 1038–1047. Cook, F. L. 2003. Living longer: living better: The challenge to policy makers, an overview. Public Policy Aging Rpt. 11(3): 2–4 Cooney, C. A., A. A. Dave and G. L. Wolff. 2002. Maternal methyl supplements in mice affect epigenetic variation and DNA methylation of offspring. J. Nutr. 132: 23935–24005. Cooper, E. L. (ed.). 1984. Stress, Immunity, and Aging. Marcel Dekker, New York. Corpas, E., S. M. Harman, M. A. Pineyro, R. Roberson and M. R. Blackman. 1992. Growth hormone (GH)-releasing hormone (1–29) twice daily reverses the decreased GH and insulin like growth factor-I levels in old men. J. Clin. Endocrinol. Metab. 75: 530–535. Corral-Debrinski, M., T. Horton, M. T. Lott, J. M. Shoffner, M. F. Beal and D. C. Wallace. 1992. Mitochondrial DNA deletions in human brain: Regional variability and increase with advanced age. Nature Genet. 2: 324–329. Cortopassi, G. A. and E. Wang. 1996. There is substantial agreement among interspecies estimates of DNA repair activity. Mech. Ageing Dev. 91: 211– 218. Corwin, J., M. Loury and A. N. Gilbert. 1997. Workplace, age, and sex as mediators of olfactory function: Data from the National Geographic Smell Survey. J. Gerontol. Psychol. Sci. 50B: P179– P186. Coschigano, K. T., D. Clemmons L. L. Bellush and J. J. Kopchick. 2000. Assessment of growth parameters and life span of GHR/BP gene-disrupted mice. Endocrinology 141: 2608–2613. Costa, P. T. and R. R. McCrae. 1980. Functional age: A conceptual and empirical critique. Pp. 23–49 in S. G. Haynes and M. Feinleib (eds.), Proceedings of the Second Conference on the Epidemiology of Aging. NIH Publ. no. 80-969, National Institutes of Health, Washington, DC. Costa, P. T. and R. R. McCrae. 1984. Concepts of functional or biological age: A critical review. Pp. 30–37 in R. Andres, E. L. Bierman and W. R. Hazzard (eds.), Principles of Geriatric Medicine. McGraw-Hill, New York. Costa, P. T. and R. R. McCrae. 1995. Design and analysis of aging studies. Pp. 25–36 in E. J. Masoro (ed.), Handbook of Physiology, Section 11: Aging. Oxford University Press, New York. Costa, P. T., R. R. McCrae and D. Arenberg. 1983. Recent longitudinal research on personality and aging. Pp. 222–265 in K. W. Schaie (ed.), Longitudinal Studies of Adult Psychological Development. Guilford, New York. Cotman, C. W., R. E. Brinton, A. Galaburda, B.

537

McEwen and D. M. Schneider. 1987. The NeuroImmune-Endocrine Connection. Raven Press, New York. Cotman, C. W. and V. K. Holets. 1985. Structural changes at synapses with age: Plasticity and regeneratrion. Pp. 617–644 in C. E. Finch and E. L. Schneider (eds.), Handbook of the Biology of Aging, 2nd ed. Van Nostrand Reinhold, New York. Cotman, C. W. and S. Neeper. 1996. Activity-dependent plasticity and the aging brain. Pp. 284–299 in E. L. Schneider and J. W. Rowe (eds.), Handbook of the Biology of Aging, 4th edition. Academic Press, San Diego, CA. Courtright, J. B., J. Sonstein and A. K. Kumaran. 1985. Age specific regulation of gene expression in Drosophila. Pp. 209–222 in R. S. Sohal, L. S. Birnbaum and R. G. Cutler (eds.), Molecular Biology of Aging: Gene Stability and Gene Expression. Raven Press, New York. Couzin, J. 2005. How much can human life span be extended? Science 309: 83. Couzin, J., G. Vogel and C. Holden. 2004. Renovating the heart. Science 304: 192–194. Craft, S., S. Asthana, D. G. Cook, L. D. Baker, M. Cherrier, K. Purganan, C. Wait, et al. 2003 . Insulin dose-response effects on memory and plasma amyloid precursor protein in Alzheimer’s disease: Interactions with apolipoprotein E genotype. Psychoneuroendocrinology 28: 809–822. Crawford, D. R., G. P. Schools, S. L. Salmon and K. J. A. Davies. 1996. Hydrogen peroxide induces the expression of adapt15, a novel RNA associated with polysomes in hamster HA-1 cells. Arch. Biochem. Biophys. 325: 256–264. Cresswell, J. L., P. Egger, C. H. Fall, C. Osmond, R. B. Fraser and D. J. Barker. 1997. Is the age of menopause determined in utero? Early Hum. Dev. 49: 143–148. Crew, M. D. 1993. Genes of the major histocompatibility complex and the evolutionary genetics of lifespan. Genetica 91: 225–238. Crews, D. E. 2003. Human Senescence: Evolutionary and Biocultural Perspectives. Cambridge University Press, Cambridge. Crimmins, E. M. 2001. Mortality and health in human life spans. Exp. Gerontol. 36: 885–897. Cristofalo, V. J., R. G. Allen, R. P. Pignolo, B. M. Martin and J. C. Beck. 1998. Relationship between donor age and the replicative life spans of human cells in culture: A re-evaluation. Proc. Natl. Acad. Sci. USA 95: 10614–10619. Cristofalo, V. J., J. Beck and R. G. Allen. 2003. An evaluation of replicative senescence in culture as a model for aging in situ. J. Gerontol. Biol. Sci. 58A: 776–779. Cristofalo, V. J. and R. J. Pignolo. 1995. Cell culture as a model. Pp. 58–82 in E. J. Masoro (ed.), Hand-

538

References

book of Physiology, Section 11: Aging. Oxford University Press, New York. Crown, W. H. 1998. Social Security and employment policy. Gerontologist 38: 132–135. Curtis, H. 1983. Biology, 4th edition. Worth Publishers, New York. Curtis, H. J. 1963. Biological mechanisms underlying the aging process. Science 141: 686–694. Curtis, H. J. and K. Miller. 1971. Chromosome aberrations in liver cells of guinea pigs. J. Gerontol. 26: 292–293. Curtsinger, J. W. 1995a. A book review of Evolution in Age-Structured Populations, 2nd edition, by Brian Charlesworth. Exp. Gerontol. 30: 663–665. Curtsinger, J. W. 1995b. Density and age-specific mortality. Genetica 96: 179–182. Curtsinger, J. W. 1995c. Density, mortality and the narrow view. Genetica 96: 187–189. Curtsinger, J. W. 1996. The old fly: Genetics of aging in Drosophila. Presented at the Gerontology Society of America meeting, November 1996. Curtsinger, J. W., H. H. Fukui, A. Khazaeli, A. Kirscher, S. D. Pletcher, D. E. L. Promislow and M. Tatar. 1995. Genetic variation and aging. Annu. Rev. Genet. 29: 553–575. Curtsinger, J. W., H. H. Fukui, A. S. Resler, K. Kelly and A. A. Khazaeli. 1998. Genetic analysis of extended life span in Drosophila melanogaster. I. RAPD screen for genetic divergence between selected and control lines. Genetica 104: 21–32. Curtsinger, J. W., H. H. Fukui, D. R. Townsend and J. W. Vaupel. 1992. Demography of genotypes: Failure of the limited life-span paradigm in Drosophila melanogaster. Science 258: 461–463. Curtsinger J. W. and A. A. Khazaeli. 2002. Lifespan, QTLs, age-specificity, and pleiotropy in Drosophila. Mech. Ageing Dev. 123: 81–93. Cutler, R. G. 1975. Evolution of human longevity and the genetic complexity governing aging rate. Proc. Natl. Acad. Sci. USA 72: 4664–4668. Cutler, R. G. 1982. Longevity is determined by specific genes: Testing the hypothesis. Pp. 25–114 in R. C. Adelman and G. S. Roth (eds.), Testing the Theories of Aging. CRC Press, Boca Raton, FL. Cutler, R. 2003. Genetic stability, dysdifferentiation, and longevity determination genes. Pp. 1146–1235 in R. G. Cutler and H. Rodriguez (eds.), Critical Reviews of Oxidative Stress and Aging: Advances in Basic Science, Diagnostics and Intervention, vol. II. World Scientific, Singapore. Cutler, R. G. and H. Rodriguez (eds.) 2003. Critical Reviews of Oxidative Stress and Aging: Advances in Basic Science, Diagnostics and Intervention, vols. I and II. World Scientific, Singapore. Cutler, R. G. 1983a. Species probes, longevity and aging. Pp. 69–144 in W. Regelson and F. M. Sinex (eds.), Intervention in the Aging Process, Part B.

Basic Research and Preclinical Screening. Alan R. Liss, New York. Cutler, R. G. 1983b. Superoxide dismutase, longevity and specific metabolic rate: A reply. Gerontology 29: 113–120. Cutler, R. G. 1984. Antioxidants, aging and longevity. Pp. 371–428 in W. A. Pryor (ed.), Free Radicals in Biology, vol. 6. Academic Press, New York. Cutler, R. G. 1985. Dysdifferentiative hypothesis of aging: A review. Pp. 307–340 in R. S. Sohal, L. S. Birnbaum and R. G. Cutler (eds.), Molecular Biology of Aging: Gene Stability and Gene Expression. Raven Press, New York. Cvejic, S., Z. Zhu, S. J. Felice, Y. Berman and X-Y. Huang. 2004. The endogenous ligand stunted of the GPCR Methuselah extends lifespan in Drosophila. Nature Cell Biology; May 9; DOI: 10.1038/ncb1133. Cypser, J. and T. E. Johnson, 2002. Multiple stressors in Caenorhabditis elegans induce stress hormesis and extended longevity. J. Gerontol. Biol. Sci. 57A: B109–B114. Cyrne, L., L. Martins, L. Fernandes and H. S. Marinho. 2003. Regulation of antioxidant enzymes gene expression in the yeast Saccharomyces cerevisiae during stationary phase. Free Radic. Biol. Med. 34: 385–393. Czermin, B. and A. Imhof. 2003. The sounds of silence—histone deacetylation meets histone methylation. Genetica 117: 159–164. da Cunha, G. L., I. B. da Cruz, P. Fiorino and A. K. de Oliveira. 1995. Paraquat resistance and starvation conditions in the selection of longevity extremes in Drosophila melanogaster populations previously selected for long and short developmental period. Dev. Genet. 17: 352–361. Dacquin, R., R. A. Davey, C. Laplace, R. Levasseur, H. A. Morris, S. R. Golding, S. Gebre-Medhin, D. L. Galson, J. D. Zajac and G. Karsenty. 2004. Amylin inhibits bone resorption while the calcitonin receptor controls bone formation in vivo. J. Cell Biol. 164: 509–514. Dahm, R. 2004. Dying to see. Sci. Am. 291 (4): 83–89. Damush, T. M., A. L. Stewart, K. M. Mills, A. C. King and P. L. Ritter. 1999. Prevalence and correlates of physician recommendations to exercise among older adults. J. Gerontol. Med. Sci. 54A: M423– M427. Danchin, A. and A. Hènaut. 1997. The map of the cell is in the chromosome. Curr. Opin. Genet. Dev. 7: 852–854. Daneva, T., E. Spinedi, R. Hadid and R. C. Gaillard. 1995. Impaired hypothalamo-pituitary-adrenal axis function in Swiss nude athymic mice. Neuroendocrinology 62: 79–86. Daniel, C. W. 1977. Cell longevity in vivo. Pp. 122– 158 in C. E. Finch and L. Hayflick (eds.), Hand-

References

book of the Biology of Aging. Van Nostrand Reinhold, New York. Davidson, E. H. 1999. A view from the genome: Spatial control of transcription in sea urchin development. Curr. Opin. Genet. Dev. 9: 530–541. Davidson, E. H., et al. 2002. A provisional regulatory gene network for specification of endomesoderm in the sea urchin embryo. Dev. Biol. 246: 162–190. Davis, R. and D. Faulds. 1996. Dexfenfluramine: An updated review of its therapeutic use in the management of obesity. Drugs 52: 696–724. Davis, R. E., S. Miller, C. Herrnstadt, S. S. Ghosh, E. Fahy, L. A. Shinobu, D. Galasko, L. J. Thal, M. F. Beal, N. Howell and W. D. Parker, Jr. 1997. Mutations in mitochondrial cytochrome c oxidase genes segregate with late-onset Alzheimer disease. Proc. Natl. Acad. Sci. USA 94: 4526–4531. Davison, A. N. 1987. Pathophysiology of aging brain. Gerontology 33: 129–135. Day, J. C. 1993. U.S. Bureau of the Census, Population Projections of the United States by Age, Sex, Race and Hispanic Origin; 1993 to 2050. Current Population Reports, PL5-1104 (middle series projections). U.S. Government Printing Office, Washington, DC. de Benedictis, G., G. Rose, G. Carrieri, M. de Luca, E. Falcone, G. Passarino, M. Bonafe, I. D. Mont, G. Baggio, S. Bertolini, D. Mari, R. Mattace and C. Francheschi. 1999. Mitochondrial DNA inherited variants are associated with successful aging and longevity in humans. FASEB J. 13: 1532– 1536. de Cabo, R., Surer-Galban, R. M. Anson, C. Gilman, M. Gorospe and M. A. Lane. 2003. An in vitro model of caloric restriction. Exp. Gerontol. 38: 631–640. de Duve, C. 1984. A Guided Tour of the Living Cell, vol. 1. Scientific American Library, New York. De Flora, S., A. Izzotti, K. Randerath, E. Randerath, H. Bartsch, J. Nair, R. Balansky, F. van Schooten, P. Degan, G. Fronza, D. Walsh and J. Lewtas. 1996. DNA adducts and chronic degenerative disease: Pathogenic relevance and implications in preventive medicine. Mutat. Res. 366: 197–238. DeFronzo, R. A., J. A. Tobin and R. Andres. 1979. Glucose clamp technique: A method for quantifying insulin in secretion and resistance. Am. J. Physiol. Endocrinol. 237(E6): E214–E223. de Grey, A. D. N. J. 1997. A proposed refinement of the mitochondrial free radical theory of aging. BioEssays 19: 161–166. de Grey, A. D. N. J. 1998. A mechanism proposed to explain the rise in oxidative stress during aging. J. Anti-Aging Med. 1: 53–66. de Grey, A. D. N. J. 1999. The Mitochondrial Free Radical Theory of Aging. R. G. Landes, Austin, Texas.

539

de Grey, A. D. N. J. 2000. Mitochondrial gene therapy: An arena for the biomedical use of inteins. Trends Biotechnol. 18: 394–399. de Grey, A. D. N. J. 2002. The reductive hotspot hypothesis of mammalian aging.: Membrane metabolishm magnifies mutant mitochondrial mischief. Eur. J. Biochem. 269: 2003–2009. de Grey, A. D. N. J. 2003a. The foreseeability of real anti-aging medicine: Focusing the debate. Exp. Gerontol. 38: 927–934. de Grey, A. D. N. J. 2003b. An engineer’s approach to the development of real anti-aging medicine. Science’s SAGE KE 2. http://sageke.sciencemag .org/cgi/content/full/sageke;2003/1/vp1. de Grey, A. D. N. J., BN Ames, J. K. Andersen, A. Bartke, J. Campisi, C. B. Heward, R. J. M. McCarter and G Stock. 2002. Time to talk SENS: Critiquing the immutability of human aging. Annals N.Y. Acad. Sci. 959: 452–465. de Haan, G. and G. Van Zant. 1999. Genetic analysis of hemopoietic cell cycling in mice suggests its involvement in organismal life span. FASEB J. 13: 707–713. de Haan, G. and G. Van Zant. 2002. Stem cells from birth to death: The history and the future. J. Am. Aging Assoc. 25: 79–88. de Magalhaes, J. P. and O. Toussaint. 2004. GenAge: A genomic and proteomic network map of human ageing. FEBS Lett. 571: 243–247. Deevey, E. S. Jr. 1947. Life tables for natural populations of animals. Q. Rev. Biol. 22: 283–314. Demple, B. and L. Harrison. 1994. Repair of oxidative damage to DNA: Enzymology and biology. Annu. Rev. Biochem. 63: 915–948. Dennett. D. 2002. The new replicators. Pp. E83–E92 in M. Pagel (ed.), Encyclopedia of Evolution, vol. 1. Oxford University Press, New York. Derry, W. B., A. P. Putzke and J. H. Rothman. 2001. Caenorhabditis elegans p53: role in apoptosis, meiosis, and stress resistance. Science 294: 591–595. Dewji, N. N. and S. J. Singer. 1996. Genetic clues to Alzheimer’s disease. Science 271: 159–160. Dhahbi, J. M., H-J. Kim, P. L. Mote, R. J. Beaver and S. R. Spindler. 2004. Temporal linkage between the phenotypic and genomic responses to caloric restriction. Proc. Natl. Acad. Sci. 101: 5524– 5529. Dhahbi, J. M. and S. R. Spindler. 2003. Aging of the liver. Pp. 271–291 in R. Aspinall (ed.), Ageing of Organs and Systems. Kluwer Academic Publishers, Dordrecht, the Netherlands. Diamond, J. 2003. The double puzzle of diabetes. Nature 423: 599–602. Dimri, G. P., X. Lee, G. Basile, M. Acosta, G. Scott, C. Roskelley, E. E. Medrano, M. Linskens, I. Rubeli, O. Pereira-Smith, M. Peacocke and J. Campisi. 1995. A biomarker that identifies

540

References

senescent human cells in culture and in aging skin in vivo. Proc. Natl. Acad. Sci. USA 92: 9363–9367. DiPietro, L. 2001. Physical activity in aging: Changes in patterns and their relationship to health and function. J. Gerontol. A 56A (special issue II): 13–22. D’Mello, N. P., A. M. Childress, D. S. Franklin, S. P. Kale, C. Pinswadi and S. M. Jazwinski. 1994. Cloning and characterization of LAG1, a longevity assurance gene in yeast. J. Biol. Chem. 269: 15451–15459. Dobzhansky, T. 1973. Nothing in biology makes sense except in the light of evolution. Amer. Biol. Teacher 35: 125–129. Docherty, J. R. 1996. Effects of aging on prejunctional control of neurotransmission in the rat. Ann. N.Y. Acad. Sci. 786: 264–273. Dolle, M. E. T., H. Giese, C. L. Hopkins, H.-J. Martus, J. M. Hausdorff and J. Vijg. 1997. Rapid accumulation of genome rearrangements in liver but not brain of old mice. Nature Genet. 17: 431–434. Dominici, F. P., D. P. Argentino, A. Bartke and D. Turyn. 2003. The dwarf mutation decreases high dose insulin responses in skeletal muscle, the opposite of effects in liver. Mech. Aging Dev. 124: 819–827. Donis-Keller, H. et al. 1987. A genetic linkage map of the human genome. Cell 51: 319–337. Dorman, J. B., B. Albinder, T. Shoyer and C. Kenyon. 1995. The age-1 and daf-2 genes function in a common pathway to control the lifespan of Caenorhabditis elegans. Genetics 141: 1399–1406. Doty, R. L., P. Shaman, S. C. Applebaum, R. Giberson, L. Sikosorski and L. Rosenberg. 1984. Smell identification ability: Changes with age. Science 226: 1441–1443. Doyle, F. H. 1969. Radiological measurements of skin thickness and bone mineral. Sci. Basis Med. Annu. Rev. 8: 133–145. Draye, X. and F. A. Lints. 1995. Geographic variations of life history strategies in Drosophila melanogaster. II. Analysis of laboratory-adapted populations. Exp. Gerontol. 30: 517–532. Drewnowski, A. and W. J. Evans. 2001. Nutrition, physical activity, and quality of life in older adults: Summary. J. Gerontol. 56A (special issue II): 89–94. Driscoll, M. 1995. Genes controlling programmed cell death: Relation to mechanisms of cell senescence and aging? Pp. 45–60 in K. Esser and G. M. Martin (eds.), Molecular Aspects of Aging. Wiley, Chichester, England. Driver, C. and N. Tawadros. 2000. Cytoplasmic genomes that confer additional longevity in Drosophila melanogaster. Biogerontology 1: 255–260. Drori, D. and Y. Folman. 1986. Interactive environmental and genetic effects on longevity in the male

rat. Litter size, exercise, castration and electronic shock. Exp. Aging Res. 12: 59–64. Duara, R., E. D. London and S. I. Rapoport. 1985. Changes in structure and energy metabolism of the aging brain. Pp. 595–616 in C. E. Finch and E. L. Schneider (eds.), Handbook of the Biology of Aging, 2nd ed. Van Nostrand Reinhold, New York. Ducy, P., T. Schinke and G. Karsenty. 2000. The osteoblast: a sophisticated fibroblast under central surveillance. Science 289: 1501–1504. Dudas, S. P. and R. Arking. 1994. The expression of the EF1 genes of Drosophila is not associated with the extended longevity phenotype in a selected long lived strain. Exp. Gerontol. 29: 645–657. Dudas, S. P. and R. Arking. 1995. A coordinate upregulation of the antioxidant gene activities is associated with the delayed onset of senescence in a long lived strain of Drosophila. J. Gerontol. Biol. Sci. 50A: B117–B127. Duffy, P. H., R. J. Feuers, J. A. Leakey, A. Turturro and R. W. Hart. 1989. Effect of chronic calories restriction on physiological variables related to energy metabolism in the male Fischer 344 rat. Mech. Ageing Dev. 48: 117–133. Duhon, S. A. and T. E. Johnson. 1993. Detection of new long-lived mutants in Caenorhabditis elegans. Gerontologist 33: 96 (abstract). Dukan, S., A. Farewell, M. Ballesteros, F. Taddei, M. Radman and T. Nystrom. 2000. Protein oxidation in response to increased transcriptional or translational errors. Proc. Natl. Acad. Sci. USA 97: 5746–5749. Dusenbery, R. L. and P. D. Smith. 1996. Cellular responses to DNA damage in Drosophila melanogaster. Mutat. Res. 364: 133–145. Duttaroy, A., T. Parkes, P. Emtage, K. Kirby, G. L. Boulianne, X. Wang, A. J. Hilliker and J. P. Phillips. 1977. The manganese superoxide dismutase gene of Drosophila: Structure, expression, and evidence for regulation by MAP kinase. DNA Cell Biol. 16(4): 391–399. Dworkin, L. D., T. H. Hostetter, H. G. Rennke and B. M. Brenner. 1984. Hemodynamic basis for glomerular injury in rats with desoxycorticosterone-salt hypertension. J. Clin. Invest. 73: 1448– 1461. Dykes, C. W. 1996. Genes, disease and medicine. Br. J. Clin. Pharmacol. 42: 683–695. Dykhuisen, D. 1974. The evolution of cell senescence, atherosclerosis and benign tumours. Exp. Cell Res. 144: 455–562. Eakin, T. and M. Witten. 1995a. A gerontological distance metric for analysis of survival dynamics. Mech. Ageing Dev. 78: 85–101. Eakin, T. and M. Witten. 1995b. How square is the survival curve of a given species. Exp. Gerontol. 30: 33–64.

References

Eakin, T., R. Shouman, Y. Qi, G. Liu and M. Witten. 1995. Estimating parametric survival model parameters in gerontological aging studies: methodological problems and insights. J. Gerontol.: Biol. Sci. 50: B166–B176. Earle, W. R. 1943. Production of malignancy in vitro. IV. The mouse fibroblast cultures and changes seen in living cells. J. Natl. Cancer Inst. 4: 165–212. Ebbesen, P. 1973. Papilloma induction in different aged skin grafts to young recipients. Nature 241: 280–281. Ebbesen, P. 1974. Aging increases susceptibility of mouse skin to PMBA carcinogenesis independent of general immune status. Science 183: 217–218. Ebeling, A. H. 1913. The permanent life of connective tissue outside of the organism. J. Exp. Med. 17: 273–285. Eberling, J. L., T. E. Nordahl, N. Kusubov, B. R. Reed, T. F. Budinger and W. J. Jagust. 1995. Reduced temporal lobe glucose metabolism in aging. J. Neuroimaging 5: 178–182. Ebert, R. H., V. A. Cherkasova, R. A. Dennis, J. H. Wu and S. Ruggles. 1993. Longevity determining genes in Caenorhabditis elegans: Chromosomal mapping of multiple noninteractive loci. Genetics 135: 1003–1010. Echtay, K. S., M. P. Murphy, R. A. Smith, D. A. Talbot and M. D. Brand. 2002. Superoxide activates mitochondrial uncoupling protein 2 from the matrix side. Studies using targeted antioxidants. J. Biol. Chem. 277: 47129–47135. Edelman, G. M. 1988. Topobiology: An Introduction to Molecular Embryology. Basic Books, New York. Edelson, R. L. and E. M. Fink. 1985. The immunologic function of skin. Sci. Am. 252(6): 46–53. Edgar, B. A. and C. F. Lehner. 1996. Developmental control of cell cycle regulators: A fly’s perspective. Science 274: 1646–1652. Edington, D. W., A. C. Cosmos and W. B. McCafferty. 1972. Exercise and longevity: Evidence for a threshold age. J. Gerontol. 27: 341–343. Edwards, M. G., D. Sarkar, R. Klopp, J. D. Morrow, R. Weindruch and T. A. Prolla. 2003. Age-related impairment of the transcriptional responses to oxidative stress in the mouse heart. Physiol. Genomics, February 25. 10.1152/physiolgenomics .00172.2002. Effros, R. B. 1996. Insights on immunological aging derived from the T lymphocyte cellular senescence model. Exp. Gerontol. 31: 21–27. Egan, K. M., J. A. Lawson, S. Fries, B. Koller, D. J. Rader, E. M. Smyth and G. A. FitzGerald. 2004. COX-2-derived prostacyclin confers atheroprotection on female mice. Science 306: 1954– 1957. Egilmez, N. K., J. B. Chen and S. M. Jazwinski. 1989. Specific alterations in transcript prevalence dur-

541

ing the yeast life span. J. Biol. Chem. 264: 14312– 14317. Egilmez, N. K. and M. Jazwinski. 1989. Specific alterations in transcript prevalence during the yeast life span. J. Biol. Chem. 264: 14312–14317. Egilmez, N. K. and M. Rothstein. 1985. The effect of aging on cell-free protein synthesis in the freeliving nematode Trubatrix aceti. Biochim. Biophys. Acta 840: 355–363. Ehrenstein, D. 1998. Immortality gene discovered. Science 279: 177. Eiseley, L. 1977. Somewhere beyond the pawnshops. Pp. 73–74 in Another Kind of Autumn. Scribner, New York. Elnum, S. and I. A. Fleming. 2000. Highly fecund mothers sacrifice offspring survival to maximize fitness. Nature xxx:xxx–xxx. Engelberg, D., C. Klein, H. Martinetto, K. Struhl and M. Karin. 1994. The UV response involving the Ras signaling pathway and AP-1 transcription factors is conserved between yeast and mammals. Cell 77: 381–390. Epel, E. S., E. H. Blackburn, J. Lin, F. S. Dhabhar, N. E. Adler, J. D. Morrow and R. M. Cawthorn. 2004. Accelerated telomere shortening in response to life stress. Proc. Natl. Acad. Sci. USA 101: 17312– 17315. Esposito, D., G. Fassina, P. Szabo, P. DeAngelis, L. Rodgers, M. Weksler and M. Siniscalco. 1989. Chromosomes of older humans are more prone to aminopterine-induced breakage. Proc. Natl. Acad. Sci. USA 86: 1302–1306. Esposito, J. L. 1987. The Obsolete Self: Philosophical Dimensions of Aging. University of California Press, Berkeley. Esser, K. 1985. Genetic control of aging: The mobile intron model. Pp. 3–20 in M. Bergener et al. (eds.), Thresholds in Aging. Academic Press, London. Esser, K. and W. Keller. 1976. Genes inhibiting senescence in the ascomycete Podospora anserina. Mol. General Genet. 144: 107–110. Etter, P. D. and M. Ramaswami. 2002. The ups and downs of daily life: Profiling circadian gene expression in Drosophila. BioEssays 24: 494–498. Evans, D. 1995. Human studies. Pp. 83–92 in E. J. Masoro (ed.), Handbook of Physiology, Section 11: Aging. Oxford University Press, New York. Evans, D. A. 1996. Descriptive epidemiology of Alzheimer’s disease. Pp. 51–60 in Z. S. Katchaturian and T. Radebaugh (eds.), Alzheimer’s Disease: Cause(s), Diagnosis, Treatment, and Care. CRC Press, Boca Raton, FL. Evans, D. A., A. A. van der Kleij, M. A. Sonnermans, J. P Burbach and F. W. van Leeuwen. 1994. Frameshift mutations at two hotspots in vasopressin transcripts in post-mitotic neurons. Proc. Natl. Acad. Sci. USA 91: 6059–6063.

542

References

Evans, W. F. 1983. Anatomy and Physiology, 3rd ed. Prentice-Hall, Englewood Cliffs, NJ. Evans, W. J. 1995. Effects of exercise on body composition and functional capacity of the elderly. J. Gerontol. Biol. Sci. 50A(special issue): B147–B150. Everitt, A. V., C. D. Shorey and M. A. Ficarra. 1985. Skeletal muscle aging in the hind limb of the old male Wistar rat: Inhibitory effect of hypophysectomy and food restriction. Arch. Gerontol. Geriatr. 4: 101–115. Evert, J., E. Lawler, H. Bogan and T. Perls. 2003. Morbidity profiles of centenarians: Survivors, delayers, and escapers. J. Gerontol. Med. Sci. 58A: 232–237. Ewbank, J. J., T. M. Barnes, B. Lakowski, M. Lussier, H. Bussey and S. Hekimi. 1997. Structural and functional conservation of the Caenorhabditis elegans timing gene clk-1. Science 275: 980–983. Exton-Smith, A. N. 1985. Mineral metabolism. Pp. 511–539 in C. E. Finch and E. L. Schneider (eds.), Handbook of the Biology of Aging, 2nd ed. Van Nostrand Reinhold, New York. Ezzell, C. 1995. Fat times for obesity research: Tons of new information but how does it all fit together? J. NIH Res. 7: 40–43. Fabian, T. J. and T. E. Johnson. 1995. Identification of genes that are differentially expressed during aging in Caenorhabditis elegans. J. Gerontol. Biol. Sci. 50A: B245–B253. Fabris, N. and E. Mocchegiani. 1985. Endocrine control of thymic serum factor production in youngadult and old mice. Cell. Immunol. 91: 325. Fabris, N. and E. Mocchegiani. 1996. Zinc, human diseases and aging. Aging 7: 77–93. Fabris, N., E. Mocchegiani and M. Provinciali. 1995. Pituitary-thyroid axis and immune system: A reciprocal neuroendocrine-immune interaction. Hormone Res. 43: 29–38. Fabrizio, P., L-L. Kiou, V. N. Moy, A. Diaspro, J. S. Valentine, E. B. Gralla and V. D. Longo. 2003. SOD2 functions downstream of Sch9 to extend longevity in yeast. Genetics 163: 35–46. Fabrizio, P. and V. D. Longo. 2003. The chronological life span of Saccharomyces cerevisiae. Aging Cell 2: 73–82. Fabrizio, P., P. Pozza, S. D. Pletcher, C. M. Gendron and V. D. Longo. 2001. Regulation of longevity and stress resistance by Sch9 in yeast. Science 292: 288–290. Familli, I., J. Forster, J. Nielsen and B. O. Palsson. 2003. Saccharomyces cerevisiae phenotypes can be predicted by using constraint-based analysis of a genome-scale reconstructed metabolic network. Proc. Natl. Acad. Sci. USA 100: 13134–13139. Farrell, P. M. and J. G. Bieri. 1975. Megavitamin supplementation in man. Am. J. Clin. Nutr. 28: 1381–1386.

Featherstone, D. E. and K. Broadie. 2002. Wrestling with pleiotropy: Genomic and topological analysis of the yeast gene expression network. BioEssays 24: 267–274. Feinleib, M., R. J. Garrison, R. Fabsitx, J. C. Christian, X. S. Hrubec, N. O. Borhani, W. B. Kannel, R. Rosenman, J. T. Schwartz and J. O. Wagner. 1977. The NHLBI twin study of cardiovascular disease risk factorsL methodology and summary of results. Am. J. Epidemiol. 106: 284–285. Feldman, M. L. 1976. Aging changes in the morphology of cortical dedrites. Pp. 211–246 in R. D. Terry and S. T. Gershon (eds.), Neurobiology of Aging. Raven Press, New York. Felkai, S., J. J. Ewbank, J. Lemieux, J. C. Labbe, G. G. Brown and S. Hekimi. 1999. CLK-1 controls respiration behavior and aging in the nematode Caenorhabditis elegans. EMBO J. 18: 1783–1792. Felty, Q. and D. Roy. 2005. Mitochondrial signals to nucleus regulate estrogen-induced cell growth. Med. Hypotheses 64: 133–141. Fernandes, G. and J. T. Venkatraman. 1994. Effect of food restriction on immunoregulation and aging. Pp. 331–348 in R. Watson (ed.), Handbook of Nutrition in the Aged, 2nd ed. CRC Press, Boca Raton, FL. Ferrucci, L., C. Cavazzini, A. Corsi, B. Bartali, C. R. Russo, F. Lauretani, S. Bandinelli and J. M. Guralnick. 2002. Biomarkers of frailty in older persons. J. Endocrinol. Invest 25 (10 Suppl): 10–15. Festing, M. F. W. 1999. Warning: The use of heterogeneous mice may seriously damage your research. Neurobiol. of Aging 20: 237–244. Feuers, R. J., D. A. Casciano, J. G. Shaddock, J. A. Leakey, P. H. Duffy, R. W. Hart, J. D. Hunter and L. E. Schering. 1991. Modification in regulation of intermediary metabolism by caloric restriction in rodents. Pp. 198–206 in L. Fishbein (ed.), Biological Effects of Dietary Restriction. SpringerVerlag, New York. Feuers, R. J., P. H. Duffy, F. Chen, V. Desai, E. Oriaku, J. G. Shaddock, J. W. Pipkin, R. Weindruch and R. W. Hart. 1995. Intermediary metabolism and antioxidant systems. Pp. 180–195 in R. W. Hart, D. A. Neumann and R. T. Robertson (eds.), Dietary Restriction: Implications for the Design and Interpretation of Toxicity and Carcinogencity Studies. ILSI Press, Washington, DC. Fiatarone, M. A., E. F. O’Neill, N. Doyle, K. M. Clements, S. B. Roberts, J. J. Kehayias, L. A. Lipsitz and W. J. Evans. 1993. The Boston FICSIT study: The effects of resistance training and nutritional supplementation on physical frailty in the oldest old. J. Am. Geriatr. Soc. 41: 333–337. Finch, C. E. 1987. Neural and endocrine determinants of senescence: Investigation of causality and reversibility by laboratory and clinical interven-

References

tions. Pp. 261–308 in H. R. Warner, R. N. Butler, R. L. Sprott and E. L. Schneider (eds.), Aging, vol. 31, Modern Biological Theories of Aging. Raven Press, New York. Finch, C. E. 1988. Neural and endocrine approaches to the resolution of time as a dependent variable in the aging processes of mammals. Gerontologist 28: 29–42. Finch, C. E. 1990. Longevity, Senescence, and the Genome. University of Chicago Press, Chicago. Finch, C. E. and E. M. Crimmins. 2004. Inflammatory exposure and historical changes in human lifespans. Science 305: 1736–1739. Finch, C. E., L. S. Felicio, C. V. Mobbs and J. F. Nelson. 1984. Ovarian and steroidal influences on neuroendocrine aging processes in female rodents. Endocrine Rev. 5: 467–497. Finch, C. E. and R. G. Gosden. 1986. Animal models for the human menopause. Pp. 3–34 in L. Mastroianni Jr. and C. A. Paulsen (eds.), Aging, Reproduction and the Climacteric. Plenum Press, New York. Finch, C. E. and T. B. L. Kirkwood. 2000. Chance, Development, and Aging. Oxford University Press, New York. Finch, C. E. and P. W. Landfield. 1985. Neuroendocrine and autonomic function in aging mammals. Pp. 567–594 in C. E. Finch and E. L. Schneider (eds.), Handbook of the Biology of Aging, 2nd ed. Van Nostrand Reinhold, New York. Finch, C. E. and M. C. Pike. 1995. Maximum life span predictions from the Gompertz mortality model. J. Gerontol. Biol. Sci. 51A: B183–B194. Finch, C. E., M. C. Pike and M. Witten. 1990. Slow mortality rate accelerations during aging in some animals approximate that of humans. Science 249: 902–905. Finch, C. E. and R. Tanzi. 1997. Genetics of aging. Science 278: 407–411. Fink, L. and F. S. Collins. 1997. The Human Genome Project: View from the National Institutes of Health. J. Am. Med. Womens Assoc. 52: 4–7, 15. Finkel, D., N. L. Pedersen, M. McGue and G. E. McClearn. 1995. Heritability of cognitive abilities in adult twins: Comparison of Minnesota and Swedish data. Behavior Genet. 25: 421–431. Finkel, T. and N. J. Holbrook. 2000. Oxidants, oxidative stress, and the biology of ageing. Nature 408: 239–247. Fischer, W., A. Bjorklund, K. Chen and F. Gage. 1991. NGF improves spatial memory in aged rodents as a function of age. J. Neurosci. 11: 1889–1906. Fisher, A. and J. E. Morley, 2002. Antiaging medicine: the good, the bad, and the ugly. J. Gerontol. Med Sci. 57A:M636–M639. Fisher, A. and J. E. Morley. 2002. Antiaging medicine:

543

the good, the bad, and the ugly. J. Gerontol.: Med. Sci. 57: M636–639. Fisher, R. A. 1930. The Genetical Theory of Natural Selection. Clarendon, Oxford. Fleming, J. E., G. Spicer, R. C. Garrison and M. R. Rose. 1993. Two-dimensional protein electrophoretic analysis of postponed aging in Drosophila. Genetica 91: 183–193. Fleming, J. E., J. K. Walton, R. Dubitsky and K. G. Bensch. 1988. Aging results in an unusual expression of Drosophila heat shock proteins. Proc. Natl. Acad. Sci. USA 85: 4099–4103. Flier, J. S. 2004. Obesity wars: Molecular progress confronts an expanding epidemic. Cell 116: 337– 350. Fliert, A., C. Jacson, B. Cottrell, D. Murdock, P. Seibert and D. C. Wallace. 2003. Targeted delivery of DNA to the mitochondrial compartment via import sequence-conjugated peptide nucleic acid. Mol. Ther. 7: 550–557. Fonager, J., R., Beedholm, B. F. C.Clark and S. I. S. Rattan. 2002. Mild stress-induced stimulation of heat-shock protein synthesis and improved functional ability of human fibroblasts undergoing aging in vitro. Exp. Gerontol. 37: 1223–1228. Fontaine, K. R., D. T. Redden, C. Wang, A. O. Westfall and D. B. Allison. 2003. Years of life lost due to obesity. JAMA 289: 187–193. Fontana, L., T. E. Meyer, S. Klein and J. O. Holloszy. 2004. Long-term calorie restriction is highly effective in reducting the risk for atherosclerosis in humans. Proc. Natl. Acad. Sci. USA 101: 6659–6663. Force, A. G., T. Staples, S. Soliman and R. Arking. 1995. A comparative biochemical and stress analysis of genetically selected Drosophila strains with different longevities. Dev. Genet. 17: 340–351. Foreman, K. E. and J. Tang. 2003. Molecular mechanisms of replicative senescence in endothelial cells. Exp. Gerontol. 38: 1251–1257. Forsyth, N., F. F. B. H. Elder, J. W. Shay and W. E Wright. 2005. Lagomorphs (rabbits, pikas and hares) do not use telomere-directed replicative aging in vitro. Mech. Aging Develop. 126: 685– 691. Fossel, M. 1998. Implications of recent work in telomeres and cell senescence. J. Anti-Aging Med. 1: 39– 43. Fossel, M. 2003a. The ethics of longevity: should we extend the healthy maximum lifespan? Pp. xx–xx in R. G. Cutler and H. Rodriguez (eds.), Critical Reviews of Oxidative Stress and Aging: Advances in Basic Science, Diagnostics, and Intervention, vol. 2. World Scientific Publishing, Singapore. Fossel, M. 2003b. What ethics are these? J. AntiAging Medicine 6: 71. Fossel, M. 2004. Cells, Aging, and Human Disease. Oxford University Press, New York.

544

References

Foster, G. D., et al. 2003. A randomized trial of a lowcarbohydrate diet for obesity. N. Engl. J. Med. 348: 2082–2090. Fozard, J. L., E. J. Metter and L. J. Brant. 1990. Next steps in describing aging and disease in longitudinal studies. J. Gerontol. 45: P116–P127. Fraidenraich, D., E. Stillwell, E. Romero, D. Wilkes, K. Manova, C. T. Basson and R. Benezra. 2004. Rescue of cardiac defects in Il knockout embryos by injection of embryonic stem cells. Science 306: 247–252. Franceschi, C., L. Motta, S. Valensin, R. Rapisarda, A. Franzone, M. Berardelli, et al. 2000. Do men and women follow different trajectories to reach extreme longevity? Italian Multicenter Study on Centenarians. Aging 12: 77–84. Frank, J. 2005. Beyond vitamin E supplementation: an alternative strategy to improve vitamin E status. J. Plant. Physiol. 162: 834–843. Frank, L. 1985. Oxygen toxicity in eukaryotes. Pp. 1– 44 in L. W. Oberly (ed.), Superoxide Dismutase, vol. 3. CRC Press, Boca Raton, FL. Fraser, G. E. and D. J. Shavlik. 2001. Ten years of life: Is it a matter of choice? Arch. Intern. Med. 161: 1645–1652. Freeman, G. and J. W. Lundelius. 1992. Evolutionary implications of the mode of D quadrant specification in coelomates with spiral cleavage. J. Evol. Biol. 5: 205–247. Friddle, C. J., T. Koga, E. M. Rubin and J Bristow. 2000. Expression profiling reveals distinct sets of genes altered during induction and regression of cardiac hypertrophy. Proc. Natl. Acad. Sci. USA 97: 6745–6750. Fried, L. P., C. M. Tangen, J. Walston, A. B. Newman, C. Hirsch, J. Gottdiener, et al. 2000. Fraility in older adults: Evidence for a phenotype. J. Gerontol. Med. Sci. 56A: M146–M156. Friedman, D. B. and T. E. Johnson. 1988a. A mutation in the age-1 gene in Caenorhabditis elegans lengthens life and reduces hermaphrodite fertility. Genetics 118: 75–86. Friedman, D. B. and T. E. Johnson. 1988b. Three mutants that extend both mean and maximum life span of the nematode, Caenorhabditis elegans, define the age-1 gene. J. Gerontol.: Biol. Sci. 43: B102–B109. Fries, J. F. 2003. Measuring and monitoring success in compressing morbidity. Ann. Int. Med. 139: 455– 459. Fries, J. F. 1980. Aging, natural death, and the compression of mortality. N. Engl. J. Med. 303: 130– 135. Fries, J. F. and L. F. Crapo. 1981. Vitality and Aging: Implications of the Rectangular Curve. Freeman, San Francisco. Frolkis, V. V. 1982. Aging and Life-Prolonging

Processes (Trans. N. Bobrov). Springer-Verlag, Vienna. Fryer, J. H. 1962. Studies of body composition on men aged 60 and over. Pp. 59–78 in N. W. Shock (ed.), Biological Aspects of Aging. Columbia University Press, New York. Frymoyer, J. W. 1986. Musculoskeletal disabilities. Pp. 273–300 in S. J. Brody and G. E. Ruff (eds.), Aging and Rehabilitation: Advances in the State of the Art. Springer, New York. Fujibayashi, Y., A. Waki, K. Wada, M. Ueno, Y. Magata, Y. Yonekura, J. Konishi, T. Takeda and A. Yokoyama. 1994. Differential aging pattern of cerebral accumulation of radiolabeled glucose and amino acid in the senescence accelerated mouse (SAM), a new model for the study of memory impairment. Biol. Pharmacol. Bull. 17: 102–105. Fukui, H. H., L. Xiu and J. W. Curtsinger. 1993. Slowing of age-specific mortality rates in Drosophila melanogaster. Exp. Gerontol. 28: 585–599. Fulop, T. Jr. and I. Seres. 1994. Age-related changes in signal transduction: Implications for neuronal transmission and potential for drug intervention. Drugs Aging 5: 366–390. Funk, S. D., C. K. Wang, D. N. Shelton, C. B. Harley, G. D. Pagon and W. K. Hoeffler. 2000. Telomerase expression restores dermal integrity to in-vitro aged fibroblasts in a reconstituted skin model. Exp. Cell Res. 258: 270–278. Furukawa, F., S. Ikehara, R. A. Good, T. Nakamura, S. Inoue, H. Tanaka, S. Imamura and Y. Hamashinma. 1988. Immunological status of nude mice engrafted with allogeneic or syngeneic thymuses. Thymus 12: 11–26. Furukawa, T. 1994. Assessment of the adequacy of the multiregression method to estimate biological age. Pp. 471–484 in A. K. Balin (ed.), Practical Handbook of Human Biologic Age Determination. CRC Press, Boca Raton, FL. Gadaleta, M. N., G. Rainaldi, A. M. Lezza, F. Milella, F. Fracasso and P. Cantore. 1992. Mitochondrial DNA copy number and mitochondrial DNA deletion in adult and senescent rats. Mutat. Res. 275: 181–193. Gafni, A. and L. Cook. 1988. Protection of rat muscle phosphoglycerate kinase from aging by specific methylation. Gerontologist 28: 229A (abstract). Gaillard, J.-M., D. Allaine, D. Pontier, N. G. Yoccoz and D. E. L. Promislow. 1994. Senescence in natural populations of mammals: A reanalysis. Evolution 48: 509–516. Galli, R. L., B. Shukitt-Hale, K. A. Youdim and J. A. Joseph. 2002. Fruit polyphenolics and brain aging: nutritional interventions targeting age-related neuronal and behavioral deficits. Ann. N.Y. Acad. Sci. 959: 128–133. Gambert, S. R. and D. A. Kassur. 1994. Protein calo-

References

rie malnutrition in the elderly. Pp. 295–302 in R. Watson (ed.), Handbook of Nutrition in the Aged, 2nd ed. CRC Press, Boca Raton, FL. Ganetzky, B. and J. R. Flanagan. 1978. On the relationship between senescence and age-related changes in two wild-type strains of Drosophila melanogaster. Exp. Gerontol. 13: 189–196. Garen, S. 1975. Bone loss and aging. Pp. 39–58 in R. Goldman (ed.), The Physiology and Pathology of Human Aging. Academic Press, New York. Garsin, D. A., J. M. Villanueva, J. Begun, D. H. Kim, C. D. Sifri, S. B. Calderwood, G. Ruvkun and F. M. Ausubel. 2003. Long-lived C. elegans daf-2 mutants are resistant to bacterial pathogens. Science 300: 1921. Gavrilov, L. and N. Gavrilova. 1991. The Biology of Life Span: A Quantitative Approach, V. P. Skulacher (ed.). Revised and updated English ed. (trans. J. and L. Payne). Harwood Academic Publishers, Chur, Switzerland. Gavrilov, L. and N. Gavrilova. 1997. Parental age at conception and offspring longevity. Rev. Clin. Geron. 7: 5–12. Gavrilov, L. A. and N. S. Gavrilova, 1999. Is there a reproductive cost for human longevity? J. AntiAging Med. 2: 121–123. Gavrilov, L. A. and N. S. Gavrilova, 2001. Biodemographic study of familial determinants of human longevity. Population—An English Selection 13: 197–222. Gavrilov, L. A. and N. S. Gavrilova, 2002. Evolutionary theories of aging and longevity. Sci. World J. 2: 339–356. Gavrilov, L. A. and N. S. Gavrilova, 2003. Early-life factors modulating lifespan. Pp. 27–50 in S. I. S. Rattan (ed.), Modulating Aging and Longevity. Kluwer Academic Publishers, Dordrecht, the Netherlands. Gavrilov, L. A. and N. S. Gavrilova. 2004. The reliability engineering approach to the problem of biological aging. Ann. N. Y. Acad. Sci. 1019: 509– 512. Gavrilova, N. S., L. A. Gavrilov, V. G. Semyonova and G. N. Evdokushkina. 2004. Does exceptional human longevity come with a high cost of infertility? Testing the evolutionary theories of aging. Ann. N.Y. Acad. Sci. 1019: 513–517. Gearing, M., S. S. Mirra, J. C. Hedreen, S. M. Sumi, L. A. Hansen and A. Heyman. 1995. The Consortium to Establish a Registry for Alzheimer’s Disease (CERAD). Part X. Neuropathology confirmation of the clinical diagnosis of Alzheimer’s disease. Neurology 45: 461–466. Gee, E. M. and J. E. Veevers. 1984. Accelerating sex differentials in mortality: An analysis of contributing factors. Social Biol. 30: 75–85. Gekakis, N., L. Saez, A. M. Delayaye-Brown, M. P.

545

Myers, A. Sehgal, M. W. Young and C. J. Weitz. 1995. Isolation of timeless by PER protein interaction: Defective interaction between timeless protein and long-period mutant PERL. Science 270: 811–815. Gelman, R., A. Watson, R. Bronson and E. Yunis. 1988. Murine chromosomal regions correlated with longevity. Genetics 118: 693–704. Gensler, H. L. and H. Bernstein. 1981. DNA damage as the primary cause of aging. Q. Rev. Biol. 56: 279–303. Gerhard, GS. 2001. Caloric restriction in nonmammalian models. J. Anti-Aging Med. 4: 205–213. Gerisch, B. and A. Antebi. 2004. Hormonal signals produced by DAF-9/cytochrome P450 regulate C. elegans dauer diapause in response to environmental signals. Development 131: 1765– 1776. Gerisch, B., C. Weitzel, C. Kober-Eisermann, V. Rottiers and A. Antebi. 2001. A hormonal signaling pathway influencing C. elegans metabolism, reproductive development, and life span. Dev. Cell 1: 841–851. Gershon, D. 1979. Current status of age altered enzymes: Alternative mechanisms. Mech. Ageing Dev. 9: 189–196. Gershon, H. and D. Gershon. 1970. Detection of inactive enzyme molecules in ageing organisms. Nature 227: 1214–1217. Gershon, H. and D. Gershon. 2000. The budding yeast, Saccharomyces cerevisiae, as a model for aging research: A critical review. Mech. Ageing Dev. 120: 1–22. Gerson, D. 1999. The mitochondrial theory of aging: Is the culprit a faulty disposal system rather than indigenous mitochondrial alterations. Exp. Gerontol. 34: 613–619. Gey, G. O., W. D. Coffman and M. Kubicel. 1952. Tissue culture studies of the proliferative capacity of the cervical carcinoma and normal epithelium. Cancer Res. 12: 264–265. Gey, G. O., M. Svotelis, M. Foard and F. B. Bang. 1974. Long-term growth of chicken fibroblasts on a collagen substrate. Exp. Cell Res. 84: 63–71. Ghirardi, O., R. Coxxolino, D. Guaraldi and A. Giuliani. 1995. Within- and between-strain variability in longevity of inbred and outbred rats under the same environmental conditions. Exp. Gerontol. 30: 485–494. Giblin, G. M., P. C. Box, I. B. Campbell, A. P. Hancock, S. Roomans, G. I. Mills, C. Molloy, G. E. Tranter, A. L. Walker, S. R. Doctrow, K. Huffman and B. Malfroy 2001. 6,6'-Bis(2hydroxyphenyl)-2,2'-bipyridine manganese(III) complexes: a novel series of superoxide dismutase and catalase mimetics. Bioorg. Med Chem. Lett. 11: 1367–1370.

546

References

Gilbert, S. 2000. Developmental Biology, 6th ed. Sinauer Associates, Sunderland MA. Gilbert, S. F. 1997. Developmental Biology, 5th ed. Sinauer Associates, Sunderland, MA. Giot, L., J. S. Bader, C. Brouwer, A. Chaudhuri, B. Kaung, Y. Li, et al. 2003. A protein interaction map of Drosophila melanogaster. Science 302: 1727–1736. Giradot, F., V. Monnier and H. Tricoire. 2004. Genome wide analysis of common and specific stress responses in adult Drosophila melanogaster. BMC Genomics 5: 74 DOI: 10.1186/1471-21645-74. Giro, M. and J. M. Davidson. 1993. Familial co-segregation of the elastin phenotype in skin fibroblasts from Hutchinson-Gilford progeria. Mech. Ageing Dev. 70: 163–176. Giroux, C. N., A. Weiss, J. DelProposto and S. Stapels. 2003. A dynamic genetic network mediates dosedependent oxidative stress responses. Toxicologist 72(suppl. 1): 92 (abstract). Giubilei, F., R. D’Antona, R. Antonini, G. L. Lenzi, G. Ricci and C. Fieschi. 1990. Serum lipoprotein pattern variations in dementia and ischemic stroke. Acta Neurol. Scand. 81: 84–86. Gjonca, A., C. Tomassini and J. W. Vaupel. 1999. Max Planck-Institut for demografische Forschung Working paper 1999-009, July. Glaser, V. 2003. Pieces of the puzzle: An interview with Michael Fossel, MD, PhD. J. Anti-Aging Med. 6: 293–297. Gleeson, M., A. W. Cripps and R. L. Clancy. 1995. Modifiers of the human mucosal immune system. Immunol. Cell Biol. 73: 397–404. Glenner, G. G. 1988. Alzheimer’s disease: its proteins and genes. Cell 52: 307–308. Gluckman, P. D. and M. A. Hanson. 2004. Living with the past: Evolution, development, and patterns of disease. Science 305: 1733–1736. Go, C. G., J. E. Brustrom, M. F. Lynch and C. M. Aldwin. 1995. Ethnic trends in survival curves and mortality. Gerontologist 35: 318–326. Goglia, R. and V. P. Skulachev. 2003. A function for novel uncoupling proteins: Antioxidant defense of mitochondrial matrix by translocating fatty acid peroxides from the inner to the outer membrane leaflet. FASEB J. 17: 1585–1591. Gogly B., G. Godeau, D. Septier, W. Hornebeck, B. Pellat and C. Jeandel. 1998. Measurement of the amounts of elastic fibers in the skin and temporal arteries of healthy aged individuals by automated image analysis. Gerontology 44: 318–323. Goldberg, P. B. 1978. Cardiac fuction of Fischer 344 rats in relation to age. Pp. 87–100 in G. Kaldor and W. J. D. Battista (eds.), Aging in Muscle. Raven Press, New York. Golden, T. R., D. A. Hinderfeld and S. Melov. 2002.

Oxidative stress and aging: Beyond correlation. Aging Cell 1: 117–123. Goldman, J. E., N. Y. Calingasan and G. E. Gibson. 1994. Aging and the brain. Curr. Opin. Neurol. 7: 287–293. Goldstein, A. L., G. G. Thurman, T. L. K. Low, G. E. Trivers and J. L. Rossie. 1979. Thymosin: The endocrine thymus and its role in the aging process. Pp. 51–60 in A. Cherkin, C. E. Finch, N. Kharasch, T. Makinodan, F. L. Scott and B. L. Strehler (eds.), Aging, vol. 8, Physiology and Cell Biology of Aging. Raven Press, New York. Goldstein, S. 1971. The role of DNA repair in aging of cultured fibroblasts from xeroderma pigmentosum and normals. Proc. Soc. Exp. Med. 137: 730–734. Golubev, A., S. Khrustalev and A. Butov. 2003. An in silico investigation into the causes of telomere length heterogeneity and its implications for the Hayflick limit. J. Theor. Biol. 225: 153–170. Gompertz, B. 1825. On the nature of the function expressive of the law of human mortality and on a new mode of determining the value of life contingencies. Phil. Trans. R. Soc. Lond. 115: 513– 585. Gonzalez, E. and O. Delbono. 2001. Age-dependent fatigue in single intact fast- and slow-fibers from mouse EDL and soleus skeletal muscles. Mech. Ageing Dev. 122: 1019–1032. Gonzalez-Polo, R. A., G. Soler, A. Alvarez, I. Fabregat and J. M. Fuentes. 2003. Vitamin E blocks early events induced by 1-methyl-4-phenylpyridinium (MPP+) in cerebellar granule cells. J. Neurochem. 84: 305–315. Goodrick, C. L. 1975. Life-span and the inheritance of longevity of inbred mice. J. Gerontol. 30: 257– 263. Goodrick, C. L. 1980. Effects of long-term voluntary wheel exercise on male and female Wistar rats. 1. Longevity, body weight and metabolic rate. Gerontology 26: 22–23. Goodrick, C. L., D. K. Ingram, M. A. Reynolds, J. R. Freeman and N. Cider. 1990. Effects of intermittent feeding upon body weight and lifespan in inbred mice: interaction of genotype and age. Mech. Ageing Dev. 55: 69–87. Gorsky, R. D, E. Pamuk, D. F. Williamson, P. A. Shaffer and J. P. Koplan. 1996. The 25 year health care costs of women who remain overweight after 40 years of age. Am. J. Prev. Med. 12: 388–394. Gosain, A. and L. A. DiPietro. 2004. Aging and wound healing. World J. Surg. 28: 321–326. Gould, S. J. 1981. The Mismeasure of Man. Norton, New York. Goya, L., F. Rivero and A. M. Pascual-Leone. 1995. Stress, glucocorticoids and aging. Pp. 249–266 in P. S. Timiras, W. D. Quay and A. Vernadakis

References

(eds.), Hormones and Aging. CRC Press, Boca Raton, FL. Grady, C. L. 1996. Age-related changes in cortical blood flow activation during perception and memory. Ann. N.Y. Acad. Sci. 777: 14–21. Grady, C. L., J. M. Maisog, B. Horwitz, L. G. Ungerleider, M. J. Mentis, J. A. Salerno, P. Pietrini, E. Wagner and J. V. Haxby. 1994. Age-related changes in cortical blood flow activation during visual processing of faces and location. J. Neurosci. 14: 1450–1462. Graf, J-D. and F. J. Ayala. 1986. Genetic variation for superoxide dismutase level in Drosophila melanogaster. Biochem. Genet. 24: 153–168. Graham, C. F. and P. F. Wareing. 1984. Developmental Control in Animals and Plants, 2nd ed. Blackwell, Oxford. Grant, S. G. N. 2003. Synapse signalling complexes and networks: Machines underlying cognition. BioEssays 25: 1229–1235. Graves, J. L. Jr. and L. D. Mueller. 1995. Population density effects on longevity revisited. Genetica 96: 183–186. Graves, J. L., E. C. Toolson,C. Jeung, L. N. Vu and M. R. Rose. 1992. Dessication, flight, glycogen, and postponed senescence in Drosophila melanogaster. Physiol. Zool. 65: 268–286. Greco, M., G. Villani, F. Maxxucchelli, N. M. Bresolin, S. Papa and G. Attardi. 2003. Marked agingrelated decline in efficiency of oxidative phosphorylation in human skin fibroblasts. FASEB J. 14: 1706–1708. Green, D. R. and G. Kroemer. 2004. The pathophysiology of mitochondrial cell death. Science 305: 626–629. Greenblatt, D. J., M. D. Allen and R. I. Shader. 1977. Toxicity of high-dose fluorazepam in the elderly. Clin. Pharmacol. Ther. 21: 355–361. Gregerman, R. I. 1967. The age-related alteration of thyroid function and thyroid hormone metabolism. Pp. 161–173 in L. Girtman (ed.), Endocrines and Aging. Charles C. Thomas, Springfield, IL. Greicus, M. D., G. Srivastava, A. L. Reiss and V. Menon. 2004. Default-mode network activity distinguishes Alzheimer’s disease from healthy aging: evidence from functional MRI. Proc. Natl. Acad. Sci. USA 101: 4637–4642. Grell, K. G. 1973. Protozoology. Springer-Verlag, New York. Gresl, T. A., R. J. Colman, T. C. Havighurst, D. B. Allison, D. A. Schoeller and J. W. Kemnitz. 2003. Dietary restriction and beta-cell sensitivity to glucose in adult male rhesus monkeys. J. Gerontol. Biol. Sci. 58A: 598–610. Griffiths, A. J. F. 1992. Fungal senescence. Annu. Rev. Genet. 26: 351–372. Groopman, J. 2004. Forward, medicine! Science, mo-

547

rality, and embryonic stem cells. The New Republic 231: 35–41. Grube, K. and A. Burkle. 1992. Poly(ADP-ribose) polymerase activity in mononuclear leukocytes of 13 mammalian species correlates with speciesspecific life span. Proc. Natl. Acad. Sci. USA 89: 11759–11763. Gruenbaum, Y., R. D. Goldman, R. Meyuhas, E. Mills, A. Margalit, A. Fridkin, Y. Dayani, et al. 2003. The nuclear lamina and its functions in the nucleus. Intl. Rev. Cytol. 226: 1–62. Guarente, L. 1999. Diverse and dynamic functions of the Sir silencing complex. Nature Genet. 23: 281– 285. Guiamet, J. J., E. Pichesky and L. D. Nooden. 1995. Senescence-inhibiting effects of the stay-green gene cytG in soybean. Plant Physiol. 108: 121 (abstract). Guralnik, J. M. and G. A. Kaplan. 1989. Predictors of healthy aging: Prospective evidence from the Alameda County study. Am. J. Public Health 79: 703–708. Guralnik, J. M. and E. M. Simonsick. 1993. Physical disability in older Americans. J. Gerontol. 48: 3– 10. Gutmann, E. and V. Hanzlikova. 1972. Age Changes in the Neuromuscular System. Scientechnia, Bristol, England. Guyton, A. C. 1966. Textbook of Medical Physiology, 3rd ed. Saunders, Philadelphia. Hacker, J. J. and T. R. Marmon. 2003. Medicare reform: Fact, fiction and foolishness. Public Policy Aging Rpt. 13(4): 1 Hadorn, E. 1978. Transdetermination. Pp. 556–617 in M. Ashburner and T. R. F. Wright (eds.), The Genetics and Biology of Drosophila, vol. 2C. Academic Press, London. Hadshiew, I. M., M. S. Eller and B. A. Gilchrest. 1999. Age-associated decreases in human DNA repair capacity: Implications for the skin. Age 22: 45–57. Hagen, G. and G. Kochert. 1980. Protein syntheis in a new system for the study of senescence. Exp. Cell Res. 127: 451–457. Hahn, J. S., Z. Hu, D. J. Thiele and V. R. Iyer. 2004. Genome-wide analysis of the biology of stress responses through heat shock transcription factor. Mol. Cell Biol. 24: 5249–5256. Hakim, A. A., H. Petrovich, C. M. Burchfiel, G. W. Ross, B. L. Rodriguez, L. R. White, K. Yano, J. D. Curb and R. D. Abbott. 1998. Effects of walking on mortality among nonsmoking retired men. N. Engl. J. Med. 338: 94–99. Halaas, J. L., K. S. Gajiwala, M. Maffei, S. L. Cohen, B. T. Chait, D. Rabinowitz, R. L. Lallone, S. K. Burney and J. M. Friedman. 1995. Weight reducing effects of the plasma protein encoded by the obese gene. Science 269: 543–546.

548

References

Haldane, J. B. S. 1927. A mathematical theory of natural and artificial selection, part IV. Proc. Cambridge Phil. Soc. 23: 607–615. Hall, G. S. 1922. Senescence: The Last Half of Life. Appleton, London. Halliwell, B. and J. M. C. Gutteridge. 1989. Free Radicals in Biology and Medicine, 2nd ed. Clarendon Press, Oxford. Halliwell, B. and J. M. C. Gutteridge. 1999. Free Radicals in Biology and Medicine, 3rd ed. Oxford University Press, New York. Halter, J. B. 1995. Carbohydrate metabolism. Pp. 119– 146 in E. J. Masoro (ed.), Handbook of Physiology, Section 11: Aging. Oxford University Press, New York. Ham, R. G. and M. J. Veomett. 1980. Mechanisms of Development. C. V. Mosby, St. Louis, MO. Hamilton, J. B. and G. E. Mestler. 1969. A comparison of eunuchs with intact men and women in a mentally retarded population. J. Gerontol. 24: 395–411. Han, E-S., S. G. Hilsenbeck, A. Richardson and J. F. Nelson. 2000. cDNA expression arrays reveal incomplete reversal of age-related changes in gene expression by caloric restriction. Mech. Ageing Dev. 155: 157–174. Hanawalt, P. C. 1987. On the role of DNA damage and repair processes in aging: Evidence for and against. Pp. 183–198 in Aging, vol. 31, H. R. Warner, R. N. Butler, R. L. Sprott and E. L. Schneider (eds.), Modern Biological Theories of Aging. Raven Press, New York. Hangartner, J. R. M., N. J. Marley, A. Whitehead, A. C. Thomas and M. I. Davies. 1985. The assessment of cardiac hypertrophy at autopsy. Histopathology 9: 1295–1306. Hardt, H. and M. Rothstein. 1985. Altered phosphoglycerate kinase from old rat muscle shows no change in primary structure. Biochim. Biophys. Acta 831: 13–21. Hari, R. V. Burde and R. Arking. 1998. Immunological confirmation of elevated levels of CuZn superoxide dismutase protein in an artificially selected long-lived strain of Drosophila melanogaster. Exp. Gerontol. 33: 227–238. Hariharan, I. K. and D. A. Haber. 2003. Yeast, flies, worms, and fish in the study of human disease. N. Engl. J. Med. 348: 2457–2463. Harley, C. A., A. B. Futcher and C. W. Greider. 1990. Telomeres shorted during ageing of human fibroblasts. Nature 345: 458–460. Harley, C. A., H. Vaziri, C. M. Counter and R. C. Allsopp. 1992. The telomere hypothesis of cellular aging. Exp. Gerontol. 27: 375–382. Harman, D. 1956. Aging: A theory based on free radical and radiation chemistry. J. Gerontol. 11: 298– 300.

Harman, D. 1972. The biologic clock: the mitochondria? J. Am. Geriatr. Soc. 20: 145–147. Harman, S. M., E. J. Metter, J. D. Tobin, J. Pearson and M. R. Blackman. 2001. Longitudinal effects of aging on serum total and free testosterone levels in healthy men. J. Clin. Endocrin. Metabol. 86: 724–731. Harper, J. M., A. T. Galecki, D. T. Burke and R. A. Miller. 2004. Body weight, hormones, and T cell subsets as predictors of life span in genetically heterogeneous mice. Mech. Ageing Dev. 125: 381–390. Harper, J. M., N. Wolf, A. T. Galecki, S. L. Pinosky and R. A. Miller. 2003. Hormone levels and cataract scores as sex-specific, mid-life predictors of longevity in genetically heterogeneous mice. Mech. Aging Dev. 124: 801–810. Harris, N., V. Costa, M. MacLean, M. Mollapour, P. Moradas-Ferreira and P. W. Piper. 2003. Mnsod overexpression extends the yeast chronological (G(0)) life span but acts independently of Sir2p histone deacetylase to shorten the replicative life span of dividing cells. Free Rad. Biol. Med. 34: 1599–1606. Harris, R. E. and R. P. Forsythe. 1973. Personality and emotional stress in essential hypertension in man. Pp. 125–132 in G. Onesta, K. E. Kim and J. H. Meyer (eds.), Hypertension: Mechanisms and Management. Grune and Stratton, New York. Harris, S. B., R. Weindruch, G. S. Smith, M. R. Mickey and R. L. Walford. 1990. Dietary restriction alone and in combination with oral ethoxyquin/2mercaptoethylamine in mice. J. Gerontol. Biol. Sci. 45: B141–B147. Harrison, D. 1982. Must we grow old? Biol. Digest 8: 11–25. Harrison, D. 1985. Cell and tissue transplantation: A means of studying the aging process. Pp. 322–356 in C. E. Finch and E. L. Schneider (eds.), Handbook of the Biology of Aging, 2nd ed. Van Nostrand Reinhold, New York. Harrison, D. E. 1994. Potential misinterpretations using models of accelerated aging. J. Gerontol. Biol. Sci. 49: B245. Harrison, D. E. and J. R. Archer. 1987. Genetic differences in effects of food restriction on aging in mice. J. Nutr. 117: 376–382. Harrison, D. E., J. R. Archer and C. M. Astle. 1982. The effect of hypophysectomy on thymic aging in mice. J. Immunol. 136: 2673. Harrison, D. E., J. R. Archer and C. M. Astle. 1984. Effects of food restriction on aging: Segregation of food intake and adiposity. Proc. Natl. Acad. Sci. USA 81: 1835–1838. Harrison, D. E. Jr., J. R. Archer, G. A. Sacher and F. M. Boyce III. 1978. Tail collagen aging in mice of thirteen different genotypes and two species:

References

Relationship to biological age. Exp. Gerontol. 13: 63–73. Harrison, D. E. and T. Roderick. 1997. Selection for maximum longevity in mice. Exp. Gerontol. 32: 65–78. Harrison, R. G. 1910. The outgrowth of the nerve fiber as a mode of protoplasmic movement. J. Exp. Zool. 9: 787–848. Harshman, L. G., K. M. Moore, M.A. Sty and M. M. Magwire. 1999. Stress resistance and longevity in selected lines of Drosophila melanogaster. Neurobiol. Aging. 20: 521–529. Hart, C. L., M. D. Taylor, G. D. Smith, L. J. Whalley, J. M. Starr, D. J. Hole, V. Wilson and I. J. Deary. 2005. Childhood IQ and all-cause mortality before and after age 65: prospective observational study linking the Scottish Mental Survey 1932 and the Midspan studies. Br. J. Health. Psychol. 10: 153–165. Hart, R. and R. B. Setlow. 1974. Correlation between deoxyribonucleic acid excision-repair and life-span in a number of mammalian species. Proc. Natl. Acad. Sci. USA 71: 2169–2173. Hart, R. W., D. A. Neumann and R. T. Robertson (eds.). 1995. Dietary Restriction: Implications for the Design and Interpretation of Toxicity and Carcinogenicity Studies. ILSI Press, Washington, DC. Hart, R. W. and A. Turturro. 1983. Theories of aging. Pp. 5–18 in M. Rothstein (ed.), Review of Biological Research in Aging, vol. 1. Alan R. Liss, New York. Hartman, P. E. and R. W. Morgan. 1985. Mutageninduced focal lesions as key factors in aging: A review. Pp. 93–136 in R. S. Sohal, L. S. Birnbaum and R. G. Cutler (eds.), Molecular Biology of Aging: Gene Stability and Gene Expression. Raven Press, New York. Hartman, P., E. Childress and T. Beyer. 1995. Nematode development is inhibited by methyl viologen and high oxygen concentrations at a rate inversely proportional to life span. J. Gerontol.: Biol. Sci. 50A: B322–B326. Hartung, G. H., E. J. Farge and R. E. Mitchell. 1981. Effects of marathon running, jogging and diet on coronary risk factors in middle-aged men. Prev. Med. 10: 316–323. Hartwell, L. H. and M. B. Kastan. 1994. Cell cycle control and cancer. Science 266: 1821–1828. Harvell, C. D. and R. K. Grossberg. 1988. The timing of sexual maturity in clonal animals. Ecology 69: 1855–1864. Hasty, P., J. Campisi, J. Hoeljmakers, H. van Steeg and J. Vijg. 2003. Aging and genome maintenance: Lessons from the mouse? Science 299: 1355–1359. Haug, H. and R. Eggers. 1991. Morphometry of the human cortex cerebri and corpus striatum during aging. Neurobiol. Aging. 12: 336–338.

549

Hausman, P. B. and M. E. Weksler. 1985. Changes in the inmmune response with age. Pp. 414–432 in C. E. Finch and E. L. Schneider (eds.), Handbook of the Biology of Aging, 2nd ed. Van Nostrand Reinhold, New York. Hayflick, L. 1965. The limited in vitro lifetime of human diploid cell strains. Exp. Cell Res. 37: 614– 636. Hayflick, L. 1977. The cellular basis for biological aging. Pp. 159–188 in C. E. Finch and L. Hayflick (eds.), Handbook of the Biology of Aging. Van Nostrand Reinhold, New York. Hayflick, L. 1982. Biological aspects of human aging. Br. J. Hosp. Med. 27: 366. Hayflick, L. 1985. Theories of biological aging. Exp. Gerontol. 20: 145–159. Hayflick, L. 1987. Origins of longevity. Pp. 21–31 in H. R. Warner, R. N. Butler, R. L. Sprott and E. L. Schneider (eds.), Aging, vol. 31, Modern Biological Theories of Aging. Raven Press, New York. Hayflick, L. 2000. The future of ageing. Nature 408: 267–269. Hayflick. L. 2002. Anarchy in gerontological terminology. Gerontologist 42: 416–421. Hayflick, L. 2003. Living forever and dying in the attempt. Exp. Gerontol. 38: 1231–1241. Hayflick, L. and P. S. Moorhead. 1961. The limited in vitro lifetime of human diploid cell strains. Exp. Cell Res. 25: 585–621. Hazzard, D. G. and J. Soban. 1988. Studies of aging using genetically defined rodents: A bibliography. Exp. Aging Res. 14: 59–81. Hazzard, W. R. 1986a. Aging, lipoprotein metabolism and atherosclerosis: A clinical conundrum. Pp. 75– 103 in S. R. Bates and E. C. Gangloff (eds.), Atherogenesis and Aging. Springer-Verlag, New York. Hazzard, W. R. 1986b. Biological basis of the sex differential in longevity. J. Am. Geriatr. Soc. 34: 455– 471. Heikkinen, E., H. Suominen, P. Era and A.-L. Lyyra. 1994. Variations in aging parameters, their sources and possibilities of predicting physiological age. Pp. 71–92 in A. K. Balin (ed.), Practical Handbook of Human Biologic Age Determination. CRC Press, Boca Raton, FL. Heilbronn, L. K. and E. Ravussin. 2003. Calorie restriction and aging: review of the literature and implications for study in humans. Am. J. Clin. Nutr. 78: 361–369. Hekimi, S. and L. S. Guarente. 2003. Genetics and the specificity of the aging process. Science 299: 1351– 1354. Hekimi, S. 2000. Crossroads of aging in the nematode Caenorhabditis elegans. Pp. 81–112 in S. Hekimi (ed.), Results and Problems in Cell Differentiation, vol. 29. The Molecular Genetics of Aging. Springer-Verlag, Berlin Heidelberg.

550

References

Hekimi, S., P. Boutis and B. Lakowski. 1995. Viable maternal-effect mutations that affect the development of the nematode Caenorhabditis elegans. Genetics 141: 1351–1364. Hekimi, S., J. Burgess, F. Bussiere, Y. Meng and C. Benard. 2001. Genetics of lifespan in C. elegans: molecular diversity, physiological complexity, mechanistic simplicity. Trends Genet. 17: 712– 718. Helfand, S. L., K. J. Blake, B. Rogina, M. D. Stracks, A. Centurion and B. Naprta. 1995. Temporal patterns of gene expression in the antenna of the adult Drosophila melanogaster. Genetics 140: 549–555. Helfand, S.L. and B. Rogina. 2000. Regulation of gene expression during aging. Results Probl. Cell Differ. 29: 67–80. Helfand, S. L. and B. Rogina. 2003a. From genes to aging in Drosophila. Adv. Genet. 49: 67–109. Helfand, S. L. and B. Rogina. 2003b. Molecular genetics of aging in the fly: Is this the end of the beginning? BioEssays 25: 134–141. Helle, S., V. Lummaa and J. Jokela. 2002. Sons reduced maternal longevity in preindustrial humans. Science 296: 1085. Hellemans, L., H. Corstjens, A. Neven, L. Declercq and D. Maes. 2003. Antioxidant enzyme variation in human strateum corneum shows seasonal variation with an age-dependent recovery. J. Invest. Dermatol. 120;434–439. Hengge-Aronis, R. 2002. Stationary phase gene regulation: what makes an Escherichia coli promoter ss -selective? Curr. Opin. Microbiol. 5: 591–595. Henon, N., M. Busson, C. Dehay-Martouchou, O. Charron and J. Hors. 1999. Familial versus longevity and MHC markers. J. Biol. Regul. Homeost. Agents 13: 27–31. Henson, S. M., J. Pido-Lopez and R. Aspinall. 2004. Reversal of thymic atrophy. Exp. Gerontol. 39: 673–678. Hercus, M. J., V. Loeschcke and S. I. Rattan. 2004. Lifespan extension of Drosophila melanogaster through hormesis by repeated mild heat stress. Biogerontology 4: 149–156. Hermann, M., G. Untergasser, H. Rumpold and P. Berger. 2000. Aging of the male reproductive system. Exp. Gerontol. 35: 1267–1279. Hermanns, J. and H. D. Osiewacz. 1995. Evidence for a life span prolonging effect of a linear plasmid in a longevity mutant of Podospora anserina. Mol. Gen. Genet. 243: 297–305. Herndon, L. A., P. H. Schmeissner, J. M. Dudaronek, P. A. Brown, K. M. Listner, Y. Sadano, M. C. Paupard, D. H. Hall and M. Droscoll. 2002. Stochastic and genetic factors influence tissue-specific decline in ageing C. elegans. Nature 419: 808–814. Herrero, A. and G. Barja. 1997a. ADP-regulation of mitochondrial free radical production is different

with complex I- or complex II-linked substrates: Implications for the exercise paradox and brain hypermetabolism. J. Bioenerg. Biomembr. 29: 241–249. Herrero, A. and G. Barja. 1997b. Sites and mechanisms responsible for the low rate of free radical production of heart mitochondria in the long-lived pigeon. Mech. Ageing Dev. 92: 95–111. Hershcopf, R. J., D. Elahi, R. Andres, H. L. Baldwin, G. S. Raizes, D. D. Schocken and N. W. Shock. 1982. Longitudinal changes in serum cholesterol in man: An epidemiologic search for etiology. J. Chronic Diseases 35: 101–114. Heuser, I. and C.-H. Lammers. 2003. Stress and the brain. Neurobiol. of Aging 24: 869–876. Heydari, A. R. and A. Richardson. 1992. Does gene expression play any role in the mechanism of the antiaging effect of dietary restriction. Ann. N.Y. Acad. Sci. 663: 384–395. Heydari, A. R., B. Wu, R. Takahashi, R. Strong and A. Richardson. 1993. Expression of heat shock protein 70 is altered by age and diet at the level of transcription. Mol. Cell. Biol. 13: 2909–2918. Heydari, A. R., S. You, R. Takahashi, A. GutsmannConrad, K. D. Sarge and A. Richardson. 2000. Age-related alterations in the activation of heat shock transcription factor 1 in rat hepatocytes. Exp Cell Res. 256(1): 83–93. Heyman, Y. 2005. Nuclear transfer: a new tool for reproductive biotechnology in cattle. Reprod. Nutr Dev. 45: 353–361. Higami, Y., B. P. Yu, I. Shimokawa, H. Bertrand, G. B. Hubbard and E. J. Masoro. 1995. Anti-tumor action of dietary restriction is lesion dependent in male Fischer 344 rats. J. Gerontol. Biol. Sci. 50A: B72–B77. Higami, Y., B. P. Yu, I. Shimokawa, E. J. Masoro and T. Ikeda. 1994. Duration of dietary restriction: An important determinant for the incidence and age of onset of leukemia in male F344 rats. J. Gerontol. Biol. Sci. 49: B239–B244. Himes, C. L. 1994. Age patterns of mortality and cause-of-death structures in Sweden, Japan and the United States. Demography 31: 633–650. Hinderhofer, K and U. Zentgraf. 2001. Identification of a transcription factor specifically expressed at the onset of leaf senescence. Planta 213: 469–473. Hinerfeld, D., M. D. Traini, R. P. Weinberger, B. Cochran, S. R. Doctrow, J. Harry and S. Melov. 2004. Endogenous mitochondrial oxidative stress: Neurodegeneration, proteomic analysis, specific respiratory chain defects, and efficacious antioxidant therapy in superoxide dismutase 2 null mice. J. Neurochem. 88: 657–667. Hirano, T., R. Yamaguchi, S. Asami, N. Iwamoto and H. Kasai. 1996. 8-Hydroxyguanine levels in nuclear DNA and its repair in rat organs associ-

References

ated with age. J. Gerontol. Biol. Sci. 51A: B303– B307. Hirsch, H. R., J. A. Coomes and M. Witten. 1989. The waste-product theory of aging: Transformation to unlimited growth in cell cultures. Exp. Gerontol. 24: 97–112. Hochschild, R. 1989. Improving the precision of biological age determinations. II. Automatic human tests, age norms and variability. Exp. Gerontol. 24: 301–316. Hochschild, R. 1994. Validating biomarkers of aging— mathematical approaches and results of a 2462 person study. Pp. 93–144 in A. K. Balin (ed.), Practical Handbook of Human Biologic Age Determination. CRC Press, Boca Raton, FL. Hocht-Zeisberg, E., H. Kahnert, K. Guan, G. Wulf, B. Hemmerlein, T. Schlott, G. Tenderich, R. Korfer, U. Raute-Kreinsen and G. Hasenfuss. 2004. Cellular repopulation of myocardial infarction in patients with sex-mismatched heart transplantation. Eur. Heart J. 25: 749–758. Hodgkin, J. 1989. Early worms. Genetics 121: 1–3. Hoff, S. F., S. W. Scheff, L. S. Bernardo and C. W. Cotman. 1982. Lesion induced synaptogenesis in the dendate gyrus of aged rats. I. Loss and reacquisition of normal synaptic density. J. Comp. Neurol. 205: 246–252. Hoffmann, A. and P. Parsons. 1989. Selection for increased desiccation resistance in Drosophila melanogaster. Additive genetic control and correlated responses for other stresses. Genetics 122: 837– 845. Hofman, M. A. 1983. Energy metabolism, brain size and longevity in mammals. Q. Rev. Biol. 58: 495– 512. Holden, C. 1987a. The genetics of personality. Science 237: 598–601. Holden, C. 1987b. Why do women live longer than men? Science 238: 158–160. Holliday, R. 1984. The unsolved problem of cellular ageing. Monogr. Dev. Biol. 17: 60–77. Holliday, R. 1989. Food, reproduction and longevity: Is the extended longevity of calorie-restricted animals an evolutionary adaptation. BioEssays 10: 125–127. Holliday, R. 1990. The limited proliferation of cultured human diploid cells: Regulation or senescence? J. Gerontol. Biol. Sci. 45: B36–B41. Holliday, R. 1994. Longevity and fecundity in eutherian mammals. Pp. 217–225 in M. R. Rose and C. E. Finch (eds.), Genetics and Evolution of Aging. Kluwer Academic Publishers, Dordrecht, the Netherlands. Holliday, R. 1995. Understanding Ageing. Cambridge University Press, Cambridge. Holliday, R. 1996a. The evolution of human longevity. Perspect. Biol. Med. 40: 100–107.

551

Holliday, R. 1996b. The urgency of research on ageing. BioEssays 18: 89–90. Holliday, R. 2001. Human ageing and the origins of religion. Biogerontology 2: 73–77. Holliday, R., L. I. Huschtscha, G. M. Tarrant and T. B. L. Kirkwood. 1977. Testing the commitment theory of cellular aging. Science 198: 366–372. Holliday, R. and T. B. L. Kirkwood. 1981. Predictions of the somatic mutation and mortalization theories of cellular ageing are contrary to experimental observations. J. Theoret. Biol. 93: 627–642. Holloszy, J. O. 1993. Exercise increases average longevity of female rats despite increased food intake and no growth retardation. J. Gerontol. Biol. Sci. 48: B97–B100. Holloszy, J. O. and W. M. Kohrt. 1995. Exercise. Pp. 633–666 in E. J. Masoro (ed.), Handbook of Physiology. Section 11: Aging. Oxford University Press, New York. Holmes, D. J. and S. N. Austad. 1995a. Birds as animal models for the comparative biology of aging: A prospectus. J. Geontol. Biol. Sci. 50A: B59–B66. Holmes, D. J. and S. N. Austad. 1995b. The evolution of avian senescence patterns: Implications for understanding primary aging processes. Am. Zool. 35: 307–317. Holmes, D. J., R. Fluckiger and S. N. Austad. 2001. Comparative biology of aging in birds: an update. Exp Gerontol. 36: 869–83. Holmes, G. E. and N. R. Holmes. 1986. Accumulation of DNA changes in aging Paramecium tetraurelia. Mol. Gen. Genet. 204: 108–114. Holzenberger, M., J. Dupont, B. Ducos, P. Leneuve, A. Geloen, P. C. Evans, P. Cervera and Y. Le Bouc. 2003. IGF-1 receptor regulates lifespan and resistance to oxidative stress in mice. Nature 421: 182– 187. Honda, Y. and S. Honda. 2001. Life span extensions associated with upregulation of gene expression of antioxidant enzymes in Caenorhabditis elegans: studies of mutation in the age-1, PI3 kinase homologue and short-term exposure to hyperoxia. J. Am. Aging Assoc. 24: 179–186. Hood, L. 2003. Systems biology: integrating technology, biology, and computation. Mech. Aging Dev. 124: 9–16. Hopkin, K. 1995. Chromosome 21 genes in Down syndrome and development. J. NIH Res. 7: 29–30. Horiouchi, S., C. E. Finch, F. Mesle and J. Vallin. 2003. Differential patterns of age-related mortality increase in middle age and old age. J. Gerontol. Biol. Sci. 58A: 495–507. Horiuchi, S. and J.R. Wilmoth, 1998. Deceleration in the age pattern of mortality at older ages. Demography 35: 391–412. Horowitz, M. C. 1993. Cytokines and estrogen in bone: Anti-osteoporotic effects. Science 260: 626–627.

552

References

Horton, D. L. 1967. The effect of age on hair growth in the CBA mouse: Observations on transplanted skin. J. Gerontol. 22: 43–46. Horvath, S. M. 1981. Aging and adaption to stressors. Pp. 437–451 in S. M. Horvath and M. K. Yousef (eds.), Environmental Physiology: Aging, Heat and Altitide. Elsevier/North Holland, New York. House, J. S., K. R. Landis and U. Umberson. 1988. Social relationships and health. Science 241: 540– 545. Hovemann, B., S. Richter, U. Walldorf and C. Cziepluch. 1988. Two genes encode related cytoplasmic elongation factors (EF-1 alpha) in Drosophila melanogaster with continuous and stage specific expression. Nucleic Acids Res. 16: 3175– 3194. Hrdy, S. B. 2002. Motherhood. Pp. E55–E65 in M. Pagel (ed.), Encyclopedia of Evolution, vol. 1. Oxford University Press, New York. Hsiao, K., P. Chapman, S. Nilsen, C. Eckman, Y. Harigaya, S. Younkin, F. Yang and G. Cole. 1996. Correlative memory deficits, Abeta elevation, and amyloid plaques in transgenic mice. Science 274: 99–102. Hsin, H. and C. Kenyon. 1999. Signals from the reproductive system regulate the lifespan of C. elegans. Nature 399: 362–366. Huang, T-T., E. J. Carlson, A. M. Gillespie, Y. Shi and C. J. Epstein. 2000. Ubiquitous overexpression of CuZn superoxide dismutase does not extend life span in mice. J. Gerontol. Biol. Sci. 55A: B5–B9. Hubert, H. B., D. A. Bloch, J. W. Oehlert and J. F. Fries. 2002. Lifestyle habits and compression of morbidity. J.Gerontol. Med. Sci. 57A: M347– M351. Hubner, K., G. Fuhrmann, L. K. Christenson, J. Kehler, R. Reinbold, R. De La Fuente, J. Wood, et al. 2003. Derivation of oocytes from mouse embryonic stem cells. Science 300: 1251–1256. Hughes, A. L. and M. K. Hughes. 1995. Small genomes for better flyers. Nature 377: 391. Hutchinson, E. W. and M. R. Rose. 1991. Quantitative genetics of postponed aging in Drosophila melanogaster. I. Analysis of outbred populations. Genetics 127: 719–727. Hutchinson, E. W., A. J. Shaw and M. R. Rose. 1991. Quantitative genetics of postponed aging in Drosophila melanogaster. II. Analysis of selected lines. Genetics 127: 729–737. Hutter, E., H. Unterluggauer, F. Uberall, H. Schramek and P. Jansen-Durr. 2002. Replicative senescence of human fibroblasts: The role of Ras-dependent signaling and oxidative stress. Exp. Gerontol. 37: 1165–1174. Hwang, W. S., S. I. Roh, B. C. Lee, S. K. Kang, D. K. Kwon, S. Kim, S. J. Kim, S. W. Park, H. S. Kwon, C. K. Lee, J. B. Lee, J. M. Kim, C. Ahn, S. H. Paek,

S. S. Chang, J. J. Koo, H. S. Yoon, J. H. Hwang, Y. Y. Hwang, Y. S. Park, S. K. Oh, H. S. Kim, J. H. Park, S. Y. Moon and G. Schatten. 2005. Patient-specific embryonic stem cells derived from human SCNT blastocysts. Science 308: 1777– 1783. Hwangbo D-S., B. Gersham, M-P. Tu, M. Palmer and M. Tatar. 2004. Drosophila dFOXO controls lifespan and regulates insulin signalling in brain and fat body. Nature. DOI:10.1038/nature02549; www.nature.com/nature. Hwangbo, D-S., M. Lane and R. Arking. 2003. Epistatic interactions of the insulin-like signaling system genes in normal- and long-lived strains of Drosophila. Submitted. Ideker, T., V. Thorsson, J. A. Ranish, R. Christmas, J. Buhler, J. K. Eng, R. Bumgarner, D. R. Goodlett, R. Aebersold and L. Hood. 2001. Integrated genomic and proteomic analyses of a systematically perturbed metabolic network. Science 292: 929–934. Ikari, H., L. Zhang, J. M. Chernak, A. Mastrangeli, S. Kato, H. Kuo, R. G. Crystal, D. K. Ingram and G. S. Roth. 1995. Adenovirus-mediated gene transfer of dopamine D2 receptor cDNA into rat striatum. Mol. Brain Res. 34: 315–320. Ikehara, S., J. Shimiu, R. Yasumizu, T. Nakamura, M. Inaba, S. Inoue, N. Oyaizu, K. Sugiura, M. M. Oo, Y. Hamashima and R. Good. 1987. Thymic rudiments are responsible for induction of functional T cells in nu/nu mice. Thymus 10: 193–205. Ikeno, Y., R. T. Bronson, G. B. Hubbard, S. Lee and A. Bartke. 2003. Delayed occurrence of fatal neoplatic diseases in Ames dwarf mice: Correlation to extended longevity. J. Gerontol. Biol. Sci. 58A: 291–296. Ikeyama, S., G. Kokkonen, S. Shack, X-T. Wang and N. J. Holbrook. 2002. Loss in oxidative stress tolerance with aging linked to reduced extracellular signal-regulated kinase and Akt kinase activities. FASEB J. 16: 114–116. Ingram, D. K. 1999. Design of cross-sectional, longitudinal, and sequential studies in gerontology. Pp. 25–42 in B. P. Yu (ed.) Methods in Aging Research. CRC Press, Boca Raton, FL. Ingram, D. K., R. G. Cutler, R. Weindruch, D. M. Renquist, J. J. Knapka, M. April, C. T. Belcher, M. A. Clark, C. D. Hatcherson, B. M. Marriott and G. S. Roth. 1990. Dietary restriction and aging: The initiation of a primate study. J. Gerontol. Biol. Sci. 45: B148–B163. Ingram, D. K., R. M. Anson, R. de Cabo, J. Mamczarz, M. Zhu, J. Mattison, M. A. Lane and G. S. Roth. 2004. Development of calorie restriction mimetics as a prolongevity strategy. Ann. N.Y. Acad. Sci. 1019: 412–423. Ingram, D. K., E. Nakamura, D. Smucny, G. S. Roth and M. A. Lane. 2001. Strategy for indentifying

References

biomarkers of aging in long-lived species. Exp. Gerontol. 36: 1025–1034. Ingram, D. K. and M. A. Reynolds. 1987. The relationship of body weight to longevity within laboratory rodent species. Pp. 247–282 in A. D. Woodhead and K. H. Thompson (eds.), Evolution of Longevity in Animals: A Comparative Approach. Plenum Press, New York. Ingram, D. K., S. Stoll and G. T. Baker III. 1995. Is attempting to assess biological age worth the effort? Gerontologist 35: 707–710 (book review). Ingram, D. K., R. Weindruch, L. E. Spangler, J. R. Freeman and R. L. Walford. 1987. Dietary restriction benefits learning and motor performance of aged mice. J. Gerontol. 42: 78–81. Inokuchi, S., H. Ishikawa, S. Iwamoto and T. Kimura. 1975. Age related changes in the histological composition of the rectus abdominis muscle of the adult human. Hum. Biol. 47: 231–249. Iovino, M., P. Monteleone and L. Steardo. 1989. Repetitive growth hormone releasing hormone administration restores the attenuated growth hormone (GH) response to GH-releasing hormone testing in normal aging. J. Clin. Endocrinol. Metab. 69: 910–913. Ivanova, R. N. Henon, V. Lepage, D. Charron, E. Vincent and F. Schachter. 1998. HLA-DR alleles display sex-dependent effects on survival and discriminate between individual and familial longevity. Hum. Mol. Genet. 7: 187–194. Iwasaki, K., C. A. Gleiser, E. J. Masoro, A. McMahan, E. J. Seo and B. P. Yu. 1988. The influence of dietary protein source on longevity and age related disease processes of Fischer rats. J. Gerontol. Biol. Sci. 43: B5–B12. Iwase, M., M. Wada, N. Shinohara, H. Yoshizumi, M. Yoshinari and M. Fujishima. 1995. Effect of maternal diabetes on longevity in offspring of spontaneously hypertensive rats. Gerontology 41: 181–186. Jackson, A. U., A. Fornes, A. Galecki, R. A. Miller and D. T. Burke. 1999. Multiple-trait quantitative trait loci analysis using a large mouse sibship. Genetics 151: 785–795. Jacob, F. 1982. The Possible and the Actual. Pantheon, New York. Jacob, R. A., G. M. Aiello, C. B. Stephensen, J. B. Blumberg, P. E. Milbury, L. M. Wallock and B. N. Ames. 2003. Moderate antioxidant supplementation has no effect on biomarkers of oxidant damage in healthy men with low fruit and vegetable intakes. J. Nutr. 133: 740–743. Jacobs, B., M. Schall and A. B. Scheibel. 1993. A quantitative dendritic analysis of Wernicke’s in humans. II. Gender, hemispheric and environmental factors. J. Comp. Neurol. 327: 97–111. Jacobs, H. T. 2003. The mitochondrial theory of aging: dead or alive? Aging Cell 2: 11–18.

553

Jacobs, P. A. and W. M. Court-Brown. 1966. Age and chromosomes. Nature 212: 823. Jaenisch, R. and A. Bird. 2003. Epigenetic regulation of gene expression: How the genome integrates intrinsic and environmental signals. Nature Genet. 33 (suppl.): 245–254. Jalavisto, E. 1951. Inheritance of longevity according to Finnish and Swedish genealogies. Ann. Med. Int. Fenn. 40: 263–274. James, A. C. and J. W. Ballard. 2003.Mitochondrial genotype affects fitness in Drosophila simulans. Genetics. 164: 187–194. Jarvik, L. F. 1979. Genetic aspects of aging. Pp. 86– 109 in I. Rossman (ed.), Clinical Geriatrics, 2nd ed. Lippincott, Philadelphia. Jarvik, L. F. 1988. Aging of the brain: How can we prevent it. Gerontologist 28: 739–747. Jazwinski, S. M. 1993. The genetics of aging in the yeast Saccharomyces cerevisiae. Genetica 91: 35–51. Jazwinski, S. M. 1996. Longevity, genes and aging. Science 273: 55–59. Jazwinski, S. M. 1999. The RAS genes: a homeostatic device in Saccharomyces cerevisiae longevity. Neurobiol. Aging 20: 471–478. Jazwinski, S. M. 2000a. Coordination of metabolic activity and stress resistance in yeast longevity. Pp. xx–xx in Hekemi (ed.), Results and Problems in Cell Differentiation, vol. 29, The Molecular Genetics of Aging. Springer-Verlag, Berlin. Jazwinski, SM. 2000b. Metabolic control and gene dysregulation in yeast aging. Ann. N.Y. Acad. Sci. xxx: 21–30. Jazwinski, S. M., N. K. Egilmez and J. B. Chen. 1989. Replication control and cellular life span. Exp. Gerontol. 24: 423–436. Jazwinski, S. M., B. H. Howard and R. K. Nayak. 1995. Cell cycle progression, aging and cell death. J. Gerontol. Biol. Sci. 50A: B1–B8. Jazwinski, S.M., S. Kim, C-Y Lai and A. Benguria. 1998. Epigenetic stratification: The role of individual change in the biological aging process. Exp. Gerontol. 33: 571–580. Jensen, R. E., C. D. Dunn, M. J. Youngman and H. Sesaki. 2004. Mitochondrial building blocks. Trends Cell Biol. 14: 215–218. Jernigan, T. L., L. M. Zatz, I. Feinberg and G. Fein. 1980. The measurement of cerebral atrophy in the aged by computed tomography. Pp. 86–96 in L. Poon (ed.), Aging in the 1980’s. APA Press, Washington, DC. Jesmin, S., Y. Hattori, I. Sakuma, M. Y. Liu, D. N. Mowa and A. Kitabatake. 2003. Estrogen deprivation and replacement modulate cerebral capillary density with vascular expression of angiogenic molecules in middle-aged female rats. J. Cereb. Blood Flow Metab. 23: 181–189.

554

References

Jessberger, S. and G. Kempermann. 2003. Adult-born hippocampal neurons mature into activity-dependent responsiveness. Eur. J. Neurosci. 18: 2707– 2712. Jessie, B. C., C. Q. Sun, H. R. Irons, F. F. Marshall, D. C. Wallace and J. A. Peros. 2001. Accumulation of mitochondrial DNA deletions in the malignant prostate of patients of different ages. Exp. Gerontol. 37: 169–174. Jeune, B. and J. W. Vaupel (eds.). 1999. Validation of Exceptional Longevity. Odense University Press, Odense, Denmark. Jia, K., P. S. Albert and D. L. Riddle. 2002. DAF-9, a cytochrome P450 regulating C. elegans larval development and adult longevity. Development 129: 221–231. Jiang, C. H., J. Z. Tsien, P. G. Schultz and Y. Hu. 2001. An evaluation of replicative senescence in culture as a model for aging in situ. J.Gerontol.: Biol.Sci. 58A: 779–781. The effects of aging on gene expression in the hypothalmus and cortex of mice. Proc. Natl. Acad. Sci. USA 98: 1930–1934. Jiang, J. C., J. Wawryn, M. M. C. Shantha Kumara and S. M. Jazwinski. 2002. Distinct roles of processes modulated by histone deacetylases Rpd3p, Hda1p, and Sir2p in life extension by caloric restriction in yeast. Exp. Gerontol. 37: 1023–1030. Jiang, D. J., P. A. Kirchman, M. Zagulski, J. Hunt and S. M. Jazwinski. 1998. Homologs of the yeast longevity gene LAG1 in Caenorhabditis elegans and human. Genome Res. 8: 1259–1272. Jiang, Q. and B. N. Ames. 2003. Gamma-tocopherol, but not alpha-tocopherol, decreases proinflammatory eicosanoids and inflammation damage in rats. FASEB J. 17: 816–822. Jing, H. C., M. H. Sturre, J. Hille and P. P. Dijkwel. 2002. Arabidopsis onset of leaf senescence mutants identify a regulatory pathway controlling leaf senescence. Plant J. 32: 51–63. Jishage, M., K. Kvint, V. Shingler and T. Nystrom. 2002. Regulation of sigma factor competition by the alarmone ppGpp. Genes Dev. 16: 1260–1270. Johnson, D. C. 2003. The social security promises not yet kept. New York Times. Date, page. Johnson, J., J. Canning, T. Kaneko, J. K. Pru and J. L. Tilly. 2004. Germline stem cells and follicular renewal in the postnatal mammalian ovary. Nature 428: 145–150. Johnson, S. A. and C. E. Finch. 1996. Changes in gene expression during brain aging: A survey. Pp. 300– 327 in E. L. Schneider and J. W. Rowe (eds.), Handbook of the Biology of Aging, 4th ed. Academic Press, San Diego, CA. Johnson, T. E. 1987. Aging can be genetically dissected into component processes using long lived lines of Caenorhabditis elegans. Proc. Natl. Acad. Sci USA 84: 3777–3781.

Johnson, T. E. 1988. Genetic specification of lifespan: Process, problems and potentials. J. Gerontol. Biol. Sci. 43: B87–B92. Johnson, T. E. 2002. Subfield history: Caenorhabditis elegans as a system for analysis of the genetics of aging. http://sageke.sciencemag.org/cgi/content/ full/sageke;2002/34/re4. Johnson, T. E. and H. Bruunsgaard. 1998. Implications of hormesis for biomedical aging research. Hum. Exp. Toxicol. 17: 263–265. Johnson, T. E., G. J. Lithgow and S. Murakami. 1996. Hypothesis: Interventions that increase the response to stress offer the potential for effective life prolongation and increased health. J. Gerontol.: Biol. Sci. 51A: B392–B395. Johnson, T. E., P. M. Tedesco and G. J. Lithgow. 1993. Comparing mutants, selective breeding and transgenics in the dissection of aging processes of Caenorhabditis elegans. Genetica 91: 65–78. Johnson, T. E. and W. G. Wood. 1982. Genetic analysis of life span in Caenorhabditis elegans. Proc. Natl. Acad. Sci. USA 79: 6603–6607. Jonassen, T., P. L. Larsen and C. F. Clarke. 2001. A dietary source of coenzyme Q is essential for growth of long-lived Caenorhabiditis elegans clk-1 mutants. Proc. Natl. Acad. Sci. USA 98: 421–426. Jones, H. B. 1959. The relation of human health to age, place, and time. Pp. 336–363 in J. E. Birron (ed.), Handbook of Aging and the Individual. University of Chicago Press, Chicago. Jost, B. C. and G. T. Grossberg. 1995. The natural history of Alzheimer’s Disease: A Brain Bank study. J. Am. Geriatr. Soc. 43: 1248–1255. Journal of Gerontology. Special issue. March 2001. Jozsi, A. C., E. E. Dupont-Versteegden, J. M. TaylorJones, W. J. Evans, T. A. Trappe, W. W. Campbell and C. A. Peterson. 2000. Aged human muscle demonstrates an altered gene expression profile consistent with an impaired response to exercise. Mech. Ageing Dev. 120: 45–56. Judd, E. M., M. T. Laub and H. H. McAdams. 2000. Toggles and oscillators: new genetic circuit designs. BioEssays 22: 507–509. Judge, S., A. Judge, T. Grune and C. Leeuwenburgh. 2004. Short-term CR decreases cardiac mitochondrial oxidant production but increases carbonyl content. Am. J. Physiol. Regul. Intergr. Comp. Physiol. 286: R254–259. Juengst, E. T. and M. Fossel. 2000. The ethics of embryonic stem cells—now and forever, cells without end. J. Am. Med. Assoc. 284: 3180–3184. Juengst, E. T. 2004. Anti-aging research and the limits of medicine. Pp. x–x in S. G. Post and R. H. Binstock (eds.), The Fountain of Youth: Cultural, Scientific, and Ethical Perspectives on a Biomedical Goal. Oxford University Press, New York. Julius, M. and C. M. Lang. 1988. Blood glutathione

References

levels reflect subjective and objective health status in Southfield elderly. Gerontologist 28: 228A (abstract). Julius, M., C. A. Lang, L. Gleiberman, E. Harburg, W. DiFranceisco and A. Schork. 1994. Glutathione and morbidity in a community-based sample of elderly. J. Clin. Epidemiol. 47: 1021–1026. Kadenbach, B., E. Bender, A. Reith, A. Becker, S. Hammerschmidt, I. Lee, S. Arnold and M. Huttemann. 1999. Possible influence of metabolic activity on aging. J. Anti-Aging Med. 2: 255–264. Kadenbach, B., C. Munscher, V. Frank, J. MullerHocker and J. Naplwotzki. 1995. Human aging is associated with stochastic somatic mutations of mitochondrial DNA. Mutat. Res. 338: 161–172. Kadenbach, B., M. Huttemann, S. Arnold , I. Lee and E. Bender. 2000. Mitochondrial energy metabolism is egulated via nuclear-coded subunits of cytochrome c oxidase. Free Rad. Biol Med. 3/4: 211–221. Kaeberlein, M., M. McVey and L. Guarente. 2001. Using yeast to discover the fountain of youth. http://sageke.sciencemag.org/cgi/content/full/ sageke;2001/1/pe1. Kahn, S. E., V. G. Larson, J. C. Beard, K. C. Cain, G. W. Fellingham, R. S. Schwartz, R. C. Veith, J. R Stratton, M. D. Cerqueira and I. B. Abrass. 1990. Effect of exercise on insulin action, glucose tolerance, and insulin secretion in aging. Am. J. Physiol. 258: E937–943. Kaiser, M., M. Gasser, R. Ackermann and S. C. Stearns. 1997. P-element inserts in transgenic flies: A cautionary tale. Heredity 78: 1–11. Kakkar, R., J. S. Bains and S. P. Sharma. 1996. Effect of vitamin E on life span, malondialdehyde content and antioxidant enzymes in aging Zaprionus paravittiger. Gerontology 42: 312–321. Kale, S. P. and S. M. Jazwinski. 1996. Differential response to UV stress and DNA damage during the yeast replicative life span. Dev. Genet. 18: 154– 160. Kallman, F. J. 1957. Twin data on the genetics of aging. Pp. 131–143 in G. E. Wolstenholme and C. M. O’Connor (eds.), Methodology of the Study of Aging. Little Brown, Boston. Kamboh, M. I. 1995. Apolipoprotein E polymorphism and susceptibility to Alzheimer’s disease. Hum. Biol. 67: 195–215. Kaminer, M. S. and B. A.Gilchrest. 1995. The many faces of acne. J. Am. Acad. Dermatol. 32: S6–14. Kang, H-L., S. Benzer and K-T. Min. 2002. Life extension in Drosophila by feeding a drug. Proc. Natl. Acad. Sci. USA 99: 838–843. Kannel, W. B. and H. Hubert. 1982. Vital capacity as a biomarker of aging. Pp. 145–160 in M. E. Reff and E. L. Schneider (eds.), Biological Markers of Aging. NIH Publ. no. 82-2221. National Institutes

555

of Health, Washington, DC. Kannisto, V., J. Lauritsen, A. R. Thatcher and J. W. Vaupel. 1994. Reductions in mortality at advanced ages: Several decades of evidence from 27 countries. Popul. Devel. Rev. 4: 793–810. Kapahi, P., M. E. Boulton and T. B. L. Kirkwood. 1999. Free Rad. Biol. Med. 26: 495–500. Kapahi, P., B. M. Zid, T. Harper, D. Koslover, V. Sapin and S. Benzer. 2004. Regulation of lifespan in Drosophila by modulation of genes in the TOR signaling pathway. Curr. Biol. 14: 885–890. Kaplan, H., J. Lancaster and A. Robson. 2003. Embodied capital and the evolutionary economics of the human life span. Pp. 152–182 in J. R. Carey and S. Tuljapurkar (eds.), Life Span: Evolutionary, Ecological, and Demographic Perspectives. Population Council, New York. Kaplan, H. 1997. The evolution of the human life course. Pp. 175–211 in K. W. Wachter and C. E. Finch (eds.), Between Zeus and the Salmon: The Biodemography of Longevity. National Academy Press, Washington, DC. Kaplan, H. J. 2002. Human life histories. Pp 627–631 in M. Pagel (ed.), Encyclopedia of Evolution, vol. 2. Oxford University Press, New York. Kaplan, H. S., K. Hill, J. B. Lancaster and A. M. Hurtado. 2000. A theory of human life history evolution: diet, intelligence, and longevity. Evol. Anthropol. 9: 156–185. Kappeler, L., D. Gourdji, P. Zizzari, M. T. Bluet-Pajot and J. Epelbaum. 2003. Age-associated changes in hypothalmic and pituitary neuroendocrine gene expression in the rat. J. Neuroendocrinol. 15: 592– 601. Kappus, H. 1985. Lipid peroxidation. Pp. 152–195 in H. Sies (ed.), Oxidative Stress. Academic Press, New York. Kass, L. R. 2004 L’Chaim and its limits: Why not immortality? Pp. x–x in S. G. Post and R. H. Binstock (eds.), The Fountain of Youth: Cultural, Scientific, and Ethical Perspectives on a Biomedical Goal. Oxford University Press, New York. Katchaturian, Z. S. and T. S. Radebaugh. 1996a. Alzheimer’s Disease: Cause(s), Diagnosis, Treatment, and Care. CRC Press, Boca Raton, FL. Katchaturian, Z. S. and T. S. Radebaugh. 1996b. A synthesis of critical topics in Alzheimer’s disease. Pp. 3–14 in Z. S. Katchaturian and T. S. Radebaugh (eds.), Alzheimer’s Disease: Cause(s), Diagnosis, Treatment, and Care. CRC Press, Boca Raton, FL. Kato, S., H. Endoh, Y. Masuhiro, T. Kitamoto, S. Uchiyama, H. Sasaki, S. Masushige, Y. Gotoh, E. Nishida, H. Kawashima, D. Metzger and P. Chambon. 1995. Activation of the estrogen receptor through phosphorylation by mitogen-activated protein kinase. Science 270: 1491–1494.

556

References

Kato, H., M. Harada, K. Tsuchiya and K. Miriwaki. 1980. Absence of correlation between DNA repair in ultraviolet irradiated mammalian cells and life span of the donor species. Jpn. J. Genet. 55: 99– 108. Kator, K., V. Cristofalo, R. Charpentier and R. G. Cutler. 1985. Dysdifferentiative nature of aging: Passage number dependency of globin gene expression in normal human diploid cells grown in tissue culture. Gerontology 31: 355–361. Katz, M. L., W. G. Robison, R. K. Herrmann, A. B. Groone and J. G. Bieri. 1984. Lipofucsin accumulation resulting from senescence and vitamin E deficiency: Special properties and tissue distribution. Mech. Ageing Dev. 25: 149–159. Kay, M. M. B. 1985. Aging of cell membrane molecules leads to appearance of an aging antigen and removal of senescent cells. Gerontology 31: 215– 235. Kay, M. M., C. Cover, S. F. Schluter, R. M. Bernstein and J. J. Marchalonis. 1995. Band 3, the anion transporter, is conserved during evolution: Implications for aging and vertebrate evolution. Cell. Mol. Biol. 41: 833–842. Kay, M. M. and J. Goodman. 1997. Brain and erythrocyte anion transporter protein, band 3, as a marker for Alzheimer’s disease: Structural changes detected by electron microscopy, phosphorylation, and antibodies. Gerontology 43: 44–66. Kay, M. M., T. Wyant and J. Goodman. 1994. Autoantibodies to band 3 during aging and disease and aging interventions. Ann. N.Y. Acad. Sci. 719: 419–447. Kay, S. A. and A. J. Millar. 1995. New models in vogue for circadian clocks. Cell 83: 361–364. Kayo, T., D. B. Allison, R. Weindruch and T. A. Prolla. 2001. Influences of aging and caloric restriction on the transcriptional profile of skeletal muscle from rhesus monkeys. Proc. Natl. Acad. Sci. USA 98: 5093–5098. Keaney, M. and D. Gems. 2003. No increase in lifespan in Caenorhabiditis elegans upon treatment with the superoxide dismutase mimetic EUK-8. Free Rad. Biol. Med. 34: 277–282. Kemnitz, J. W., R. Weindruch, E. B. Roecker, K. Crawford, P. L. Kaufmann and W. B. Ershler. 1993. Dietary restriction of adult male rhesus monkeys: Findings from the first year of study. J. Gerontol. Biol. Sci. 48: B17–B26. Kennedy, B. K., N. R. Austriaco Jr., J. Zhang and L. Guarente. 1995. Mutation in the silencing gene SIR4 can delay aging in S. cerevisiae. Cell 80: 485– 496. Kenny, A. M., L. Dawson, A. Kleppinger, M. IannuzziSucich and J. O. Judge. 2003. Prevalence of sarcopenia and predictors of skeletal muscle mass in nonobese women who are long-term users of

estrogen-replacement therapy. J. Gerontol. Med. Sci. 58A: 436–440. Kenyon, C., J. Chang, E. Gensch, A. Rudner and R. Tabtiang. 1993. A C. elegans mutant that lives twice as long as wild type. Nature 366: 461–464. Kerr, J. F. R., J. Searle, B. V. Harman and C. J. Bishop. 1987. Apoptosis. Pp. 93–128 in C. S. Potten (ed.), Perspectives on Mammalian Cell Death. Oxford University Press, Oxford. Kerwin, J. M., C. M. Morris, M. Johnson, R. H. Perry and E. K. Perry. 1993. Hippocampal p75 nerve growth factor receptor immunoreactivity in development, normal aging and senescence. Acta Anatom. 147: 216–222. Keuther K and R. Arking. 1999. Drosophila selected for extended longevity are more sensitive to heat shock. Age 22: 175–180. Kevles, D. J. and L. Hood. 1992. The Code of Codes: Scientific and Social Issues in the Human Genome Project. Harvard University Press, Cambridge, MA. Khazaeli, A. A. and J. W. Curtsinger. 2000. Genetic analysis of extended lifespan in Drosophila melanogaster. III. On the relationship between artificially selected and wild stocks. Genetica 109: 245–253. Khazaeli, A. A., M. Tatar, S. D. Pletcher and J. W. Curtsinger. 1997. Heat-induced longevity extension in Drosophila. I. Heat treatment, mortality, and thermotolerance. J. Gerontol. Biol. Sci. Med. Sci. 52: B48–B52. Khodr, B., J. Howard, K. Watson and Z. Khalil. 2003. Effects of short-term and long-term antioxidant therapy on primary and secondary ageing neurovascular processes. J. Gerontol. Biol. Sci. 58A: 698–708. Kibler, K. H. and H. D. Johnson. 1961. Metabolic rate and aging in rats during exposure to cold. J. Gerontol. 16: 13–19. Kim, H. J., K. W. Kim, B. P. Yu and H. Y. Chung. 2000. The effect of age on cyclooxygenase-2 gene expression: NF-kB activation and IkBa degradation. Free Rad. Biol. Med. 28: 683–692. Kim, H. J., B. P. Yu and H. Y. Chung. 2002. Molecular exploration of age-related Nf?B/IKK downregulation by calories restriction in rat kidney. Free Rad. Biol. Med. 32: 991–1005. Kim, N. W., M. Platyszek, K. R. Prowse, C. B. Harley, M. D. West, P. L. C. Ho, G. M. Coviello, W. E. Wright, S. L. Weinrich and J. W. Shay. 1994. Specific association of human telomerase activity with immortal cells and cancer. Science 266: 2011–2015. Kim, Y. S., H. J. Nam, H. Y. Chung, N. D. Kim, J. H. Ryu, W. J. Lee, R. Arkling and M. A. Yoo. 2001. Role of xanthine dehydrogenase and aging on the innate immune response of Drosophila. J. Am. Aging Assoc. 24: 187–194.

References

Kimura, K. D., H. A. Tissenbaum, Y. Liu and G. Ruvkun. 1997. daf-2, an insulin receptor-like gene that regulates longevity and diapause in Caenorhabditis elegans. Science 277: 942–946. King, T. J. and R. Briggs. 1956. Serial transplantation of embryonic nuclei. Cold Spring Harbor Symp. Quant. Biol. 21: 271–290. King, V. and J. Tower. 1999. Aging-specific expression of Drosophila hsp22. Dev. Biol. 207: 107–118. Kirby, G. C. 1974. Greying with age: A coat-color variant in wild Australian populations of mice. J. Hered. 65: 126–128. Kirby, K., J. Hu, A. J. Hilliker and J. P. Phillips. 2002. RNA interference-mediated silencing of Sod2 in Drosophila leads to early adult-onset mortality and elevated endogenous oxidative stress. Proc. Natl. Acad. Sci. USA 99: 16162–16167. Kirchman, P. A., S. Kim, C. Y. Lai and S. M. Jazwinski. 1999. Interorganelle signaling is a determinant of longevity in Saccharomyces cerevisiae. Genetics 152: 179–190. Kirk, D. L. 1988. The ontogeny and phylogeny of cellular differentiation in Volvox. Trends Genet. 4: 32–36. Kirk, D. L. 2001. Germ-soma differentiation in Volvox. Dev. Biol. 238: 213–223. Kirk, K. L. 2001. Dietary restriction and aging: comparative tests of evolutionary hypotheses. J. Gerontol. Biol. Sci. 56A: B123–129. Kirk, M. M., K. Stark, S. M. Miller, W. Muller, B. E. Taillon, H. Gruber, R. Schmitt and D. L. Kirk. 1999. regA, a Volvox gene that plays a central role in germ-soma differentiation, encodes a novel regulatory protein. Development 126: 639–647. Kirkwood, T. B. L. 1985. Comparative and evolutionary aspects of longevity. Pp. 27–44 in C. E. Finch and E. L. Schneider (eds.), Handbook of the Biology of Aging, 2nd ed. Van Nostrand Reinhold, New York. Kirkwood, T. B. L. 1987. Immortality of the germ-line versus disposability of the soma. Pp. 209–218 in A. D. Woodhead and K. H. Thompson (eds.), Evolution of Longevity in Animals: A Comparative Approach. Plenum Press, New York. Kitado, H., K. Higuchi and T. Takeda. 1994a. Molecular genetic characterization of the senescenceaccelerated mouse (SAM) strains. J. Gerontol. Biol. Sci. 49: B247–B254. Kitado, H., K. Higuchi and T. Takeda. 1994b. Response to guest editorial. J. Gerontol. Biol. Sci. 49: B245–B246. Klarsfeld, A and F Revah. 2004. The Biology of Death: Origins of Mortality (trans. L. Brady). Cornell University Press, Ithaca, NY. Klass, M. 1977. Aging in the nematode Caenorhabditis elegans: major biological and environmental factors influencing life-span. Mech. Ageing Dev. 6: 413– 429.

557

Klebba-Goodman, A. A. 1986. Osteoporosis. Unpublished essay, Wayne State University, Detroit, MI. Klichko, V. I., S. V. Radyuk and W. C. Orr. 1999. CuZnSOD promoter-driven expression in the Drosophila central nervous system. Neurobiol. Aging 20: 557–543. Kligman, A. M., G. L. Grove and A. K. Balin. 1985. Aging of human skin. Pp. 820–841 in C. E. Finch and E. L. Schneider (eds.), Handbook of the Biology of Aging, 2nd ed. Van Nostrand Reinhold, New York. Klionsky, D. J. 2004. Regulated self-cannibalism. Nature 431: 31–32. Klocke, R. A. 1977. Influence of aging on the lung. Pp. 432–444 in C. E. Finch and L. Hayflick (eds.), Handbook of the Biology of Aging. Van Nostrand Reinhold, New York. Knauf, F., B. Rogina, Z. Jiang P. S. Aronson and S. L. Helfand. 2002. Functional characterization and immunolocalization of the transporter encoded by the the life extending gene Indy. Proc. Natl. Acad. Sci. USA 99: 14312–14319. Koenig, H. L., M. Schumacker, B. Ferzaz, A. N. D. Thi, A. Ressouches, R. Guennoun, I. Jung-Testas, R. Robel, Y. Akwa and E-E. Baulieu. 1995. Progesterone synthesis and myelin formation by Schwann cells. Science 268: 1500–1503. Kohn, R. R. 1977. Heart and cardiovascular system. Pp. 281–317 in C. E. Finch and L. Hayflick (eds.), Handbook of the Biology of Aging. Van Nostrand Reinhold, New York. Kohn, R. R. 1978. Principles of Mammalian Aging, 2nd ed. Prentice-Hall, Englewood Cliffs, NJ. Kolodner, R. D., C. D. Putnam and K. Myung. 2002. Maintenance of genome stability in Saccharomyces cerevisiae. Science 297: 552–558. Kolta, M. G., R. Holson, P. Duffy and R. W. Hart. 1989. Effect of long-term caloric restriction on brain monoamines in aging male and female Fischer 344 rats. Mech. Ageing Dev. 48: 191– 198. Korenman, S. 1982. Introduction. Pp. 1–8 in L. V. Avidi and S. G. Korenman (eds.), Endocrine Aspects of Aging. Elsevier, New York. Kormondy, E. J. 1969. Concepts of Ecology. PrenticeHall, Englewood Cliffs, NJ. Korpelainen, H. 2000. Fitness, reproduction and longevity among European aristocratic and rural Finnish families in the 1700s and 1800s. Proc. R. Soc. Lond. B Biol. Sci. 267: 1765–1770. Kowald, A. and T. B. Kirkwood. 1994. Towards a network theory of ageing: A model combining the free radical theory and the protein error theory. J. Theoret. Biol. 168: 75–94. Kowald, A. and T. B. Kirkwood. 1996. A network theory of ageing: The interactions of defective mitochondria, aberrant proteins, free radicals and

558

References

scavengers in the ageing process. Mutat. Res. 316: 209–236. Kowald, A. and T. B. L. Kirkwood. 1999. Modeling the role of mitochondrial mutations in cellular aging. J. Anti-Aging Med. 2: 243–244. Koward, A. 1999. The mitochondrial theory of aging: Do damaged mitochondria accumulate by delayed degradation? Exp. Gerontol. 34: 605–612. Krabbe, K. S., M. Pedersen and H. Bruunsgaard. 2004. Inflammatory mediators in the elderly. Exp. Gerontol. 39: 687–699. Kraehenbuhl, J.-P. and M. Corbett. 2004. Keeping the gut microflora at bay. Science 303: 1624–1625. Kramer, J. M., J. T. Davidge, J. M. Lockyer and B. E. Staveley. 2003. Expression of Drosophila FOXO regulates growth and can phenocopy starvation. BMC Dev. Biol. 3: 5. Krause, F., C. Q. Scheckhuber, A. Werner, S. Rexroth, N. H. Reifschneider, N. A. Dencher and H. D. Osiewacz. 2004. Supramolecular organization of cytochrome c oxidase- and alternative oxidasedependent respiratory chains in the filamentous fungus Podospora anserina. J. Biol. Chem. 27: 26453–26461. Krauss, S., C-Y. Zhang and B. B. Lowell. 2002. A significant portion of mitochondrial proton leak in intact thymocytes depends on expression of UCP2. Proc. Natl. Acad. Sci. USA 99: 118–122. Kreil, G. 1994. Conversion of L- to D-amino acid: A post-translational reaction. Science 266: 996– 997. Kristal, B. S. and B. P. Yu. 1992. An emerging hypothesis: Synergistic induction of aging by free radicals and Millard reactions. J. Gerontol. Biol. Sci. 47: B107–B114. Kristal, B. S. and B. P. Yu. 2000. Oxidant-mediated repression of mitochondrial DNA transcription. Pp. x–x in Antioxidant and Redox Regulation of Genes. Academic Press, San Diego, CA. Krohn, P. L. 1966. Transplantation and aging. Pp. 125– 138 in P. L. Krohn (ed.), Topics of the Biology of Aging. Wiley, New York. Kruk, P. A., A. S. Balajee, K. S. Rao and V. A. Bohr. 1996. Telomere reduction and telomerase inactivation during neuronal cell differentiation. BBRC 224: 487–492. Kuck, U., H. D. Osiewacz, U. Schmidt, B. Kappelhoff, E. Schulte, U. Stahl and K. Esser. 1985. The onset of senescence is affected by DNA rearrangements of a discontinuous mitochondrial gene in Podospora anserina. Curr. Genet. 9: 373–382. Kuether, K. and Arking, R. 1999. Drosophila selected for extended longevity are more senesitive to heat shock. Age 22: 175–180. Kumaran, S., S. Savitha, M. A. Devi and C. Panneerselvam. 2004. L-carnitine and DL-a-lipoic acid reverse the age related deficit in glutathione redox

state in skeletal muscle and heart tissues. Mech. Aging Dev. 125: 507–512. Kurapati, R., H. B. Passananti, M. R. Rose and J. Tower. 2000. Increased hsp22 RNA levels in Drosophila lines genetically selected for increased longevity. J. Gerontol. Biol. Sci. Med. Sci. 55: B552–B559. Kurland, C. G. and R. Mikkola. 1993. The impact of nutritional state on the microevolution of ribosomes. Pp. 225–238 in S. Kjelleberg (ed.), Starvation in Bacteria. Plenum Press, New York. Kurz, D. J., S. Decary, Y. Hong, E. Trivier, A. Akhmedov and J. D. Erusalimsky. 2004. Chronic oxidative stress comprises telomere integrity and accelerates the onset of senescence in human endothelial cells. J. Cell Sci. 117: 2417–2426. Kuzmin, E. V., O. V. Karpova, T. E. Elthon and K. J. Newton. 2004. Mitochondrial respiratory deficiencies signal up-regulation of genes for heat shock proteins. J. Biol. Chem., March 11, ms M400640200. Lack, D. 1943. The age of some more British birds. Br. Birds 36: 193–197. Laganiere, S. and B. P. Yu. 1993. Modulation of membrane phospholipid fatty acid composition by food, age, and food restriction. Gerontology 39: 7–18. Lai, C-Y., E. Jaruga, C. Borghouts and S. M. Jazwinski. 2002. A mutation in the ATP2 gene abrogates the age asymmetry between mother and daughter cells of the yeast Saccharomyces cerevisiae. Genetics 162: 73–87. Lakatta, E. G. 1985. Heart and circulation. Pp. 377– 413 in C. E. Finch and E. L. Schneider (eds.), Handbook of the Biology of Aging, 2nd ed. Van Nostrand Reinhold, New York. Lakatta, E. G. 2003. Arterial and cardiac aging: major shareholders in cardiovascular disease enterprises; part III: cellular and molecular clues to heart and arterial aging. Circulation 107: 490–497. Lakowski, B. and S. Hekimi. 1996. Determination of life-span in Caenorhabditis elegans by four clock genes. Science 272: 1010–1013. Lal, S. B., J. J. Ramsey, S. Monemdjou, R. Weindruch and M. E. Harper. 2001. Effects of caloric restriction on skeletal muscle mitochondrial proton leak in aging rats. J. Gerontol. Biol Sci 56: B116–B122. La Marche, V. C. 1969. Environment in relation to age of bristlecone pines. Ecology 50: 53–59. Lamb, M. J. 1977. Biology of Ageing. Halsted (Wiley), New York. Lamb, M. J. 1978. Ageing. Pp. 43–104 in M. Ashburner and T. R. F. Wright (eds.), The Genetics and Biology of Drosophila, vol. 2c. Academic Press, London. Lambert, A. J. and B. J. Merry. 2003. Effect of caloric restriction on mitochondrial reactive oxygen species production and bioenergetics: Reversal by insulin. Am. J. Physiol. Regul. Integr.

References

Comp. Physiol., Sept. 11. 10.1152/ajpregu.00341 .2003. Lambert, A. J., B. Wang, J. Yardley, J. Edwards and B. J. Merry. 2004. The effect of aging and caloric restriction on mitochondrial protein density and oxygen consumption. Exp. Gerontol. 39: 289– 295. Lammer, E. J., D. T. Chen, R. M. Hoar, N. D. Agnish, P. J. Benke, J. T. Braun, C. J. Curry, P. M. Fernhoff, A. W. Grix, I. T. Lott, et al. 1985. Retinoic acid embryopathy. New England J. Med. 313: 837–841. Landfield, P. W., S. K. Sundberg, M. Smith, J. C. Eldridge and M. Morris. 1980. Mammalian aging: Theoretical implications of changes in brain and endocrine systems during mid and late life. Peptides 1(suppl. 1): 185–196. Lane, M. A., D. J. Baer, E. M. Tilmont, W. V. Rumpler, D. K. Ingram, G. S. Roth and R. G. Cutler. 1995. Energy balance in Rhesus monkeys (Macaca mulatta) subjected to long-term dietary restriction. J. Gerontol. Biol. Sci. 50A: B295– B302. Lansing, A. I. 1947. A transmissible, cumulative and reversible factor in aging. J. Gerontol. 2: 228–239. Lansing, A. I. 1954. A nongenic factor in the longevity of rotifers. Ann. N.Y. Acad. Sci. 57: 455–464. Larsen, P. L. 1993. Aging and resistance to oxidative damage in Caenorhabditis elegans. Proc. Natl. Acad. Sci. USA 90: 8905–8909. Larsen, P. L. and C. F. Clarke. 2002. Extension of lifespan in Caenorhabditis elegans by diet lacking coenzyme Q. Science 295: 534–540. Larsen, P., P. S. Albert and D. L. Riddle. 1995. Genes that regulate both development and longevity in Caenorhabditis elegans. Genetics 139: 1567–1583. Laughlin, S. B. and T. J. Sejnowski. 2003. Communication in neural networks. Science 301: 1870– 1873. Laun P., A. Pichova, F. Madeo, J. Fuchs, A. Ellinger, S. Kohlwein, I. Dawes, K. U. Frohlich and M. Breitenbach. 2001. Aged mother cells of Saccharomyces cerevisiae show markers of oxidative stress and apoptosis. Mol. Microbiol. 39: 1166–1173. Lavie, L., A. Reznick and D. Gershon. 1982. Decreased protein and puromycinil-peptide degradation in livers of senescent mice. Biochem. J. 202: 47–51. Lawrence, P. A. 1992. The Making of a Fly: The Genetics of Animal Design. Blackwell, Oxford. Lazzaro, B. P., B. K. Sceurman and A. G. Clark. 2004. Genetic basis of natural variation in D. melanogaster antibacterial immunity. Science 303: 1873– 1876. Le Bourg, E., B. Thon, J. Legare, B. Desjardins and H. Charbonneau. 1993. Exp. Gerontol. 28: 217– 223. Le Bourg, E., P. Valenti, P. Lucchetta and Payre, F. 2001. Effects of mild heat shocks at young age on

559

aging and longevity in Drosophila melanogaster. Biogerontology 2: 155–164. Leaf, A. 1984. Long-lived populations (extreme old age). Pp. 82–86 in R. Andres, E. L. Bierman and W. R. Hazzard (eds.), Principles of Geriatric Medicine. McGraw-Hill, New York. Leakey, J. A., H. C. Cunny, J. Bazare, Jr., P. J. Webb, J. C. Lipscomb, W. Slikker, Jr., R. J. Feuers, P. H. Duffy and R. W. Hart. 1989. Effects of aging and caloric restriction on hepatic drug metabolizing enzymes in the Fischer 344 rat. II: Effects on conjugating enzymes. Mech. Ageing Dev. 48: 157–166. Lebo, C. P. and R. C. Reddell. 1972. The prebycusis component in occupational hearing loss. Laryngoscope 82: 1399–1409. Lee, C-K., D. B. Allison, J. Brand, R. Weindruch and T. A. Prolla. 2002. Transcriptional profiles associated with aging and middle age-onset caloric restriction in mouse hearts. Proc. Natl. Acad. Sci. USA 99: 14988–14993. Lee, C-K., R. G. Kloop, R. Weindruch and T. A. Prolla. 1999. Gene expression profile of aging and its retardation by caloric restriction. Science 285: 1390– 1393. Lee, C-K., R. Weindruch and T. A. Prolla. 2000. Geneexpression profile of the ageing brain in mice. Nature Genet. 25: 294–297. Lee, D. W. and B. P. Yu. 1990. Modulation of free radicals and superoxide dismutases by age and dietary restriction. Aging 2: 357–362. Lee, M. H., B. Ahn, I. S. Choi and H. S. Koo. 2002. The gene expression and deficiency phenotypes of Cockayne syndrome B protein in Caenorhabditis elegans. FEBS Lett. 522: 47–51. Lee, S. S., S. Kennedy, A. C. Tolonen and G. Ruvkun. 2003. DAF-16 target genes that control C. elegans life-span and metabolism. Science 300: 644–647. Leeson, C. F. and T. S. Leeson. 1967. Histology. Saunders, Philadelphia. Leeson, C. F. and T. S. Leeson. 1970. Histology, 2nd ed. Saunders, Philadelphia. Leffelaar, D. and T. A. Grigliatti. 1984. A mutation in Drosophila that appears to accelerate aging. Dev. Genet. 4: 199–210. Leiers, B., A. Kampkotter, C. G. Grevelding, C. D. Link, T. E. Johnson and K. Henkle-Duhrsen. 2003. A stress responsive glutathione-S-transferase confers resistance to oxidative stress in Caenorhabiditis elegans. Free Rad. Biol. Med. 34: 1405–1415. Lemaire, V., M. Koehl, M. Le Moal and D. N. Abrous. 2000. Prenatal stress produces learning deficits associated with an inhibition of neurogenesis in the hippocampus. Proc. Natl. Acad. Sci. USA 97: 11032–11037. Lemos, B., C. D. Meiklejohn and D. L. Hartl. 2004. Regulatory evolution across the protein interaction network. Nature Genet. 36: 1059–1060.

560

References

Lenton, K. J. and C. L. Greenstock. 1999. Ability of human plasma to protect against ionising radiation is inversely correlated with age. Mech. Ageing Dev. 107: 15–20. Lerner, S. P. and C. E. Finch. 1991. The major histocompatibility complex and reproductive functions. Endocrine Rev. 12: 78–90. Lesher, S., R. J. M. Fry and H. I. Kohn. 1961. Age and the generation time of the mouse duodenal epithelial cells. Exp. Cell Res. 24: 334–343. Leslie, M. 2004. The Goldilocks Genes. Sci. Aging Knowl. Environ. 2004: 43. Lewin, R. 1988. Disappointing brain graft results. Science 240: 1407. Lewis, R. 2004. Drs. Atkins and Agatson, you were so right. The Scientist, May 24, p. 64. Lewis, V. M., J. J. Twoney, P. Beatmar, G. Goldstein and R. A. Good. 1978. Age, thymic involution and circulating thymic hormone activity. J. Clin. Endocrinol. Metab. 47: 145–150. Lewontin, R. 2000. It Ain’t Necessarily So: The Dream of the Human Genome and Other Illusions. New York Review Books, New York. Li, J. and N. J. Holbrook. 2003. Common mechanisms for declines in oxidative stress tolerance and proliferatin with aging. Free Rad. Biol. Med. 35: 292–299. Li, M., C. Li and W. S. Parkhouse. 2003. Age-related differences in the des IGF-1-mediated activation of Akt-1 and p70 S6K in mouse skeletal muscle. Mech. Ageing Develop. 124: 771–778. Liang, H., E. J. Masoro, J. F. Nelson, R. Strong, C. A. McMahan and A. Richardson. 2003. Genetic mouse models of extended lifespan. Exp. Gerontol. 38: 1353–1364. Liao, X. and R. A. Butow. 1993. RTG1 and RTG2: Two yeast genes required for a novel path of communication from mitochondria to the nucleus. Cell 72: 61–71. Lichtig, C., J. Levy, D. Gershon and A. Z. Reznick. 1987. Effect of aging and exercise on the kidney. Gerontology 33: 40–48. Lieb, J. and T. F. C. Mackay. 2002. The complex genetic architecture of Drosophila melanogaster. Exp. Aging Res 28: 361–390. Lifton, R. P., A. G. Gharavi and D. S. Geller. 2001. Molecular mechanisms of human hypertension. Cell 104: 545–556. Lightowers, R. N., H.T. Jacobs and O. A. Kajander. 1999. Mitochondrial DNA—all things bad? Trends Genet. 15: 91–93. Lin, K., J. B. Dorman, A. Rodan and C. Kenyon. 1997. Daf-16: An HNF-2/forkhead family member that can function to double the life-span of Caenorhabditis elegans. Science 278: 1319–1322. Lin, S-J., P-A. Defossez and L. Guarente. 2000. Requirement of NAD and SIR2 for life-span exten-

sion by caloric restriction in Saccharomyces cerevisiae. Science 289: 2126–2128. Lin, S-J., M. Kaeberfein. A. A. Andalis, L. A. Strutz, P-A. Defossez, V. C. Culotta, G. R. Fink and L. Guarente. 2002. Caloric restriction extends Saccharomyces cerevisiae lifespan by increasing respiration. Nature 418: 344–348. Lin, Y-J., L. Seroude and S. Benzer. 1998. Extended life-span and stress resistance in the Drosophila mutant methuselah. Science 282: 943–946. Lindeman, R. D., J. D. Tobin and N. W. Shock. 1985. Longitudinal studies on the rate of decline in renal function with age. J. Am. Geriatr. Soc. 33: 278– 285. Linnen, C., M. Tatar and D. Promislow. 2001. Cultural artifacts: a comparison of senescence in natural, laboratory-adapted and artificially selected lines of Drosophila melanogaster. Evol. Ecol. Res. 3: 877–888. Linton, M. F., J. B. Atkinson and S. Fazio. 1995. Prevention of atherosclerosis in apolipoprotein E-deficient mice by bone marrow transplantation. Science 267: 1034–1038. Lints, F. A. 1978. Genetics and Aging. Interdisciplinary Topics in Gerontology, vol. 14. S. Karger, Basel, Switzerland. Lints, F. A. 1988. Parental age effects. Pp. 176–185 in F. A. Lints and M. H. Soliman (eds.), Drosophila as a Model Organism for Ageing Studies. Blackie, Glasgow. Lints, F. A. 1989. The rate of living theory revisited. Exp. Gerontol. 35: 36–57. Lints, F. A. and C. V. Lints. 1971. Influence of preimaginal environment on fecundity and ageing in Drosophila melanogaster hybrids. II. Developmental speed and life span. Exp. Gerontol. 6: 427–445. Lipman, J. M., A. Turturro and R. W. Hart. 1989. The influence of dietary restriction on DNA repair in rodents: A preliminary study. Mech. Ageing Dev. 48: 135–143. Lippman, R. D. 1983. Lipid peroxidation and metabolism in aging. Pp. 315–342 in M. Rothstein (ed.), Review of Biological Research in Aging, vol. 1. Alan R. Liss, New York. Lipsitz, L. 2002. Dynamics of stability: the physiologic basis of functional health and stability. J. Gerontol. Biol. Sci. 57A: B115. Lissner, L., P. M. Odell, R. B. D’Agostino, J. Stokes III, B. E. Kreger, A. J. Belanger and K. D. Brownell. 1991. N. Engl. J. Med. 324: 1839–1844. Lithgow, G. J. and T. B. Kirkwood. 1996. Mechanisms and evolution of aging. Science 273: 80. Lithgow, G. J., T. M. White, S. Melov and T. E. Johnson. 1995. Thermotolerance and extended life-span conferred by single-gene mutations and induced by thermal stress. Proc. Natl. Acad. Sci. USA 92: 7540–7544.

References

Liu, J., E. Head, A. M. Gharib, W. Yuan, R. T. Ingersoll, T. M. Hagen, C. W. Cotman and B. N. Ames. 2002. Memory loss in old rats is associated with brain mitochondrial decay and RNA/DNA oxidation: Partial reversal by feeding acetyl-Lcarnitine and/or R-a-lipoic acid. Proc. Natl. Acad. Sci. USA 99: 2356–2361. Liu, R., I. Y. Liu, X. Bi, R. F. Thompson, S. R. Doctrow, B. Malfroy and M. Baudry. 2003. Reversal of age-related learning deficits and brain oxidative stress in mice with superoxide dismutase/catalase mimetics. Proc. Natl. Acad. Sci. USA. Liu, Z. and R. A. Butow. 1999. A transcriptional switch in the expression of yeast tricarboxylic acid cycle genes in response to a reduction or loss of respiratory function. Mol. Cell. Biol. 19: 6720–6728. Lockshin, R. A. and Z. Zakeri-Milovanovic. 1984. Nucleic acids in cell death. Pp. 243–268 in I. Davies and D. C. Siegel (eds.), Cell Ageing and Cell Death, I. Cambridge University Press, Cambridge. Loeb, J. and J. H. Northrop. 1917. On the influence of food and temperatures on the duration of life. J. Biol. Chem. 32: 103–121. Longini, I. M., M. W. Higgins, P. C. Hinton, P. C. Moll and J. B. Keller. 1984. Environmental and genetic sources of familial aggregation of blood pressure in Tecumseh, Michigan. Am. J. Epidemiol. 120: 131–144. Longo, V. D. and C. E. Finch. 2003. Evolutionary medicine: from dwarf model systems to healthy centenarians? Science 299: 1342–1346. Longo, V. D., E. B., Gralla and J. S. Valentine. 1996. Superoxide dismutase activity is essential for stationary phase survival in Saccharomyces cerevisiae: mitochondrial production of toxic oxygen species in vivo. J. Biol. Chem. 271: 12275–12280. Lonn, E., J. Bosch, S. Yusuf, P. Sheridan, J. Pogue, J. M. Arnold, C. Ross, A. Arnold, P. Sleight, J. Probstfield, G. R. Dagenais and the HOPE and HOPE-TOO Trial Investigators. 2005. Effects of long-term vitamin E supplementation on cardiovascular events and cancer: a randomized controlled trial. JAMA 293: 1338–1347. Loomis, W. F. and P. W. Sternberg. 1995. Genetic networks. Science 269: 649. Lopez-Torres, M., R. Gredilla, A. Sanz and G. Barja. 2002. Influence of aging and long-term caloric restriction on oxygen radical generation and oxidative DNA damage in rat liver mitochondria. Free Rad. Biol. Med. 32: 882–889. Luckinbill, L. S., R. Arking, M. J. Clare, W. C. Cirocco and S. A. Buck. 1984. Selection for delayed senescence in Drosophila melanogaster. Evolution 38: 996–1003. Luckinbill, L. S., M. J. Clare, W. L. Krell, W. C. Cirocco and P. Richards. 1987. Estimating the

561

number of genetic elements that defer senescence in Drosophila. Evol. Ecol. 1: 37–46. Luckinbill, L. S., J. L. Graves, A. H. Reed and S. Koetsawang. 1988. Localizing genes that defer senescence in Drosophila melanogaster. Heredity 60: 367–374. Luckinbill, L. S., T. A. Grudzien, S. Rhine and G. Weisman. 1989. The genetic basis of adaptation to selection for longevity in Drosophila melanogaster. Evol. Ecol. 3: 31–39. Luhtala, T. A., E. B. Roecker, T. Pugh, R. J. Feuers and R. Weindruch. 1994. Dietary restriction attenuates age-related increases in rat skeletal muscle antioxidant activities. J. Gerontol. Biol. Sci. 49: B231–B238. Lumpkin, C. K., J. K. McClung, O. M. Pereira-Smith and J. R. Smith. 1986. Existence of high abundance antiproliferative mRNAs in senescent human diploid fibroblasts. Science 232: 393–395. Lundblad, V. and J. W. Szostak. 1989. A mutant with a defect in telomere elongation leads to senescence in yeast. Cell 57: 633–643. Luscombe, N. M., M. M. Babu, H. Yu, M. Snyder, S. A. Teichmann and M. Gerstein. 2004. Genomic analysis of regulatory network dynamics reveals large topological changes. Nature 431: 308–312. MacArthur, R. H. and E. O. Wilson. 1967. The Theory of Island Biogeography. Princeton University Press, Princeton, NJ. Ma, Y. X., Y. Zhu, Z. S. Wang, C. F. Wang, S. Y. Chen, M. T. Zhao, G. L. Zhang, S. Q. Zheng, J. G. Zhang, Q. Gu and L. He. 1997. HLA and longevity or aging among Shanghai Chinese. Mech. Ageing Dev. 94: 191–198. Mackay, W. J. and G. C. Bewley. 1989. The genetics of catalase in Drosophila melanogaster: Isolation and characterization of acatalesemic mutants. Genetics 122: 643–652. Maclean MJ, R Aamodt, N Harris, I alseth, E Seeberg, M Bjoras and PW Piper. 2003. Base excision repair activities required for yeast to attain a full chronological life span. Aging Cell 2: 93–104. Maeda, H., C. A. Gleiser, E. J. Masoro, I. Murata, C. A. McMahan and B. P. Yu. 1985. Nutritional influences on aging of Fischer 344 rats: II. Pathology. J. Gerontol. 40: 671–688. Magalhaes, J. M. and O. Toussaint. 2004. FEBS Lett. Magwere, T., T. Chapman and L. Partridge. 2004. Sex differences in the effect of dietary restriction on life span and mortality rates in female and male Drosophila melanogaster. J. Gerontol. Biol. Sci. 59A: 3–9. Mair, W., P. Gomeyer, S. D. Pletcher and L. Partridge. 2003. Demography of dietary restriction and death in Drosophila. Science 301: 1731–1733. Mair, W., C. M. Sgro, A. P. Johnson, T. Chapman and L. Partridge. 2004. Lifespan extension by dietary

562

References

restriction in female Drosophila melanogaster is not caused by a reduction in vitellogenesis or ovarian activity. Exp. Gerontol. 39: 1011–1019. Mak, H. Y. and G. Ruvkun. 2004. Intercellular signaling of reproductive development by the C. elegans DAF-9 cytochrome P450. Development 131: 1777–1786. Makrides, S. C. 1983. Protein synthesis and degradation. Biol. Rev. 58: 344–422. Mandavilli, B. S., J. H. Santos and B. Van Houten. 2002. Mitochondrial DNA repair and aging. Mutat. Research 509: 127–151. Manson, J. E., W. C. Willet, M. J. Stampfer, G. A. Colditz, D. J. Hunter, S. E. Hakinson, C. H. Hennekens and F. E. Speizer. 1995. Body weight and mortality among women. N. Engl. J. Med. 333: 677–685. Manton, K. G. 1999. Dynamic paradigms for human mortality and aging. J. Gerontol. Biol.Sci 54A: B247–254. Manton, K. 2000. Commentary on “Mortality oscillations induced by periodic starvation alter sexmortality differential in Mediterranean fruit flies.” J. Gerontol. Biol. Sci. 55A: B54. Manton, K. G., L. Corder and E. Stallard. 1997. Chronic disability trends in elderly United States populations: 1982–1994. Proc. Natl. Acad. Sci. USA 94: 2593–2598. Manton, K. G. and E. Stallard. 1996. Longevity in the United States: Age and sex-specific evidence on life span limits from mortality patterns 1960– 1990. J. Gerontol. Biol. Sci. 51A: B362–B375. Manton, K. G., E. Stallard and L. Corder. 1995. Changes in morbidity and chronic disability in the U.S. elderly population: Evidence from the 1982, 1984, and 1989 National Long Term Care Surveys. J. Gerontol. Psychol. Sci. 50B: S194–S204. Manton, K. G., E. Stallard, M. A. Woodbury and J. E. Dowd. 1994. Time varying covariates in models of human mortality and aging: Multidimensional generalizations of the Gompertz. J. Gerontol. Biol. Sci. 49: B169–B190. Manton, K. G. and J. W. Vaupel. 1995. Survival after the age of 80 in the United States, Sweden, France, England, and Japan. N. Engl. J. Med. 333: 1232– 1235. Manton, K. G., M. A. Woodbury and E. Stallard. 1995b. Sex differences in human mortality and aging at late ages: The effect of mortality selection and state dynamics. Gerontologist 35: 597–608. Markowsha, A. L. and S. J. Breckler, 1999. Behavioral biomarkers of aging: Illustration of a multivariate approach for detecting age-related behavioral changes. J. Gerontol. Biol. Sci. 54A: B549–B566. Marshall, E. 2004. Getting the noise out of gene arrays. Science 306: 630–631. Martin, G. M. 1978. Genetic syndromes in man with

potential relevance to the pathobiology of aging. Pp. 5–39 in D. Bergsma and D. E. Harrison (eds.), Genetic Effects on Aging. Alan R. Liss, New York. Martin, G. M. 1992. Clonal attenuation and cell senescence: The next 30 years. Exp. Gerontol. 27: 455–459. Martin, G. M. 2002. Gene action in the aging brain: an evolutionary biological perspective. Neurobiol. Aging 23: 647–654. Martin, G. M., S. N. Austad and T. E. Johnson. 1996. Genetic analysis of ageing: Role of oxidative damage and environmental stresses. Nature Genet. 13: 25–34. Martin, G. M., M. Bergener, C. R. Harrington, J. W. Heinecke, H. Hoehm, R. A. Miller, R. Stocker and C. M. Wischik. 1995. Group report: Do common underlying mechanisms of aging contribute to the pathogenesis of major geriatric disorders? (Dahlem Workshop report). Pp. 281–292 in K. Esser and G. M. Martin (eds.), Molecular Aspects of Aging. Wiley, Chichester, England. Martin, G., M. Fry and L. A. Loeb. 1985. Somatic mutation and aging in mammalian cells. Pp. 7–21 in R. S. Sohal, L. S. Birnbaum and R. G. Cutler (eds.), Molecular Biology of Aging: Gene Stability and Gene Expression. Raven Press, New York. Martin, G. M., S. M. Gartler, C. J. Epstein and A. G. Motulsky. 1965. Diminished lifespan of cultured cells in Werner’s syndrome. Fed. Proc. 24: 678. Martin, G. M. and J. Oshima. 2000. Lessons from human progeroid syndromes. Nature 408: 263– 266. Martin, G. M., C. A. Sprague and C. J. Epstein. 1970. Replicative life-span of cultivated human cells: Effects of donor’s age, tissue and genotype. Lab. Invest. 23: 867–892. Martin, G. M. and M. S. Turker. 1988. Model systems for the genetic analysis of mechanisms of aging. J. Gerontol. Biol. Sci. 43: B33–B39. Martin, J. A., A. J. Klingelhutz, F. Moussavi-Harami and J. A. Buckwalter. 2004. Effects of oxidative damage and telomerase activity on human articular cartilage chrondrocyte senescence. J. Gerontol. Biol. Sci. 59A: 324–337. Martin, L. J., MC Maheny, A. M. Bronikowski, K. D. Carey, B. Dyke and A. G. Comuzzie. 2002. Lifespan in captive baboons is heritable. Mech. Ageing Dev. 123: 1461–1467. Martindale, J. L. and N. J. Holbrook. 2002. Cellular response to oxidative stress: Signaling for suicide and survival. J. Cell. Physiol. 192: 1–15. Martinez, D. E. and J. S. Leviton. 1992. Asexual metazoans undergo senescence. Proc. Natl. Acad. Sci. USA 89: 9921–9923. Marzban, G., J. Grillari, E. Reisinger, T. Hemetsberger, R. Grabherr and T. Katinger. 2002. Age-related alterations in the protein expression profile of

References

C57BL/6J pituitaries. Exp. Gerontol. 37: 1451– 1460. Mason, J. and H. Biessmann. 1995. The unusual telomeres of Drosophila. Trends Genet. 11: 58–63. Masoro, E. J. 1985. Metabolism. Pp. 540–563 in C. E. Finch and E. L. Schneider (eds.), Handbook of the Biology of Aging, 2nd ed. Van Nostrand Reinhold, New York. Masoro, E. J. 1988a. Food restriction in rodents: An evaluation of its role in the study of aging. J. Gerontol. Biol. Sci. 43: B59–B64. Masoro, E. J. 1988b. Physiological system markers of aging. Exp. Gerontol. 23: 391–394. Masoro, E. J. 1992a. Aging and proliferative homeostasis: Modulation by food restriction in rodents. Lab. Animal Sci. 42: 132–137. Masoro, E. J. 1992b. Potential role of the modulation of fuel use in the antiaging action of dietary restriction. Ann. N.Y. Acad. Sci. 663: 403–411. Masoro, E. J. 1995a. Aging: Current concepts. Pp. 3– 21 in E. J. Masoro (ed.), Handbook of Physiology. Section 11: Aging. Oxford University Press, New York. Masoro, E. J. 1995b. Dietary restriction. Exp. Gerontol. 30: 291–298. Masoro, E. J. 1998. Hormesis and the antiaging action of dietary restriction. Exp. Gerontol. 33: 61–66. Masoro, E. J. 1999. Choice of rodent model for aging research. Pp. 237–248 in B. P. Yu (ed.), Methods in Aging Research. CRC Press, Boca Raton, FL. Masoro, E. J. 2002. Caloric Restriction: A Key to Understanding and Modulating Aging, vol. 1, Research Profiles in Aging (J. Vijg, ed.). Elsevier, Amsterdam. Masoro, E. J. 2004. Masoro, E. J. and S. N. Austad. 1996. The evolution of the antiaging action of dietary restriction: A hypothesis. J. Gerontol.: Biol. Sci. 51A: B387– B391. Masoro, E. J., M. S. Katz and C. A. McMahan. 1989. Evidence for the glycation hypothesis of aging from the food-restricted rodent model. J. Gerontol. Biol. Sci. 44: B20–B22. Mastermak, M. M., K. A. Al-Regaiey, M. S. Bonkowski, J. A. Panici and A. Bartke. 2005. Effect of every-other-day feeding diet on gene expression in normal and in long-lived Ames dwarf mice. Exp. Gerontol. 40: 491–497. Mastermak, M. M., K. Al-Regaiey, M. S. Bonkowski, J. Panici, L. Sun, J. Wang, G. Z. Przybylski and A. Bartke. 2004. Divergent effects of caloric restriction on gene expression in normal and longlived mice. J. Gerontol. Biol. Sci. 59A: 784–788. Masters, W. H. and V. E. Johnson. 1966. Human Sexual Response. Little, Brown, Boston. Matrovic, V., K. Kostial, I. Simorivoc, R. Buzina, A. Brodarec and B. E. Nordin. 1979. Bone status and

563

fracture rates in two regions of Yugoslavia. Am. J. Clin. Nutr. 35: 540–549. Mattison, J. A., M. A. Lane, G. S. Roth and D. K. Ingram. 2003. Caloric restriction in rhesus monkeys. Exp. Gerontol. 38: 35–46. Mattison, J. A., G. S. Roth, D. K. Ingram and M. A. Lane. 2001. Endocrine effects of dietary restriction and aging: The National Institute on Aging study. J. Anti-Aging Med. 3: 215–223. Mattson, M. P., W. Duan, J. Lee, Z. Guo, G. S. Roth, D. K. Ingram and M. A. Lane. 2001a. Progress in the development of caloric restriction mimetic dietary supplements. J. Anti-Aging Med. 3: 225–232. Mattson, M. P., W. Duan, J. Lee and Z. Guo. 2001b. Suppression of brain aging and neurodegenerative disorders by dietary restriction and environmental enrichment: Molecular mechanisms. Mech. Ageing Dev. 122: 757–778. Mayer, P. J. 1991. Inheritance of longevity evinces no secular trend among members of six New England families born 1650–1874. Am. J. Hum. Biol. 3: 49–58. Maynard, S. J. 1962. Review lectures on senescence. I. The causes of ageing. Proc. R. Soc. Lond. B 157: 115–127. Maynard Smith, J. 1988. The evolution of recombination. Pp. 106–125 in R. E. Michod and B. R. Levin (eds.), The Evolution of Sex. Sinauer Associates, Sunderland, MA. Mayne, R. 1984. The different types of collagen and collagenous peptides. Pp. 33–42 in R. L. Trelstad (ed.), The Role of Extracellular Matrix in Development. Alan R. Liss, New York. Mayr, E. 1961. Cause and effect in biology. Science 134: 1501–1506. Mazzeo, R. S. 1994. The influence of exercise and aging on immune function. Med. Sci. Sports Exercise 26: 586–592. McAdams, H. H. and L. Shapiro. 1995. Circuit simulation of genetic networks. Science 269: 650–656. McAdams, H. H. and L. Shapiro. 2003. A bacterial cell-cycle regulatory network operating in time and space. Science 301: 1874–1877. McArdle, A., W. H. Dillman, R. Mestril, J. A. Faulkner and M. J. Jackson. 2004. Overexpression of HSP70 in mouse skeletal muscle protects against muscle damage and age-related muscle dysfunction. FASEB J. 18: 355–357. McCann, J. 2001. Wanna bet? Two scientists wager on whether humans can live to 130 or 150 years. Scientist 15: 8. McCarter, R. 1978. Effects of age on contraction of mammalian skeletal muscle. Pp. 1–21 in G. Kaldor and W. J. DiBattista (eds.), Aging in Muscle. Raven Press, New York. McCarter, R., W. Mejia, H. P. Bertrand and B. P. Yu. 1996. Anti-aging effect of mild calorie restriction

564

References

(CR) in combination with low level voluntary wheel running. Gerontologist 36: 165 (abstract). McCay, C. M. and M. F. Crowell. 1934. Prolonging the life span. Sci. Monthly 39: 405–414. McCay, C., M. Crowell and L. Maynard. 1935. The effect of retarded growth upon the length of life and upon ultimate size. J. Nutr. 10: 63–79. McClearn, G. E. 1997. Biomarkers of age and aging. Exp. Gerontol. 32: 87–94. McClearn, G. E. 1999. Exotic mice as models for aging research: polemic and prospectus by R. Miller et al. Neurobiol. of Aging 20: 233–236. McClearn G. E. 2004. Nature and nurture: interaction and coaction. Am. J. Med. Genet. 124B: 124–130. McClintock, B. 1941. The stability of broken ends of chromosomes in Zea mays. Genetics 41: 234–282. McGandy, R. B., C. H. Barrows, Jr., A. Spania, A. Meredith, J. L. Stone and A. H. Norris. 1966. Nutrient intake and energy expenditure in men of different ages. J. Gerontol. 21: 581–587. McGeer, E. G. and P. L. McGeer. 1999. Brain inflammation in Alzheimer disease and the therapeutic implications. Curr. Pharm. Des. 5: 821–836. McGeer, P. L. and E. G. McGeer. 2002. Innate immunity, local inflammation, and degenerative disease. www.sagke.sciencemag.org/cgi/content/full/ sageke;2002/29/re3. McGue, M., J. W. Vaupel, N. Holm and B. Harvald. 1993. Longevity is moderately heritable in a sample of Danish twins born 1871–1880. J. Gerontol.: Biol. Sci. 48: B237–B244. McGuire, H. L., L. P. Svetkey, D. W. Harsha, P. J. Elmer, L. J. Appel and J. D. Ard. 2004. Comprehensive lifestyle modification and blood pressure control: A review of the PREMIER trial. J. Clin Hypertension 6: 383–390. McKenzie, D., E. Bua, S. McKiernan, Z. Cao, J. Wanagat and J. M. Aiken. 2002. Mitochondrial DNA deletion mutants: A causal role in sarcopenia. Eur. J. Biochem. 269: 2010–2015. McKusick, V. A. and F. H. Ruddle. 1978. The status of the gene map of the human chromosomes. Science 196: 390–405. McLachlan, M. S. F. 1978. The ageing kidney. Lancet 2(8081): 143–145. McMichael, A. J. 1992. Vegetarians and longevity: Imagining a wider reference population (editorial). Epidemiology 3: 389–391. Meany, M. J., D. O’Donnell, W. Rowe, B. Tannenbaum, A. Steverman, M. Walker, N. P. V. Nair and S. Lupien. 1995. Individual differences in hypothalamic-pituitary-adrenal activity in later life and hippocampal aging. J. Gerontol. 30: 229–251. Medawar, P. B. 1946. Old age and natural death. Mod. Q. 1: 30–56. Medawar, P. B. 1952. An Unsolved Problem of Biology. H. K. Lewis, London.

Medvedev, Z. A. 1981. On the immortality of the germ line: Genetic and biochemical mechanisms. A review. Mech. Ageing Devel. 17: 331–359. Medvedev, Z. A. 1986. Age structure of Soviet populations in the Caucasus: Facts and myths. Pp. 181– 200 in A. H. Bittles and K. J. Collins (eds.), The Biology of Human Ageing. Cambridge University Press, Cambridge. Melov, S., P. Coskun, M. Pater, R. Tuinstra, B. Cottrell, A. Jun, T. H. Zastawny, M. Dizdaroglu, S. I. Goodman, T. T. Huang et al. 1999. Mitochondrial disease in superoxide dismutase 2 mutant mice. Proc. Natl. Acad. Sci. USA 96: 846–851. Melov, S., S. R. Doctorow, J. A. Schneider, J. Haberson, M. Patel, P. E. Coskun, K. Huffman, D. C. Wallace and B. Malfroy. 2001. Lifespan extension and rescue of spongiform encephalopathy in superoxide dismutase 2 nullizygous mice treated with superoxide-dismutase-catalase mimetics. J. Neurosci. 21: 8348–8353. Melov, S., G. J. Lithgow, D. R. Fischer, P. M. Tedesco and T. E. Johnson. 1995. Increased frequency of deletions in the mitochondrial genome with age of Caenorhabditis elegans. Nucleic Acids Res. 23: 1419–1425. Melov, S., J. Ravenscroft, S. Malik, M. S. Gill, D. S. Walker, P. E. Clayton, C. D. Wallace, B. Malfroy, S. R. Doctrow and G. J. Lithgow. 2000. Extension of life span with superoxide dismutase/catalase mimetics. Science 289: 1657–1569. Melov, S., J. A. Schneider, B. J. Day, D. Hinerfeld, P. Coskun, S. S. Mirra, J. D. Crapo and D. C. Wallace. 1998. A novel neurological phenotype in mice lacking mitochondrial manganese superoxide dismutase. Nature Genet. 18: 159–163. Melvin, R. G., J. T. Miller, D. R. Spitz and J. W. O. Ballard. 2005. Interspecific variation in mitochondrial oxidative phosphorylation, oxygen consumption, and survival in Drosophila simulans. Submitted. Menzies, R. A. and P. H. Gold. 1972. The apparent turnover of mitochondria, ribosomes and sRNA of the brain in young adult and aged rats. J. Neurochem. 19: 1671–1683. Merkle, F. T., A. D. Tramontin, J. M. Garcia-Verdugo and A. Alvarez-Buylla. 2004. Radial glia give rise to adult neural stem cells in the subventricular zone. Proc. Natl. Acad. Sci. USA 101: 17528–17532. Merry, B. J. and A. M. Holehan. 1994a. Aging of the female reproductive system: The menopause. Pp. 147–170 in P. S. Timiras (ed.), Physiological Basis of Aging and Geriatrics, 2nd ed. CRC Press, Boca Raton, FL. Merry, B. J. and A. M. Holehan. 1994b. Effects of diet on aging. Pp. 171–178 in P. S. Timiras (ed.), Physiological Basis of Aging and Geriatrics, 2nd ed. CRC Press, Boca Raton, FL.

References

Metter E. J., L. A. Talbot, M., Schrager and R. Conwit. 2002. Skeletal muscle strength as a predictor of allcause mortality in healthy men. J. Gerontol. Biol. Sci. Med. Sci. 57: B359–365. Meydani, M. 2001. Nutrition interventions in aging and age-associated disease. Ann. N.Y. Acad. Sci. 928: 226–235. Meyer, J. S., S. Takashima, Y. Terayama, K. Obara, K. Muramatsu and S. Weathers. 1994. CT changes associated with normal aging of the human brain. J. Neurol. Sci. 123: 200–208. Miall, W. E. and H. G. Lovell. 1967. Relation between change of blood pressure and age. Br. Med. J. 2: 660–664. Migliaccio, E., M. Giorgio, S. Mele, G. Pelicci, P. Eboldi, P. P. Pandolfi, L. Lanfrancone and P. G. Pelicci. 1999. The p66shc adaptor protein controls oxidative stress response and life span in mammals. Nature 402: 309–313. Miller, A. R. 1988. A set of test life tables for theoretical gerontology. J. Gerontol. Biol. Sci. 43: B43– B49. Miller, E. R., III, R. Pastor-Barriuso, D. Dalal, R. A. Riemersma, L. J. Appel and E. Guallar. 2004. Meta-analysis: High-dosage vitamin E supplementation may increase all-cause mortality. Ann. Int. Med. 142. www.annals.org. Miller, R. A., J. M. Harper, A. Galecki and D. T. Burke. 2002. Big mice die young: Early body weight predicts longevity in genetically heterogeneous mice. Aging Cell 1: 22–29. Miller, R. A. 1995. Immune system. Pp. 555–590 in E. J. Masoro (ed.), Handbook of Physiology. Section 11: Aging. Oxford University Press, New York. Miller, R. A. 1996. Aging and the immune response. Pp. 355–392 in E. L. Schneider and J. W. Rowe (eds.), Handbook of the Biology of Aging, 4th edition. Academic Press, San Diego. Miller, R. A. 2001a. Biomarkers of aging. http://sageke .sciencemag.org/cgi/content/full/sageke ;2001/1/pe2. Miller, R. A. 2001b. Biomarkers of aging: Prediction of longevity by using age-sensitive T-cell subset determinations in a middle-aged, genetically heterogeneous mouse population. J. Gerontol. Biol. Sci. 56A: B180–B186. Miller, R. A., S. Austad, D. Burke, C. Chrisp, R. Dysko, A. Galecki, A. Jackson and V. Monnier. 1999. Exotic mice as models for aging research: polemic and prospectus. Neurobiol. Aging 20: 217–231. Miller, R.A., Y. Chang, A. T. Galecki, K. Al-Regaiey, J. J. Kopchick and A. Bartke. 2002a. Gene expression patterns in calorically restricted mice: Partial overlap with long-lived mutant mice. Mol. Endocrinol. 16: 2657–2666. Miller, R. A., C. Chrisp, A. U. Jackson, A. T. Galecki and D. T. Burke. 2002b. Coordinated genetic con-

565

trol of neoplastic and nonneoplastic diseases in mice. J. Gerontol. Biol.Sci. 57A: B3–B8. Miller, R. A., J. M. Harper, A. Galecki and D. T. Burke. 2002c. Big mice die young: Early body weight predicts longevity in genetically heterogeneous mice. Aging Cell 1: 22–29. Miller, R. M., F. Bookstein, J. van der Meulen, S. Engle, J. Kim, L. Mullins and J. Faulkner. 1997. Candidate biomarkers of aging: Age-sensitive indices of immune and muscle function covary in genetically heterogenous mice. J. Gerontol.: Biol. Sci. 52A: B39–B47. Miller, S. M. and D. L. Kirk, 1999. glsA, a Volvox gene required for asymmetric division and germ cell specification, encodes a chaperone-like protein. Development 126: 649–658. Millis, A. J., M. Hoyle, H. M. McCue and H. Martini. 1992. Differential expression of metalloproteinase and tissue inhibitor of metalloproteinase genes in aged human fibroblasts. Exp. Cell Res. 201: 373– 379. Minaker, K. L., G. S. Meneilly and J. W. Rowe. 1985. Endocrine systems. Pp. 433–456 in C. E. Finch and E. L. Schneider (eds.), Handbook of the Biology of Aging, 2nd ed. Van Nostrand Reinhold, New York. Minamino, T., H. Miyauchi, T. Yoshida, K. Tateno, T. Kuneida and I. Komuro. 2004. Vascular cell senescence and vascular aging. J. Mol. Cell. Cardiol. 36: 175–183. Minois, N. 2000. Longevity and aging: Beneficial effects of exposure to mild stress. Biogerontology 1: 15–29. Miquel, J. 2002. Can antioxidant diet supplementation protect against age-related mitochondrial damage? Ann. NY Acad. Sci. 959: 508–516. Miquel, J. and J. E. Fleming. 1984. A two step hypothesis on the mechanism of in vitro cell aging: Cell differentiation followed by intrinsic mitochondrial mutagenesis. Exp. Gerontol. 19: 31–36. Miquel, J., A. C. Economos, J. Fleming and J. E. Johnson Jr. 1980. Mitochondrial role in cell aging. Exp. Gerontol. 15: 575–591. Mishkin, M. and T. Appenzeller. 1987. The anatomy of memory. Sci. Am. 256(6): 80–89. Missirlis, F., J. P. Phillips and H. Jackle. 2001. Cooperative action of antioxidant defense systems in Drosophila. Curr. Biol. 11: 1272–1277. Mitchell-Olds, T. 1995. The molecular basis of quantitative genetic variation in natural populations. Trends Ecol. Evol. 10: 324–328. Mitnitski, A.B., A.J. Mogilner, C. MacKnight and K. Rockwood, 2002. The mortality rate as a function of accumulated deficits in a frailty index. Mech. Ageing Develop. 123: 1457–1460. Miwa, S., J. St-Pierre, L. Partridge and M. D. Brand. 2003. Superoxide and hydrogen peroxide produc-

566

References

tion by Drosophila mitochondria. Free Rad. Biol. Med. 35: 938–948. Miyadera, H, H Amino, A Hiraishi, H Taka, K Myrayama, H Miyoshi, K Sadamoto, N Ishii, S Hekimi and K Kita. 2001. Altered quinone biosynthesis in the long-lived clk-1 mutant of Caenorhadbitis elegans. J. Biol. Chem. 276: 7713–7716. Miyashita, M., S. Haga and T. Mozota. 1978. Training and detraining effects on aerobic power in middle-aged and older men. J. Sports Med. 18: 131–137. Miyata, M. and J. D. Smith. 1996. Apolipoprotein E allele-specific antioxidant activity and effects on cytotoxicity by oxidative insults and beta-amyloid peptides. Nature Genet. 14: 55–61. Mlekusch, W., M. Lamprecht, K. Ottl, M. Tillian and G. Reibnegger. 1996a. A glucose-rich diet shortens longevity of mice. Mech. Ageing Dev. 92: 43–51. Mlekusch W., M. Tillian, M. Lamprecht, K. Oettl, H. Krainz and G. Reibnegger. 1998. The life-shortening effect of reduced physical activity is abolished by a fat rich diet. Mech. Ageing Dev. 105: 61–73. Mlekusch, W., H. Tillian, M. Lamprecht, H. Trutnovsky, R. Horejsi and G. Reibnegger. 1996b. The effect of reduced physical activity on longevity of mice. Mech. Ageing Dev. 88: 159–168. Mobbs, C. V. and C. E. Finch. 1992. Estrogen-induced impairments as a mechanism in reproductive senescence of female C57BL/6J mice. J Gerontol. 47: B48–B51. Mobbs, C. V., G. A. Bray, R. L. Atkinson, A. Bartke, C. E. Finch, E. Maratos-Flier, J. N. Crawley and J. F. Nelson. 2001. Neuroendocrine and pharmacological manipulations to assess how caloric restriction increases life span. J. Gerontol. 56A: 34–44 (special issue I). Mobbs, C. V. 1996. Neuroendocrinology of aging. Pp. 234–283 in E. L. Schneider and J. W. Rowe (eds.), Handbook of the Biology of Aging, 4th ed. Academic Press, San Diego, CA. Mocchegiani, E., L. Santarelli, M. Muzzoli and N. Fabris. 1995. Reversibility of the thymic involution and of age-related peripheral immune dysfunctions by zinc supplementation in old mice. Int. J. Immunopharmacol. 17: 703–718. Mockett, R. J., W. C. Orr, J. J. Rahmandar, J. J. Benes, S. V. Radyuk, V. I. Klichko and R. S. Sohal. 1999. Overexpression of Mn-containing superoxide dismutase in transgenic Drosophila melanogaster. Arch. Biochem. Biophys. 371: 260–269. Moeller, J. R., T. Ishikawa, V. Dhawan, P. Spetsieris, F. Mandel, G. E. Alexander, C. Grady, P. Pietrini and D. Eidelberg. 1996. The metabolic topography of normal aging. J. Cerebral Blood Flow Metab. 16: 385–398. Moffat, S. D., A. B. Zanderman, S. M. Harman, M. R. Blackman, C. Kawas and S. M. Resnick. 2000. The

relationship between longitudinal declines in dehydroepiandrosterone sulfate concentrations and cognitive performance in older men. Arch. Intern. Med. 160: 2193–2198. Mohs, M. E. 1994a. Adult protein-calorie malnutrition among special populations and developing countries. Pp. 11–36 in R. R. Watson (ed.), Handbook of Nutrition in the Aged, 2nd ed. CRC Press, Boca Raton, FL. Mohs, M. E. 1994b. Assessment of nutritional status in the aged. Pp. 145–164 in R. R. Watson (ed.), Handbook of Nutrition in the Aged, 2nd ed. CRC Press, Boca Raton, FL. Moldave, K., J. Harris, W. Sabo and I. Sadnik. 1979. Protein synthesis and aging: Studies with cell-free mammalian systems. Fed. Proc. 38: 1979–1983. Moment, G. 1982. Theories of aging: An overview. Pp. 2–23 in R. C. Adelman and G. S. Roth (eds.), Testing the Theories of Aging. CRC Press, Boca Raton, FL. Montooth, K. L., J. H. Marden and A. G. Clark. 2003. Mapping determinants of variation in energy metabolism, respiration, and flight in Drosophila. Genetics 165: 623–635. Moore, C. J. and A. G. Schwartz. 1978. Inverse correlation between species life span and capacity of cultured fibroblasts to convert benzo[a]pyrene to water-soluble metabolities. Exp. Cell Res. 116: 359–364. Moore, W. R., M. E. Anderson, A. Meister, K. Murata and A. Kimura. 1989. Increased capacity of glutathione synthesis enhances resistance to radiation in Escherichia coli: A possible model for mammalian cell protection. Proc. Natl. Acad. Sci. USA 86: 1461–1464. Moraes, C. T. 2001. What regulates mitochondrial DNA copy number in animal cells? Trends Genet. 17: 199–205. Morell, V. 1995. Zeroing in on how hormones affect the immune system. Science 269: 773–775. Mori, H., J. Kendo and Y. Ihara. 1987. Ubiquitin in a component of paired helical filaments in Alzheimer’s disease. Science 235: 1641–1644. Morimoto, M., N. Morita and M. Kawata. 1994. The effects of NGF and glucocorticoid on the cytological features of rat chromaffin cells in vitro. Neuroreport 5: 954–956. Morley, J. E. 2001. Decreased food intake with aging. J. Gerontol. 56A (special issue II): 81–88. Morley, J. E. 2004. The metabolic syndrome and aging. J. Gerontol. Med. Sci. 59A: 139–142. Morley, J. E., H. R. Brignull, J. J. Weyers and R. I. Morimoto. 2002. The threshold for polyglutamine expansion protein aggregation and cellular toxicity is dynamic and influenced by aging in Caenorhabditis elegans. Proc. Natl. Acad. Sci. USA 99: 10417–10422.

References

Morley, J. E. and R. I. Morimoto. 2004. Regulation of longevity in Caenorhabditis elegans by heat shock factor and molecular chaperones. Mol. Biol. Cell 15: 657–664. Morre, D. J. 2000. Chemical hormesis in cell growth: A molecular target at the cell surface. J. Appl. Toxicol. 20: 157–163. Morris, J. Z., H. A. Tissenbaum and G. Ruvkum. 1996. A phosphatidylinositol-3–OH kinase family member regulating longevity and diapause in Caenorhabditis elegans. Nature 382: 536–539. Morrow, G., S. Battistini, P. Zhang and R. M. Tanguay. 2004a. Decreased lifespan in the absence of expression of the mitochondrial small heat shock protein hsp22 in Drosophila. J. Biol. Chem. 279: 43382–43385. Morrow, G., M. Samson, S. Michaud and R. M. Tanguay. 2004b. Overexpression of the small mitochondrial Hsp22 extends Drosophila life span and increases resistance to oxidative stress. FASEB J. 18(3): 598–599. Morse, H. C., R. A. Yetter, J. H. Stimpfling, O. M. Pitts, T. N. Fredrickson and J. W. Hartley. 1980. Greying with age in mice: Relation to expression of murine leukemia viruses. Cell 41: 439–448. Mortimer, R. K. and J. R. Johnson, 1959. Life span of individual yeast cells. Nature 183: 1751–1752. Moskovitz, J., S. Bar-Noy, W. M. Williams, J. Requena, B. S. Berlett and E. R. Stadtman. 2001. Methionine sulfoxide reductase (Msr A) is a regulator of antioxidant damage and lifespan in mammals. Proc. Natl. Acad. Sci. USA 98: 12920–12925. Motoyama, H., F. Wang, K. A. Roth, H. Sawa, K. Nakayama, K. Nakayama, I. Negishi, S. Senju, Q. Zhang, S. Fujii and D. Y. Loh. 1995. Massive cell death of immature hematopoietic cells and neurons in Bclx-deficient mice. Science 267: 1506–1510. Moyer, J. H., M. J. Lee-Tischler, H-Y. Kwon, J. J. Schrick, E. D. Avner, W. E. Sweeney, V. L. Godfrey, N. L. A. Cacheiro, J. E. Wilkinson and R. P. Woychik. 1994. Candidate gene associated with a mutation casing recessive polycystic kidney disease in mice. Science 264: 1329–1333. Muggleton-Harris, A. L. 1979. Reassembly of cellular components for the study of aging and finite lifespan. Intl. Rev. Cytol. 9(suppl.): 279–301. Muller, F. 2001. Genetically altered mice. http:// sageke.sciencemag.org/resources/experimental/ transgenic/. Muller, H. J. 1938. The remaking of chromosomes. Collect. Net 13: 1181–1198. Muller, H. J. 1964. The relation of recombination to mutational advance. Mutat. Res. 1: 2–9. Muller, H.-G., J-M. Chiou, J. R. Carey and J-L. Wang. 2002. Fertility and life span: Late children enhance female longevity. J. Gerontol. Biol. Sci. 57A: B202–B206.

567

Muller, I., M. Zimmermann, D. Becker and M. Flomer. 1980. Calendar life span versus budding life span of Saccharomyces cerevisiae. Mech. Aging Dev. 12: 47–52. Munkres, K. D. 1985. The role of genes, antioxidants and antioxygenic enzymes in the aging of Neurospora: A review. Pp. 237–248 in L. W. Oberly (ed.), Superoxide Dismutases, vol. 3. Pathological States. CRC Press, Boca Raton, FL. Munkres, K. D. 1990. Genetic coregulation of longevity and anti oxienzymes in Neurospora crassa. Free Rad. Biol. Med. 8: 355–361. Munkres, K. D. 1992. Selection and analysis of superoxide dismutase mutants of Neurospora. Free Rad. Biol. Med. 13: 305–318. Munkres, K. D. and C. Furtek. 1984. Linkage of conidial longevity determinant genes in Neurospora crassa. Mech. Ageing Dev. 25: 63–77. Munkres, K. D., R. S. Rana and E. Goldstein. 1984. Genetically determined conidial longevity is positively correlated with superoxide dismutase, catalase, glutathionine peroxidase, cytochrome c peroxidase and ascorbate free radical reductase activities in Neurosopora crassa. Mech. Ageing Dev. 24: 83–100. Munnell, J. F. and R. Getty. 1968. Rate of accumulation of cardiac lipofuscin in the aging canine. J. Gerontol. 23: 154–158. Munoz, M. J. 2003. Longevity and heat stress regulation in Caenorhabditis elegans. Mech. Ageing Dev. 124: 43–48. Munscher, C., J. Muller-Hocher and J. Naplwotzki. 1993. Human aging is associated with various point mutations in tRNA genes of mitochondrial DNA. Biol. Chem. Hoppe Seyler 374: 1099– 1104. Murphy, E. A. 1978. Genetics of longevity in man. Pp. 261–302 in E. L. Schneider (ed.), The Genetics of Aging. Plenum, New York. Murphy, K. and R. Topel. 2003. Diminishing returns: The costs and benefits of improving health. Perspect. Biol. Med. 46 (suppl.): S108–S128. Murphy, W. J., S. K. Durum and D. L. Longo. 1993. Differentiation effects of growth hormone and prolactin on murine T cell development and function. J. Exp. Med. 178: 231. Murray, V. and R. Holliday. 1981. Increased error frequency of DNA polymerases from senescent human fibroblasts. J. Mol. Biol. 146: 55–76. Myers, G. C. and K. G. Manton. 1984. Compression of mortality: Myth or reality? Gerontologist 24: 346–353. Nabarra, B. and I. Andrianarison. 1996. Ultrastructural study of thymic microenvironment involution in aging mice. Exp. Gerontol. 31: 489–506. Naito, H., S. K. Powers, H. A. Demirel and J. Aoki. 2001. Exercise training increase heat shock protein

568

References

in skeletal muscles of old rats. Med. Sci. Sports Exerc. 33: 729–734. Nakagawa, I., A. Amano, N. Mizushima, A. Yamamoto, H. Yamaguchi, T. Kamimoto, A. Nara, J. Funao, M. Nakata, K. Tsuda, S. Hamada and T. Yoshimori. 2004. Autophagy defends cells against invading group A Streptococcus T. Science 306: 1037– 1040. Nakamura, E. 1994. Statistical approach for the assessment of biological age. Pp. 439–456 in A. K. Balin (ed.), Practical Handbook of Human Biologic Age Determination. CRC Press, Boca Raton, FL. Nakamura, E. and K. Miyao. 2003. Further evaluation of the basic nature of the human biological aging process based on a factor analysis of age-related physiological variables. J. Gerontol. Biol. Sci. 58A: 196–204. Nakamura, E., K. Miyao and T. Ozeki. 1988. Assessment of biological age by principal component analysis. Mech. Ageing Dev. 46: 1–18. Nakamura, E., T. Moritani and A. Kanetaka. 1996. Effects of habitual physical exercise on physiological age in men aged 20–85 years as estimated using principal component analysis. Eur. J. Appl. Physiol. Occup. Physiol. 73: 410–418. Nakamura, E., T. Moritani and A. Kanetaka. 1998. Further evaluation of physical fitness age versus physiological age in women. Eur. J. Appl. Physiol. Occup. Physiol. 78: 195–200. Nakashima, K., T. Kiyosue, K Yamaguchi-Shinozaki and K Shinozaki. 1997. A nuclear gene, erd1, encoding a chloroplast-targeted Clp protease regulatory subunit homolog is not only induced by water stress but also developmentally up-regulated during senescence in Arabidopsis thaliana. Plant J. 12: 851–861. Nandy, K. 1984. Effects of antioxidant on neuronal lipofuscin pigment. Pp. 223–233 in D. Armstrong, R. S. Sohal, R. G. Cutler and T. F. Slater (eds.), Free Radiation in Molecular Biology, Aging and Disease. Raven Press, New York. Nandy, K. and H. Schneider. 1976. Lipofucsin pigment formation in neuroblastoma cells in culture. Pp. 245– 264 in R. D. Terry and S. Gershon (eds.), Neurobiology of Aging. Raven Press, New York. Napoli, C., I. N. Martin-Padura, F. de Nigris, M. Giorgio, G. Mansueto, P. Somma, M. Condorelli, G. Sica, G. De Rosa and P. G. Pelicci. 2003. Deletion of the p66shc longevity gene reduces systemic and tissue oxidative stress, vascular cell apoptosis, and early atherogenesis in mice fed a high-fat diet. Proc. Natl. Acad. Sci. USA 100: 2112–2116. Nash, M. S. 1994. Exercise and immunology. Med. Sci. Sports Exerc. 26: 125–127. National Center for Health Statistics. 1999. U.S. decennial life tables for 1989–91, vol. 1, no. 3. Some

trends and comparisons of United States life table data: 1990–1991. NCHS, Hyattsville, MD. National Research Council. 1981. Mammalian Models for Research on Aging. National Academy Press, Washington, DC. Nawata, H., T. Yanase, K. Goto, T. Okabe and K. Ashida. 2002. Mechanism of action of anti-aging DHEA-S and the replacement of DHEA-S. Mech. Ageing Dev. 123: 1101–1106. Neafsey, P. J. 1990. Longevity hormesis: A review. Mech. Ageing Dev. 51: 1–31. Neel, J. V. 1962. Diabetes mellitus: a “thrifty” genotype rendered detrimental by “progress”? Am. J. Hum. Genet. 14: 353–362. Nelson, J. F. 1995. The potential role of selected endocrine systems in aging processes. Pp. 377–394 in E. J. Masoro (ed.), Handbook of Physiology, Section 11: Aging. Oxford University Press, New York. Nemoto, S. and T. Finkel. 2002. Redox regulation of forkhead proteins through a p66shc-dependent signaling pathway. Science 295: 2450–2452. Ness J., D. Johnson and N. Nisly. 2003. “Polypharmacy”: herbal supplements as a form of polypharmacy in older adults. (Letter to the editor). J. Gerontol. Med. Sci. 58AS: 478. Nesse, R. M. and G. C. Williams. 1994. Why We Get Sick: The New Science of Darwinian Medicine. Vintage Books, New York. Neumann, H., A. Calalie, D. E. Jenne and H. Wekerle. 1995. Induction of MHC Class 1 genes in neurons. Science 269: 549–552. Newman, A. B., J. S. Gottdiener, M. A. McBurnie, C. H. Hirsch, W. J. Kop, R. Tracy, J. D. Walston and L. P. Fried. 2001. Associations of subclinical cardiovascular disease with fraility. J. Gerontol. Med Sci. 56A: M158–M166. Newton, J. R. and R. Yernas. 1986. Changes in the contractile properties of the human first dorsal interosseous muscle with age. Gerontology 32: 98– 104. Nicholls, D. 2002. Mitochondrial bioenergetics, aging, and aging-related disease. Sci. Aging Knowl. Environ., August 7. http://sageke.sciencemag.org/ cgi/content/full/sageke;2002/31/pei12. Nicholls, D. (ed). 2004. Mitochondria and aging: facts and fancies. Aging Cell 3(1): 1–44. Niedermuller, H. 1985. DNA repair during aging. Pp. 173–194 in R. S. Sohal, L. S. Birnbaum and R. G. Cutler (eds.), Molecular Biology of Aging: Gene Stability and Gene Expression. Raven Press, New York. Niedzwiecki, A. and J. E. Fleming. 1990. Changes in protein turnover after heat shock are related to accumulation of abnormal proteins in aging Drosophila melanogaster. Mech Ageing Dev. 52(2–3): 295–304.

References

Nieschlag, E. and E. Michael. 1986. Reproductive functions in grandfathers. Pp. 59–71 in L. Mastroianni, Jr. and C. A. Paulsen (eds.), Aging, Reproduction and the Climacteric. Plenum Press, New York. Nigg, E. A. 1995. Cyclin-dependent protein kinases: Key regulators of the eucaryotic cell cycle. BioEssays 17: 471–480. Nikitin, A. G. and R. C. Woodruff. 1995. Somatic movement of the mariner transposable element and lifespan of Drosophila species. Muta. Res. 338: 43–49. Nilsen, J. and R. D. Brinton. 2004. Mitochondria as therapeutic targets of estrogen action in the central nervous system. Curr. Drug Targets CNS Neurol. Disord. 3: 297–313. Ninedzwiecki, A., A. M. Kongapchith and J. E. Fleming. 1991. Aging affects expression of 70-kDa heat shock proteins in Drosophila. J. Biol. Chem. 266: 9332–9338. Ning, Y. and O. M. Pereira-Smith. 1991. Molecular genetic approaches to the study of cellular senescence. Mutat. Res. 256: 303–310. Nnodim, J. O. 2000. Satellite cell numbers in senile rat levator ani muscle. Mech Ageing Dev. 112: 99– 111. Nooden, L. D. 1988a. Postlude and prospects. Pp. 499– 517 in L. D. Nooden and A. C. Leopold (eds.), Senescence and Aging in Plants. Academic Press, San Diego. Nooden, L. D. 1988b. Whole plant senescence. Pp. 392–441 in L. D. Nooden and A. C. Leopold (eds.). 1988. Senescence and Aging in Plants. Academic Press, San Diego. Nooden, L. D. and J. J. Guiamet. 1996. Genetic control of senescence and aging in plants. Pp. 94–120 in E. L. Schneider and J. W. Rowe (eds.), Handbook of the Biology of Aging, 4th edition. Academic Press, San Diego. Nooden, L. D. and A. C. Leopold (eds.). 1988. Senescence and Aging in Plants. Academic Press, San Diego. Nooden, L. D. and D. S. Letham. 1993. Cytokinin metabolism and signalling in the soybean plant. Australian J. Plant Physiol. 20: 639–653. Nooden, L.D. and J. P. Penney. 2001. Correlative controls of senescence and plant death in Arabidopsis thaliana (Brassicaceae). J. Exp. Botany 52: 2151– 2159. Nooden, L. D., M. J. Schneider, J. Hifllsberg and S. Levy. 1995. Induction of senescence in Arabidopsis by long days and light dosage. Plant Physiol. 108: 61 (abstract). Nooden, L. D. and J. E. Thompson. 1985. Aging and senescence in plants. Pp. 105–127 in C. E. Finch and E. L. Schneider (eds.), Handbook of the Biology of Aging, 2nd edition. Van Nostrand Reinhold, New York.

569

Norris, A. H. and N. W. Shock. 1974. Exercise in the adult years. Pp. 346–365 in W. R. Johnson and E. R. Buskirk (eds.), Science and Medicine of Exercise and Sport. Harper and Row, New York. Norton, H. T. J. 1928. Natural selection and Mendelian variation. Proc. Lond. Math. Soc. 28: 1–45. Norwood, T. H. and J. R. Smith. 1985. The cultured fibroblast-like cell as a model for the study of aging. Pp. 291–321 in C. E. Finch and E. L. Schneider (eds.), Handbook of the Biology of Aging, 2nd ed. Van Nostrand Reinhold, New York. Novoseltsev, V. N., R. Arking, J. A. Novoseltseva and A. I. Yashin. 2002. Evolutionary optimality applied to Drosophila experiments: hypothesis of constrained reproductive efficiency. Evolution 56: 1136–1149. Novoseltsev, V. N., J.A. Novoseltseva and A. I. Yashin. 2003. What does a fly’s individual fecundity pattern look like? The dynamics of resource allocation in reproduction and ageing. Mech. Ageing Dev. 124: 605–617. Noymer, A. and M. Garenne, 2000. The 1918 influenza epidemic’s effects on sex differentials in mortality in the United States. Popul. Dev. Rev. 26: 565–581. Nussbaum, T. J., L. D. Mueller and M. R. Rose 1996. Evolutionary patterns among measures of aging. Exp. Gerontol. 31: 507–516. Nystrom, T. 2002. Aging in bacteria. Curr. Opin. Microbiol. 5: 596–601. O’Connor, P. J., F. C. Manning, A. T. Gordon, M. A. Billett, D. P. Cooper, R. H. Elder and G. P. Margison. 2000. DNA repair: kinetics and thresholds. Toxicol. Pathol. 28: 375–378. O’Donnell, A. B., A. B. Arajuo and J. B. McKinlay. 2004. The health of normally aging men: The Massachusetts Male Aging Study (1987–2004). Exp. Gerontol. 39: 975–984. Oeppen, J. and J. W. Vaupel, 2002. Broken limits to life expectancy. Science 296: 1029–1031. Ogborn, C. E. and G. M. Martin. 1985. Age-related declines in the replicative potentials of aortic cells from rhesus monkeys: Evidence from primary cloning and organoid culture techniques. Pp. 101– 106 in R. Davis (ed.), Behavior and Pathology of Aging in Rhesus Monkeys. Alan R. Liss, New York. Ogburn, C. E., K. Carlberg, M. A. Ottinger, D. J. Holmes, G. M. Martin and S. N. Austad. 2001. Exceptional cellular resistance to oxidative damage in long-lived birds requires active gene expression. J. Gerontol. Biol Sci. 56A: B468–B474. Ogg, S., S. Paradis, S. Gottlieb, S. I. Patterson, L. Lee, H. A. Tissenbaum and G. Ruvkun. 1997. The Forkhead transcription factor DAF-16 transduces insulin-like metabolic and longevity signals in C. elegans. Nature 389: 994–999.

570

References

Oh, SA, JH Park, GI Lee, KH Paek, SK Park and HG Nam. 1997. Identification of three genes controlling leaf senescence in Arabidopsis thaliana. Plant J. 12: 527–535. Ohtani, S., S. Kato and H. Sugeno. 1996. Changes in D-aspartic acid in human deciduous teeth with age from 1–20 years. Growth Dev. Aging 60: 1–6. Olovnikov, A. M. 1971. Principles of marginotomy in template synthesis of polynucleotides. Doklady Adad. Nauk SSSR 201: 1496–1499. Olovnikov, A. M. 1973. A theory of marginotomy: The incomplete copying of template margin in enzymic synthesis of polynucleotides and biological significance of the phenomenon. J. Theoret. Biol. 41: 181–190. Olovnikov, A. M. 1996. Telomeres, telomerase and aging: Origin of the theory. Exp. Gerontol. 31: 443–448. Olshansky, S. J., B. A. Carnes and C. Cassel. 1990. In search of Methuselah: Estimating the upper limits to human longevity. Science 250: 634–640. Olshansky, S. J., B. A.Carnes and A. Desesquelles. 2001. Prospects for human longevity. Science 291: 1491–1492. Olshansky, S. J., L. Hayflick and B. A. Carnes. 2002. Position statement on human aging. J. Gerontol. Biol. Med. Sci. 57A: B292–B297. Omholt, S. W. and G. V. Amdam. 2004. Epigenetic regulation of aging in honeybee workers. Sci. Aging Knowl. Environ., June. http://sageke. sciencemag .org/cgi/content/full/2004/26/pe28. O’Nell, O. 2002. Reason and passion in bioethics. Science 298: 2335 (book review). Ono, T. and R. G. Cutler. 1978. Age-dependent relaxation of gene repression: Increase of endogenous murine leukemia virus-related and globin-related RNA in brain and liver of mice. Proc. Natl. Acad. Sci. USA 75: 4431–4435. Ono, T. and S. Okada. 1978. Does the capacity to rejoin radiation-induced DNA breaks decline in senescent mice? Intl. J. Radiation Biol. 33: 403– 407. Ono, T., S. Okada and T. Sugahara. 1976. Comparative studies of DNA size in various tissues of mice during the aging process. Exp. Gerontol. 11: 127– 132. Orentreich, D. S. and N. Orentreich. 1994. Hair changes with aging as a parameter to utilize in the estimation of human biological age. Pp. 375–390 in A. K. Balin (ed.), Practical Handbook of Human Biologic Age Determination. CRC Press, Boca Raton, FL. Orentreich, N., J. L. Brind, R. L. Rizer and J. H. Vogelman. 1984. Age changes and sex differences in serum dehydroepiandrosterone sulfate concentrations throughout childhood. J. Clin. Endocrinol. Metab. 59: 551–555.

Orentreich, N., J. A. Zimmerman and J. R. Matias. 1994. Dehydroepiandrosterone: Marker or modifier of aging. Pp. 391–396 in A. K. Balin (ed.), Practical Handbook of Human Biologic Age Determination. CRC Press, Boca Raton, FL. Orgel, L. E. 1963. The maintenance of the accuracy of protein synthesis and its relevance to aging. Proc. Natl. Acad. Sci. USA 49: 517–521. Orive, M. E. 1995. Senescence in organisms with clonal reproduction and complex life histories. Am. Nat. 145: 90–108. Ornish, D. 1993. Regression of coronary artery disease by a multifactorial approach: The Lifestyle Heart Trial. Pp. 319–329 in P. C. Weber and A. Leaf (eds.), Atherosclerosis Reviews, vol. 25. Raven Press, New York. Orr, W. C. and R. C. Sohal. 1992. The effects of catalase gene overexpression on life span and resistance to oxidative stress in transgenic Drosophila melanogaster. Arch. Biochem. Biophys. 297: 35–41. Orr, W. C. and R. C. Sohal. 1993. Effects of Cu/Zn superoxide dismutase overexpression on life span and resistance to oxidative stress in transgenic Drosophila melanogaster. Arch. Biochem. Biophys. 301: 34–40. Orr, W. C. and R. C. Sohal. 1994. Extension of lifespan by overexpression of superoxide dismutase and catalase in Drosophila melanogaster. Science 263: 1128–1130. Orr, W. C. and R. S. Sohal. 2003. Does overexpression of CuZnSOD extend life span in Drosophila melanogaster? Exp. Gerontol. 38: 227–230. Orrenius, S. 2004. Mitochondrial regulation of apoptotic cell death. Toxicol. Lett. 149: 19–23. Osborne, T. B., L. B. Mendel and E. L. Ferry. 1917. The effect of retardation of growth upon the breeding period and duration of life of rats. Science 45: 294–295. Osiewacz, H. D. 1995. Aging and genetic instabilities. Pp. 29–44 in K. Esser and G. M. Martin (eds.), Molecular Aspects of Aging. Wiley, Chichester, England. Ostfeld, A., C. M. Smith and B. A. Stotsky. 1977. The systemic use of procaine in the treatment of the elderly: A review. J. Am. Geriatr. Soc. 25: 1–9. Otero, P., M. Viana, E. Herrera and B. Bonet. 1997. Antioxidant and prooxidant effects of ascorbic acid, dehydroascorbic acid, and flavonoids on LDL submitted to different degrees of oxidation. Free Rad. Res. 27: 619–626. Ottinger, M. A., R. E. Ricklefs and C.E. Finch (eds.). 2003. Proceedings of the 2nd Symposium on Organisms with Slow Aging. Exp. Gerontol. 38(7): 721–811. Ouwor, E. D. and A-N. T. Kong. 2002. Antioxidants and oxidants regulate signal transduction pathways. Biochem. Pharmacol. 64: 765–770.

References

Overall, C. 2003. Aging, Death, and Human Longevity: A Philosophical Inquiry. University of California Press, Berkeley CA. Ozanne, S. E. and C. N. Hales. 2004. Catch-up growth and obesity in mice. Nature 427: 411–412. Packer, L. 1991. Protective role of vitamin E in biological systems. Am J. Clin. Nutr. 53: 1050S– 1055S. Paffenbarger, R. S. Jr., R. T. Hyde, A. L. Wing and C. Hsieh. 1986. Physical activity, all cause mortality, and longevity of college alumni. N. Engl. J. Med. 315: 399–401. Pahlavani, M. A. and A. Richardson. 1996. The effect of age on the expression of interleukin-2. Mech. Ageing Dev. 89: 125–154. Painter, R. S., J. M. Clarkson and B. R. Young. 1973. Ultraviolet-induced repair replication in aging diploid human cells (WI-38). Radiat. Res. 56: 560– 564. Pak, J. W., A. Herbst, E. Bua, N. Gokey, D. McKenzie and J. M. Aiken. 2003. Mitochondrial DNA mutations as a fundamental mechanism in physiological declines associated with aging. Aging Cell 2: 1–8. Pakkenberg, B., D. Pelvig, L. Marner, M. J. Bundgaard, H. J. G. Gundersen J. R. Nyengaard and L, Regeur. 2003. Aging and the human neocortex. Exp. Gerontol. 38: 95–99. Palmore, E. B. 1982. Predictors of the longevity difference: A 25 year follow-up. Gerontologist 22: 513–518. Pamplona, R. and G. Barja. 2003. Aging rate, mitochondrial free radical production, and constituitive sensitivity to lipid peroxidation: Insights from comparative studies. Pp. 47–64 in T. von Zglinicki (ed.), Aging at the Molecular Level. Kluwer Academic Publishers, Dordrecht, the Netherlands. Pamplona, R., G. Barja and M. Portero-Otin. 2002. Membrane fatty acid unsaturation, protection against oxidative stress, and maximum life span. Ann. N.Y. Acad. Sci. 959: 475–490. Paolisso, G., S. Ammendola, A. del Buono, A. Gambardella, M. Riondino, M. R. Tagliamonte, M. R. Rizzo, C. Carella and M. Varricchio. 1997. Serum levels of insulin-like growth factor -1 (IGF-1) and IGF-binding protein-3 in healthy centenarians; Relationship with plasma leptin and lipid concentrations, insulin action, and cognitive function. J. Clin. Endocrin. Metab. 82: 2204–2209. Papaconstantinou, J., P. D. Reisner, L. Liu and D. T. Kuninger. 1996. Mechanisms of altered gene expression with aging. Pp. 150–183 in E. L. Schneider and J. W. Rowe (eds.), Handbook of the Biology of Aging, 4th ed. Academic Press, San Diego, CA. Parikh, V. S., M. M. Morgan, R. Scott, L. S. Clements and R. A. Bulow. 1987. The mitochondrial geno-

571

type can influence nuclear gene expression in yeast. Science 235: 576–580. Park, D. U., C. H. Kim, S. E. Hong, B. P. Yu and H. C. Chung. 2003. AgingDB: A database for oxidative stress and calorie restriction in the study of aging. J. Am. Aging Assoc. 26: 11–18. Park, H. K., J. S. Yook, H. S. Koo, I. S. Choi and B. Ahn. 2002. The Caenorhabiditis elegans XPA homolog of human XPA. Mol. Cells 14: 50–55. Park, J. H., S. A. Oh, Y. H. Kim, H. R. Woo and H. G. Nam. 1998. Differential expression of senescence-associated mRNAs during leaf senesence induced by different senescence-inducing factors in Arabidopsis. Plant Mol. Biol. 37: 445–454. Park, J-W., C-H. Choi, M-S. Kim and M-H. Chung. 1996. Oxidative status in senescence-accelerated mice. J. Gerontol.: Biol. Sci. 51A: B337–B345. Parkes. T. L., A. J. Elia, D. Dickinson, A. J. Hilliker, J. P. Phillips and G. L. Boulianne. 1998. Extension of Drosophila lifespan by overexpression of human SOD1 in motorneurons. Nature Genet. 19: 171–174. Parkes, T. L., A. J. Hilliker and J. P. Phillips. 1999. Motorneurons, reactive oxygen and life span in Drosophila. Neurobiol. Aging 20: 531–535. Parrinnello et al. 2003 ?? Parsons, P. A. 1978. The genetics of aging in optimal and stressful environments. Exp. Gerontol. 13: 357–363. Parsons, P. A. 1995. Inherited stress resistance and longevity: A stress theory of aging. Heredity 75: 216–221. Parsons, P. A. 1996. The limit to human longevity: An approach through a stress theory of ageing. Mech. Ageing Dev. 87: 211–218. Parsons, P. A. 1997. Success in mating: a coordinated approach to fitness through genotypes incorporating genes for stress resistance and heterozygous advantage under stress. Behav. Genet. 27: 75–81. Parsons, P. A. 2000. Hormesis: An adaptive expectation with emphasis on ionizing radiation. J. Appl. Toxicol. 20: 103–112. Parsons, P. A. 2002. Life span: Does the limit to survival depend upon metabolic efficiency under stress? Biogerontology 3: 233–241. Parsons, P. A. 2003. From the stress theory of aging to energetic and evolutionary expectations for longevity. Biogerontology 4(2): 63–73. Partridge, L. and N. H. Barton. 1993. Optimality, mutation and the evolution of ageing. Nature 362: 305–311. Partridge, L. and D. Gems. 2002. Mechanisms of ageing: public or private? Nature Rev. Genet. 3: 165– 175. Partridge, L. and K. Fowler. 1992. Direct and correlated responses to selection on age at reproduction in Drosophila melanogaster. Evolution 46: 76–91.

572

References

Patel, N. V. and C. E. Finch. 2002. The glucocorticoid paradox of caloric restriction in slowing brain aging. Neruobiol. Aging 23: 707–717. Paterson, A. H., E. S. Lander, J. D. Hewitt, S. Peterson, S. E. Lincoln and S. D. Tanksley. 1988. Resolution of quantitative traits into Mendelian factors by using a complete linkage map of restriction fragment length polymorphisms. Nature 335: 721–726. Patti, M. E., A. J. Butte, S. Crunkhorn, K. Cusi, R. Berria, et al. 2003. Coordinated reduction of genes of oxidative metabolism in humans with insulin resistance and diabetes: Potential role of PGC1 and NRF1. Proc. Natl. Acad. Sci. USA 100: 8466– 8471. Pawson, T. and T. M. Saxton 1999. Signaling networks: Do all roads lead to the same gene. Cell 97: 675– 678. Pearl, R. 1922. A comparison of the laws of mortality in Drosophila and in man. Am. Nat. 61: 398–405. Pearl, R. 1928. The Rate of Living. University of London Press, London. Pearl, R. and J. R. Miner. 1935. Experimental studies on the duration of life. XIV. The comparative mortality of certain lower organisms. Q. Rev. Biol. 10: 60–79. Pedigo, N. W., Jr. 1994. Neurotransmitter receptor plasticity in aging. Life Sci. 55: 1985–1991. Pelleymounter, M. A., M. J. Cullen, M. B. Baker, R. Hecht, D. Winters, T. Boone and F. Collins. 1995. Effects of the obese gene product on body weight regulation in ob/ob mice. Science 269: 540–543. Pereira-Smith, O. M. and J. R. Smith. 1983. Evidence for the recessive nature of cellular immortality. Science 221: 964–966. Pereira-Smith, O. M. and Y. Ning. 1992. Molecular genetic studies of cellular senescence. Exp. Gerontol. 27: 519–522. Perez, G. J., R. Robles, C. M. Knudson, J. A. Flaws, S. J. Korsmeyer and J. L. Tilly. 1999. Prolongation of ovarian lifespan into advanced chronological age by Bax-deficiency. Nature Genet. 21: 200–203. Perillo, N. L., R. L. Walford, M. A. Newman and R. B. Effros. 1989. Human T lymphocytes possess a limited in vitro life span. Exp. Gerontol. 24: 177– 187. Perls, T. T., K. Bochen, M. Freeman, L. Alpert and M. H. Silver. 1999. Validity of reported age and centenarian prevalence in New England. Age Ageing 28: 193–197. Perls, T. T., L. Kunkel and A. Puca. 2002a. The genetics of aging. Curr. Opin. Genet. Dev. 12: 362–369. Perls, T. and D Terry. 2003a. Genetics of exceptional longevity. Exp. Gerontol. 38: 725–730. Perls, T. T. and D. Terry. 2003b. Understanding the deteminants of exceptional longevity. Ann. Intern. Med. 139: 445–449.

Perls, T. T., D. F. Terry, M. Silver, M. Shea, J. Bowen, E. Joyce, S. B. Ridge, R. Fretts, et al. 2000. Centenarians and the genetics of longevity. Pp. xx–xx in S. Hekimi (ed.), Results and Problems in Cell Differentiation, vol. 29, The Molecular Genetics of Aging. Springer-Verlag, Berlin. Perls, T. T., J. Wilmoth, R. Levenson, M. Drinkwater, M. Cohen, H. Bogan, E. Joyce, S. Brewster, L Kunkel and A Puca. 2002b. Life-long sustained mortality advantage of siblings of centernarians. Proc. Natl. Acad. Sci. USA 99: 8442–8447. Perls, T. T. and E. R. Wood. 1996. Acute care costs of the oldest old: They cost less, their care intensity is less and they go to nonteaching hospitals. Arch. Internal Med. 156: 754–760. Petersen, J. L. 1998. Global implications of extending the life span. J. Anti-Aging Med. 1: 67–90. Petersen, K. F., D. Befroy, S. Dufour, J. Dziura, C. Ariyan, D. L. Rothman, L. DiPeitro, G. W. Cline and G. O. Shulman. 2003. Mitochondrial dysfunction in the elderly: Possible role in insulin resistance. Science 300: 1140–1142. Petersen, S., G. Saretzki and von Zglincki. 1998. Preferential accumulation of single-stranded regions in telomeres of human fibroblasts. Exp. Cell Res. 239: 152–160. Peterson, P. G. 2004. Running On Empty; How the Democratic and Republican Parties Are Bankrupting Our Future and What Americans Can Do About It. Farrar, Straus and Giroux, New York. Petkov, V. D., V. V. Petkov and S. L. Stancheva. 1988. Age-related changes in brain neurotransmission. Gerontology 34: 14–21. Phair, J. P. 1983. Host defense in the aged. Pp. 1–12 in R. A. Gleckman and N. M. Gantz (eds.), Infections in the Elderly. Little Brown, Boston. Phelan, J. 1992. Genetic variability and aging models. Exp. Gerontol. 27: 147–159. Phelan, J. P. and S. N. Austad. 1994. Selecting animal models of human aging: inbred strains often exhibit less biological uniformity than F1 hybrids. J. Gerontol. 49: B1–11. Phillips, J. P., Campbell, Michaud, Charbonneau and A. J. Hilliker. 1989. Null mutations of copper/zinc superoxide dismutase in Drosophila confers hypersensitivity to paraquat and reduced longevity. Proc. Natl. Acad. Sci. USA 86: 2761–2765. Phillips, J. P. and A. J. Hilliker. 1990. Genetic analysis of oxygen defense mechanisms in Drosophila melanogaster. Adv. Genet. 28: 43–71. Phillips, J. P., T. L. Parkes and A. J. Hilliker. 2000. Targeted neuronal gene expression and longevity in Drosophila. Exp. Gerontol. 35: 1157–1164. Pierpaoli, W. and W. Regelson. 1994. Pineal control of aging: Effect of melatonin and pineal grafting in aging mice. Proc. Natl. Acad. Sci. USA 91: 787– 791.

References

Pierpaoli, W. and W. Regelson. 1995. The Melatonin Miracle. Simon and Schuster, New York. Pikarsky, E., R. M. Porat, H. Stein, R. Abramovitch, S. Amit, S. Kasem, E. Gutkorich-Pyest, S. UrieliShoval, E. Galun and Y. Ben-Neriah. 2004. NfkB functions as a tumour promoter in inflammationassociated cancer. Nature 431: 461–466. Piper, P. W., G. W. Jones, D. Bringloe, N. Harris, M. MacLean and M. Mollapour. 2002. The shortened replicative life span of prohibitin mutants of yeast appears to be due to defective mitochondrial segregation in old mother cells. Aging Cell 1: 149– 157. Pires-da Silva, A. and R. J. Sommer. 2003. The evolution of signaling pathways in animal development. Nature Rev. Genet. 4: 39–49. Pi-Sunyer, X. 2003. A clinical view of the obesity problem. Science 299: 859–860. Pitskhelauri, G. Z. 1982. The Longliving of Soviet Georgia (trans. and ed. G. Lesneff-Caravaglia). Human Sciences Press, New York. Pletcher, S. D., S. J. Macdonald, R. Marguerie, U. Certa, S. C. Stearns, D. B. Goldstein and L. Partridge. 2002. Genome-wide transcript profiles in aging and calorically restricted Drosophila melanogaster. Curr. Biol. 12: 712–723. Plomin, R., N. L. Pedersen, P. Lichtenstein and G. E. McClearn. 1994. Variability and stability in cognitive abilities are largely genetic later in life. Behav. Genet. 24: 207–215. Poehlman, E. T., A. Turturro, N. Bodkin, W. Cefalu, S. Heymsfield, J. Holloszy and J. Kemnitz. 2001. Caloric restriction mimetics: Physical activity and body composition changes. J. Gerontol. 56A: 45– 54 (special issue). Pohley, H. J. 1987. A formal mortality analysis for populations of unicellular organisms (Saccharomyces cerevisiae). Mech. Ageing Dev. 38: 231–243. Pohof, J., G. van Haaften, K. Thijssen, R. S. Kamath, A. G. Fraser, J. Ahringer, R. H. A. Pasterk and M. Jijsterman. 2003. Identification of genes that protect the C. elegans genome against mutations by genome-wide RNAi. Genes Dev. 17: 443–448. Poirier, J. et al. 1995. Apolipoprotein E4 allele as a predictor of cholinergic deficits and treatment outcome in Alzheimer disease. Proc. Natl. Acad. Sci. USA 92: 12260–12264. Polson, C. D. A. and G. C. Webster. 1982. Agerelated DNA fragmentation in two varieties of Drosophila melanogaster, Phaseolus (cotyledons) and three tissues of the mouse. Exp. Gerontol. 17: 11–17. Popp, R. A., E. G. Bailiff, G. P. Hirsch and R. A. Conrad. 1976. Errors in human hemoglobin as a function of age. Interdiscipl. Topics Gerontol. 9: 209–218. Porter, M. B. and J. R. Smith. 1988. Generation of

573

monoclonal antibodies reacting specifically with senescent cells. Gerontologist 28: 230A (abstract). Post, S. G. 2004. Decelerated aging: Should I drink from a fountain of youth? Pp. x–x in S. G. Post and R. H. Binstock (eds), The Fountain of Youth: Cultural, Scientific, and Ethical Perspectives on a Biomedical Goal. Oxford University Press, New York. Poulsen, H. E., S. Loft and K. Vistisen. 1996. Extreme exercise and oxidative DNA modification. J. Sports Sci. 14: 343–346. Promislow, D. E. L. 1993. Minireview: On size and survival: Progress and pitfalls in the allometry of life span. J. Gerontol. Biol. Sci. 48: B115–B123. Promislow D. E. L. 2004. Protein networks, pleiotropy and the evolution of senescence. Proc. R. Soc. Lond. B. http://www.journals.royalsoc.ac.uk/DOI 10.1098/rspb.2004.2732. Promislow D. E. and S. D. Pletcher 2002. Advice to an aging scientist. Mech Ageing Dev. 123: 841– 850. Prothero, J. and K. D. Jurgens. 1987. Scaling of maximal lifespan in mammals: A review. Pp. 44–74 in A. D. Woodhead and K. H. Thompson (eds.), Evolution of Longevity in Animals: A Comparative Approach. Plenum Press, New York. Proust, J., R. Moulias, F. Fumeron, F. Bekkhoucha, M. Busson, M. Schmid and J. Hors. 1982. HLA and longevity. Tissue Antigens 19: 168–173. Prowse, K. R. and C. W. Greider. 1995. Developmental and tissue-specific regulation of mouse telomerase and telomere length. Proc. Natl. Acad. Sci. USA 92: 4818–4822. Pryde, F. E., H. C. Gorham and E. J. Lowis. 1997. Chromosome ends: All the same under their caps. Curr. Opin. Genet. Dev. 7: 822–828. Ptashne, M. 1988. How eucaryotic transcriptional activators work. Nature 335: 683–689. Ptashne, M. and A. Gann. 2002. Genes & Signals. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Pu, M-P., T. Flatt and M. Tatar. 2005. Juvenile and steroid hormones in Drosophila melanogaster longevity. Pp. x–x in E. J. Masoro and S. N. Austad (eds.), Handbook of the Biology of Aging, 6th ed. In press. Puca, A. A., M. J. Daly, S. J. Brewster, T. C. Matise, J. Barrett, M. Shea-Drinkwater, S. Kang, E. Joyce, E. Benson, L. M. Kunkel and T. Perls. 2001. A genome-wise scan for linkage to human exceptional longevity identifies a locus on chromosome 4. Proc. Natl. Acad. Sci. USA 98: 10505–10508. Quinn, J. F. and M. Kozy. 1996. The role of bridge jobs in the retirement transition: Gender, race and ethnicity. Gerontologist 36: 363–372. Raff, M. C. 1992. Social controls on cell survival and cell death. Nature 356: 397–400.

574

References

Ramirez, R. D., W. E. Wright, J. W. Shay and R. S. Taylor. 1997. Telomerase activity concentrates in the mitotically active segments of human hair follicles. J. Invest. Dermatol. 108: 113–117. Raven, P. H. and G. B. Johnson. 1985. Biology. Times Mirror/Mosby, St. Louis, MO. Rea, S. and T. E. Johnson. 2003. A metabolic model for life span determination in Caenorhabditis elegans. Dev. Cell 5: 197–203. Rebar, R. W. and T. B. Spitzer. 1987. The physiology and measurement of hot flashes. Am. J. Obstetr. Gynecol. 156: 1284–1288. Reed, D. M., D. J. Foley, L. R. White, H. Heimovitz, C. M. Burchfiel and K. Masaki, 1998. Predictors of healthy aging in men with high life expectancies. Am. J. Public Heath 88: 1463–1468. Reff, M. E. and E. L. Schneider (eds.). 1982. Biological Markers of Aging. NIH Publ. no. 82-2221, National Institutes of Health, Washington, DC. Reiter, R. J. 1995. The pineal gland and melatonin in relation to aging: A summary of the theories and of the data. Exp. Gerontol. 30: 199–212. Rejeski, J. and S. L. Mihalko, 2001. Physical activity and quality of life in older adults. J. Gerontol. 56A (special issue II): 23–35. Reppert, S. M. and D. R. Weaver. 1995. Melatonin madness. Cell 83: 1059–1062. Reveillaud, I., J. Phillips, B. Duyf, A. Hilliker, A. Kongpachith and J. E. Fleming. 1994. Phenotypic rescue by a bovine transgene in a Cu/Zn superoxide dismutase-null mutant of Drosophila melanogaster. Mol. Cell. Biol. 14: 1302–1307. Reynolds, T. H. IV, K. M. Krajewski, L. M. Larkin, P. Reid, J. B. Halter, M. A. Supiano and D. R. Dengel. 2002. Effect of age on skeletal muscle proteolyis in extenson digitorum longus muscles of B6C3F1 mice. J. Gerontol. Biol. Sci. 57A: B198–B201. Reznick, D. N., M. J. Bryant, D. Roff, C. K. Ghalambor and D. E. Ghalambor. 2004. Effect of extrinsic mortality on the evolution of senescence in guppies. Nature 431: 1095–1099. Rice, T., G. P. Vogler, L. Perusse, C. Bouchard and D. C. Rao. 1989. Cardiovascular risk factors in a French Canadian population: Resolution of genetic and familial environmental effects on blood pressure using twins, adoptees, and extensive information on environmental correlates. Genet. Epidemiol. 6: 571–588. Richardson, A. and M. C. Birchenall-Sparks. 1983. Age-related changes in protein synthesis. Pp. 255– 274 in M. Rothstein (ed.), Review of Biological Research in Aging, vol. 1. Alan R. Liss, New York. Richardson, A., J. A. Butler, M. S. Rutherford, I. Semsei, M. Z. Gu, G. Fernandes and W. H. Chiang. 1987. Effect of age and dietary restriction on the expres-

sion of alpha 2m-globulin. J. Biol. Chem. 262: 12821–12825. Richardson, A. and H. T. Cheung. 1982. The relationship between age-related changes in gene expression, protein turnover and the responsiveness of an organism to stimuli. Life Sci. 31: 605–613. Richardson, A. and M. A. Pahlavani. 1994. Thoughts on the evolutionary basis of dietary restriction. Pp. 226–231 in M. R. Rose and C. E. Finch (eds.), Genetics and Evolution of Aging. Kluwer Academic Publishers, Dordrecht, the Netherlands. Richardson, A., M. S. Roberts and M. S. Rutherford. 1985. Aging and gene expression. Pp. 395–419 in M. Rothstein (ed.), Review of Biological Research in Aging, vol. 2. Alan R. Liss, New York. Richardson, A. and I. Semsei. 1987. Effect of ageing on translation and transcription. Pp. 467–483 in M. Rothstein (ed.), Review of Biological Research on Aging. Alan R. Liss, New York. Richel, T. 2003. Will human life expectancy quadruple in the next hundred years? Sixty gerontologists say public debate on life extension is necessary. J. Anti-Aging Med. 6: 309–314. Ricklefs, R. E. and C. E. Finch, 1995. Aging: A Natural History. Scientific American Library, New York. Riddiford, L. M., P. Cherbas and J. W. Truman. 2000. Ecdysone receptors and their biological actions. Vit. Hormones 60: 1–73. Riddle, D. L., M. M. Swanson and P. S. Albert. 1981. Interacting genes in nematode dauer larva formation. Nature 290: 268–271. Ridley, R. M. and H. F. Baker. 1993. Behavioral effects of cholinergic grafts. Ann. N.Y. Acad. Sci. 695: 274–277. Riggs, B. L. and L. J. Melton. 1986. Involutional osteoporosis. N. Engl. J. Med. 314: 1676–1686. Riha, V. F. and L. S. Luckinbill. 1996. Selection for longevity favors stringent metabolic control in Drosophila melanogaster. J. Gerontol. Biol. Sci. 51A: B284–B294. Rimm, E. B., M. J. Stampfer, A. Ascherio, E. Giovannucci, G. A. Colditz and W. C. Willet. 1993. Vitamin E consumption and the risk of coronary heart disease in men. N. Eng. J. Med. 328: 1450–1456. Rinkevitch, B., R. J. Lauzon, B. W. M. Brown and I. L. Weissman. 1992. Evidence for a programmed life span in a colonial protochordate. Proc. Natl. Acad. Sci. USA 89: 3546–3550. Rivett, A. J., S. Bose, A. J. Pemberton, P. Brooks, D. Onion, D. Shirley, F. L. L. Stratford and K. Forti. 2002. Assays of proteasome activty in relation to aging. Exp. Gerontol. 37: 1217–1222. Robert, A. M., M. Schaeverbeke, J. Schaeverbeke and L. Robert. 1997. Viellissement et circulation cerebrale: Role de la matrice extracellulaire des microvaisseaux du cerveau. Comptes Rendus des

References

Seances de la Societe de Biologie et de Ses Filiales 191: 253–260. Robert, C., B. Lesty and A. M. Robert. 1988. Ageing of the skin: Study of elastic fiber network modifications by computerized image analysis. Gerontology 34: 291–296. Roberts, L. 1988. Questions raised about anti-wrinkle cream. Science 239: 564. Robertson, M. 1987. Molecular genetics of the mind. Nature 325: 755. Robine, J. M. 2001. A new biodemographic model to explain the trajectory of mortality. Exp. Gerontol. 36: 899–914. Robine, J. M. and Y. Saito. 2003. Survival beyond age 100: The case of Japan. Pp. 208–228 in J. R. Carey and S. Tuljapurkar (eds.), Life Span: Evolutionary, Ecological, and Demographic Perspectives Population Council, New York. Robine, J. M. and J. W. Vaupel. 2001. Supercentenarians: Slower ageing individuals or senile elderly? Exp. Gerontol. 36: 915–930. Robinson, S. M. and D. J. P. Barker. 2002. Coronary heart disease: A disorder of growth. Proc. Nutr. Soc. 61: 537–542. Robinson, T. F., S. M. Factor and E. H. Sonnenblick. 1986. The heart as a suction pump. Sci. Am. 254(6): 84–91. Rodeheffer, R. J., G. Gerstenblith, L. C. Becker, J. L. Fleg, M. L. Weisfeldt and E. G. Lakatta. 1984. Exercise cardiac output is maintained with advancing age in healthy human subjects: Cardiac dilation and increased stroke volume compensate for a diminished heart rate. Circulation 69: 203–213. Rodin, J. 1986. Aging and health: Effects of the sense of control. Science 233: 1271–1276. Rogers, J. and F. E. Bloom. 1985. Neurotransmitter metabolism and function in the aging central nervous system. Pp. 645–691 in C. E. Finch and E. L. Schneider (eds.), Handbook of the Biology of Aging, 2nd edition. Van Nostrand Reinhold, New York. Rogina, B. and S. L. Helfand. 1996. Timing of expression of a gene in the adult Drosophila is regulated by mechanisms independent of termperature and metabolic rate. Genetics 143: 1643–1651. Rogina, B. and S. L. Helfand. 2000. CuZn superoxide dismutase deficiency accelerates the time course of an age-related marker in Drosophila melanogaster. Biogerontology 1: 163–169. Rogina, B., S. L. Helfand and S. Frankel. 2002. Longevity regulation by Drosophila Rpd3 deacetylase and caloric restriction. Science 298: 1745–174x. Rogina, B., R. A. Reenan, S. P. Nilsen and S. L. Helfand. 2000. Extended life span conferred by cotransporter gene mutations in Drosophila. Science 290: 2137–2140 Roitberg, B. D., J. Sircom, C. A. Roitberg, J. J. M. van

575

Alphen and M. Mangel. 1993. Life expectancy and reproduction. Nature 364: 108. Rose, G. M. 1999. Behavioral effects of neurotrophic factor supplementation in aging. Age 22: 1–7. Rose, G., G. Passarino, G. Carrieri, K. Altomare, V. Greco, S. Bertolini, M. Bonafe, C. Francheschi and G. De Benedictis. 2001. Paradoxes in longevity: sequence analysis of mtDNA haplogroup J in centenarians. Eur. J. Human Genetics 9: 701–707. Rose, M. R. 1984. Laboratory evolution of postponed senescence in Drosophila melanogaster. Evolution 38: 1004–1010. Rose, M. R. 1991. Evolutionary Biology of Aging. Oxford University Press, New York. Rose, M. R. 1996. Genetic analysis of mechanisms of aging. Curr. Opin. Genet. Dev. 6: 366–370. Rose, M. R. 2004. The metabiology of life extension. Pp. x–x in S. G. Post and R. H. Binstock (eds.), The Fountain of Youth: Cultural, Scientific, and Ethical Perspectives on a Biomedical Goal. Oxford University Press, New York. Rose, M. R. and B. Charlesworth. 1981. Genetics of life history in Drosophila melanogaster. II. Exploratory selection experiments. Genetics 97: 187–196. Rose, M. R., L. N. Vu, S. U. Park and J. L. Graves, Jr. 1992. Selection on stress resistance increases longevity in Drosophila melanogaster. Exp. Gerontol. 27: 241–250. Rosenberg, M. B., T. Friedmann, R. C. Robertson, M. Tuszynski, J. A. Wolff, X. O. Breakefield and F. H. Gage. 1988. Grafting genetically modified cells to the damaged brain: Restorative effects of NGF expression. Science 242: 1575–1577. Rosenblum, J. S., N. B. Gilula and R. A. Lerner. 1996. On signal sequence polymorphisms and diseases of distribution. Proc. Nat. Acad. Sci. USA 93: 4471–4473. Rosenthal, G. E. and C. S. Landefeld. 1993. Do older Medicare pateints cost hospitals more? Evidence from an academic medical center. Arch. Int. Med. 153: 89–96. Ross, R. E. 2000. Age specific decreases in aerobic efficiency associated with increase in oxygen free radical production in Drosophila melanogaster. J. Insect Physiol. 46: 1477–1480. Rossle, R. and F. Roulet. 1932. Zahl und Mass in Pathologie. Springer-Verlag, Berlin. Rossman, I. 1977. Anatomic and body composition changes with aging. Pp. 189–221 in C. E. Finch and L. Hayflick (eds.), Handbook of the Biology of Aging. Van Nostrand Reinhold, New York. Rossman, I. 1979. Clinical Geriatrics, 2nd ed. Lippincott, Philadelphia. Roth, G. S. and J. A. Joseph. 1994. Cellular and molecular mechanisms of impaired dopaminergic function during aging. Ann. N.Y. Acad. Sci. 719: 129–135.

576

References

Roth, G. S., M. A. Lane, D. K. Ingram, J. A. Mattison, D. Elahi, J. D. Tobin, D. Muller and E. J. Metter. 2002. Biomarkers of caloric restriction may predict longevity in humans. Science 297: 811. Rothermel, G. A., J. L. Thornton and R. A. Butow. 1997. Rtg3p, a basic helix-loop-helix/leucine zipper protein that functions in mitochondrialinduced changes in gene expression, contains independent activation domains. J. Biol. Chem. 272: 19801–19807. Rothstein, M. 1982. Biochemical Approaches to Aging. Academic Press, New York. Rothstein, M. 1983. Detection of altered proteins. Pp. 1–8 in R. C. Adelman and G. S. Roth (eds.), Altered Proteins and Aging, CRC Press, Boca Raton, FL. Rothstein, M. 1987. Evidence for and against the error catastrophe hypothesis. Pp. 139–154 in H. R. Warner, R. N. Butler, R. L. Sprott and E. L. Schneider (eds.), Aging, vol. 31, Modern Biological Theories of Aging. Raven Press, New York. Rowe, J. W., R. Andres, J. D. Tobin, A. H. Norris and N. W. Shock. 1976. The effect of age on creatinine clearance in men: A cross-sectional and longitudinal study. J. Gerontol. 31: 155–163. Rowe, J. W. and R. L. Kahn. 1987. Human aging: Usual and successful. Science 237: 143–149. Rowe, J. W. and R. L. Kahn. 1997. Successful aging. Gerontologist 37: 433–440. Rowe, J. W. and K. L. Minaker. 1985. Geriatric medicine. Pp. 932–959 in C. E. Finch and E. L. Schneider (eds.), Handbook of the Biology of Aging, 2nd ed. Van Nostrand Reinhold, New York. Rowley, M. J., H. Buchanan and I. R. Mackay. 1968. Reciprocal change with age in antibody to extrinsic and intrinsic allergens. Lancet 2: 24–26. Roy, A. K. and B. Chatterjee. 1984. Hormonal regulation of hepatic gene expression during aging. Pp. 143–166 in A. K. Roy and B. Chatterjee (eds.), Molecular Basis of Aging. Academic Press, Orlando, FL. Rubin, H. 2002. Promise and problems in relating cellular senescence in vitro to aging in vivo. Arch. Gerontol. Geriatr. 34: 275–286. Rubner, M. 1908. Das Problem der Levensdauer und seine Beziehungen zum Wachstom und Ernahrung Oldenbourg, Munich. Rudan, I., N. Smolej-Narancic, H. Campbell, A. Carothers, A. Wright, B. Janicijevic and P. Rudan. 2003. Inbreeding and the genetic complexity of human hypertension. Genetics 163: 1011–1021. Rudman, D, A. G. Feller, H. S. Nagraj, G. A. Gergans, P. Y. Lalitha, A. F. Goldberg, R. A. Sclenker, L. Cohn, L. W. Rudman and D. E. Mattson. 1990. Effects of human growth homone in men over 60 years old. N. Engl. J. Med. 323: 1–6. Rudman, D. and K. Shetty. 1994. Unanswered ques-

tions concerning the treatment of hyposomatorpism and hypogonadism in elderly men. J. Am. Geriatr. Soc. 42: 522–527. Rueppell, O., G. V. Amdam, R. E. Page Jr. and J. R. Carey. 2004. From genes to societies. Sci. Aging. Knowl. Environ., February. http://ageke .sciencemag.org/cai/content/full/2004/5/pe5. Ruhe, R. C., D. L. Curry and R. B. McDonald. 1997. Altered cellular heterogeneity as a possible mechanism for the maintenance of organ function in senescent animals. J. Gerontol. Biol. Sci. 52A: B53–B58. Ruhe, R. C. and R. B. McDonald. 1994. Aging, insulin secretion and cellular senescence. Pp. 93–110 in R. Watson (ed.), Handbook of Nutrition in the Aged, 2nd ed. CRC Press, Boca Raton, FL. Ruiz-Larrea, M. B., A. M. Leal, C. Martin, R. Martinez and M. Lacort. 1997. Antioxidant action of estrogens in rat hepatocytes. Rev. Esp. Fisol. 53: 225– 230. Rulifson E. J., S. K. Kim and R. Nusse. 2002. Ablation of insulin-producing neurons in flies: growth and diabetic phenotypes. Science 296: 1118– 1120. Russell, R. L. and L. A. Jacobson. 1985. Some aspects of aging can be studied easily in nematodes. Pp. 128–145 in C. E. Finch and E. L. Schneider (eds.), Handbook of the Biology of Aging, 2nd ed. Van Nostrand Reinhold, New York. Rustin, P., J-C. von Kleist-Retzow, Z. vajo, A. Rotig and A. Munnich. 2000. For debate: defective mitochondria, free radicals, cell death, ageing-reality or myth-ochondria? Mech. Ageing Dev. 114: 201– 206. Rutter, G. A. and R. Rizzuto. 2000. Regulation of mitochondrial metabolism by ER Ca2+ release: an intimate connection. Trends Biol. Sci. 25: 215– 221. Ryan, J. M., D. G. Ostrow, X. O. Breakefield, E. S. Gershon and L. Upchurch. 1981. A comparison of the proliferative and replication lifespan kinetics of cell cultures derived from monozygotic twins. In Vitro Cell. Dev. Biol. 17: 20–27. Sacher, G. A. 1959. Relation of lifespan to brain weight and body weight in mammals. Pp. 115–133 in G. E. W. Wolstenholme and M. O’Conner (eds.), The Life Span of Animals (CIBA Foundation Colloquium on Aging, vol. 5). Churchill, London. Sacher, G. A. 1975. Maturation and longevity in relation to cranial capacity in hominid evolution. Pp. 417–430 in R. Tuttle (ed.), Antecedents of Man and After. Primates: Functional Morphology and Evolution, vol. 1. Mouton, The Hague. Sacher, G. A. 1977. Life table modification and life prolongation. Pp. 582–638 in C. E. Finch and L. Hayflick (eds.), Handbook of the Biology of Aging. Van Nostrand Reinhold, New York.

References

Sacher, G. A. 1978. Longevity and aging in vertebrate evolution. BioScience 28: 497–501. Sacher, G. A. and R. W. Hart. 1978. Longevity, aging, and comparative cellular and molecular biology of the house mouse, Mus musculus, and the whitefooted mouse, Peromyscus leucopus. Pp. 71–96 in D. Bergsma and D. E. Harrison (eds.), Genetic Effects on Aging. Alan R. Liss, New York. Saksela, E. and P. S. Moorhead. 1963. Aneuploidy in the degenerative phase of serial cultivation of human cell strains. Proc. Natl. Acad. Sci. USA 50: 390–395. Salen. 2000. Salk, D., K. An, H. Hoehn and G. M. Martin. 1981. Cytogenetics of Werner’s syndrome cultured skin fibroblasts: Variegated translocation mosaicism. Cytogenet. Cell Genet. 30: 92–107. Salmon, TB, BA Evert, B Song and PW Doetsch. 2004. Biological consequences of oxidative-stress induced DNA damage in Saccharomyces cerevisiae. Nucleic Acids Res. 32: 3712–3723. Samaha, F. F., et al. 2003. A low carbohydrate as compared with a low fat diet in severe obesity. N. Engl. J. Med. 348: 2074–2081. Samani, N. J., R. Boultby, R. Butler, J. R. Thompson and A. H. Goodall. 2001. Telomere shortening in atherosclerosis. Lancet 358: 472–473. Samiy, A. H. 1983. Renal disease in the elderly. Med. Clin. N. Am. 3: 463–480. Samper, E., D. G. Nicholls and S. Melov. 2003. Mitochondrial oxidative stress causes chromosomal instability of mouse embryonic fibroblasts. Aging Cell 2: 277–285. Samuels et al. 2004. Human PI3K isoloci. Sandberg, R., R. Yasuda, D. G. Pankratz, T. A. Carter, J. A. Del Rio, L. Wodicka, M. Marford, D. J. Lockhart and C. Barlow. 2000. Regional and strain-specific gene expression mapping in the adult mouse brain. Proc. Natl. Acad. Sci. USA 97: 11038–11045. Sandbrink, R., C. L. Masters and K. Beyreuther. 1996. APP gene family: Alternative splicing generates functionally related isoforms. Ann. N.Y. Acad. Sci. 777: 281–287. Sanderson, W. C. and S. Scherbov. 2005. Average remaining lifetimes can increase as human populations age. Nature 435: 811–813. Sapolsky, R. M. 1990. The adrenocortical axis. Pp. 330–348 in E. L. Schneider and J. W. Rowe (eds.), Handbook of the Biology of Aging, 3rd ed. Academic Press, San Diego, CA. Sapolsky, R. M. 2004. Organismal stress and telomeric aging: an unexpected connection. Proc. Natl. Acad. Sci. USA 101: 17323–17324. Sapolsky R. M., L. C. Krey and B. S. McEwen. 1984. Glucocorticoid-sensitive hippocampal neurons are involved in terminating the adrenocortical stress

577

response. Proc. Natl. Acad. Sci. USA 81: 6174– 6177. Sapolsky, R. M., L. C. Krey and B. S. McEwen. 1986. The neuroendocrinology of stress and aging: The glucocorticoid cascade hypothesis. Endocr. Rev. 7: 284–301. Sapolsky, R. M., J. H. Vogelman, N. Orentreich and J. Altmann. 1993. Senescent decline in serum dehydroepiandrosterone sulfate concentrations in a population of wild baboons. J. Gerontol. Biol. Sci. 48: B196–B200. Schacter, F., L. Faure-Delanef, F. Geunot, H. Rouger, P. Froguel, L. Leseuer-Ginot and D. Cohen. 1994. Genetic associations with human longevity at the APOE and ACE loci. Nature Genet. 6: 29–34. Schaie, K. W. 1965. A general model for the study of developmental problems. Psychol. Bull. 64: 92– 107. Scheibel, A. B. 1978. Structural aspects of the aging brain: Spine systems and the dendritic arbor. Pp. 353–373 in R. Katzman, R. D. Terry and K. L. Bick (eds.), Aging, vol. 1, Alzheimer’s Disease: Senile Dementia and Related Disorders. Raven Press, New York. Scheiner, S. M. 1993. Genetics and evolution of phenotypic plasticity. Annu. Rev. Ecol. Syst. 24: 35–68. Schellenberg, G. D. 1995. Genetic dissection of Alzheimer disease, a heterogeneous disorder. Proc. Natl. Acad. Sci. USA 92: 8852–8859. Schlenker, E. D. 1984. Nutrition in Aging. Times Mirror/Mosby, St. Louis, MO. Schlessinger, D and G Van Zant. 2001. Does functional depletion of stem cells drive aging? Mech. Ageing Dev. 122: 1537–1553. Schnebel, E. M. and J. Grossfield. 1988. Antagonistic pleiotropy: An interspecific Drosophila comparison. Evolution 42: 306–311. Schneider, E. L. 1985. Cytogenetics of aging. Pp. 357– 376 in C. E. Finch and E. L. Schneider (eds.), Handbook of the Biology of Aging, 2nd ed. Van Nostrand Reinhold, New York. Schneider, E. L. and J. A. Brody. 1983. Aging, natural death and the compression of mortality: Another view. N. Engl. J. Med. 309: 854–856. Schotta, G. A., Ebert and G. Reuter. 2003. SU(VAR)3-9 is a conserved key function in heterochromatic gene silencing. Genetics 117: 149– 158. Schriner, S. E., N. J. Linford, G. M. Martin, P. Treuting, C. E. Ogburn, M. Edmond, P. E. Coskun, W. Ladiges, N. Wolf, H. Van Remmen, D. C. Wallace and P. S. Rabinovitch. 2005. Extension of murine life span by overexpression of catalase targeted to mitochondria. Science 308: 1909–1911. Schwartz, A. G. and L. L. Pashko. 2004. Dehydroepiandrosterone, glucose-6-phosphate dehydrogenase, and longevity. Ageing Res. Rev. 3: 171–187.

578

References

Schwartz, J. E., H. S. Friedman, J. S. Tucker, C. Tomlinson-Kennedy, D. L. Wingard and M. H. Criqui. 1995. Sociodemographic and psychosocial factors in childhood as predictors of adult mortality. Am. J. Public Health 85: 1237–1245. Schwartz, R. S. 1995. Trophic factor supplementation: Effect on the age-associated changes in body composition. J. Gerontol.: Biol. Sci. 50A(special issue): 151–156. Schwartz, T. and H. D. Osiewacz. 1996. Telomere length does not change during senescence of the ascomycete Podospora anserina. Mutat. Res. 316: 193–199. Schwarze, S. R., R. Weindruch and J. M. Aiken. 1998a. Decreased mitochondrial RNA levels without accumulation of mitochondrial DNA deletions in aging Drosophila melanogaster. Mutat. Res. 382: 99–107. Schwarze, S. R., Weindruch R. and Aiken J.M. 1998b. Oxidative stress and aging reduce COX I RNA and cytochrome oxidase activity in Drosophila. Free Rad. Biol. & Med. 25: 740–747. Schwetz, B. A. 1995. Reproductive effects associated with dietary restriction. Pp 341–350 in R. W. Hart, D. A. Neumann and R. T. Robertson (eds.), Dietary Restriction: Implications for the Design and Interpretation of Toxicity and Carcinogenicity Studies. ILSI Press, Washington, DC. Seeman, T. and X. Chen, 2002. Risk and protective factors for physical functioning in older adults with and without chronic conditions: MacArthur studies of successful aging. J. Gerontol. Soc. Sci. 57B: S135–S144. Sehgal, A., A. Rothenfluh-Hilfiker, M. Hunter-Ensor, Y. Chen, M. P. Myers and M. W. Young. 1995. Rhythmic expression of timeless: A basis for promoting circadian cycles in period gene autoregulation. Science 270: 808–811. Seidel, J. C. 1995. The impact of obesity on health status: Some implications for health care costs. Int. J. Obesity Related Metab. Dis. 19(suppl. 6): S13– S16. Selkoe, D. J. 1997. Alzheimer’s disease: Genotypes, phenotypes and treatments. Science 275: 630–631. Selkoe, D. J. 2002. Alzheimer’s disease is a synaptic failure. Science 298: 789–791. Sell, D. R. and V. M. Monnier. 1995. Long lived proteins: Extracellular matrix (collagens, elastins, proteoglycans) and lens crystallins. Pp. 235–308 in E. J. Masoro (ed.), Handbook of Physiology, Section 11: Aging. Oxford University Press, New York. Sellem, C. H., C. Lemaire, S. Lorin, G. Dujardin and A. Sainsard-Chanet. 2004. Interaction between the oxa1 and rmp1 genes modulates respiratory complex assembly and life span in Podospora anserina. Genetics 169: 1379–1389.

Sen, S., G. Talukder and A. Sharma. 1987. Age-related alterations in human chromosome composition and DNA content in vitro during senescence. Biol. Rev. 62: 25–44. Seplaki, C. L., N. Goldman, M. Weinstein and Y-H. Lin. 2004. How are biomarkers related to physical and methal well-being? J. Gerontol. Biol. Sci. 59A: 201–217. Seroude, L., T. Brummel, P. Kapahi and S. Benzer. 2002. Spatio-temporal analysis of gene expression during aging in Drosophila. Aging Cell 1: 47–56. Service, P. M., E. W. Hutchinson and M. R. Rose. 1988. Multiple genetic mechanisms for the evolution of senescence in Drosophila melanogaster. Evolution 42: 708–716. Severino, J., R. G. Allen, S. Balin A. Balin and V. J. Cristofalo. 2000. Is beta-galactosidase staining a marker of senescence in vitro and in vivo? Exp. Cell Res. 257: 162–171. Seydel, C. 2002 Stay mellow, stay young. http://sageke .sciencemag.org/cgi/content/full/sageke;2002/11/ nf5. Shakespeare, W. 1600. As you like it: Act 2, Scene 7. P. 701 in S. Wells and G. Taylor (eds.), The Complete Oxford Shakespeare. Oxford University Press, Oxford, 1987. Shanley, B. P. and T. B. L. Kirkwood, 2000. Calorie restriction and aging: A life history analysis. Evolution 54: 740–750. Sharma, H. K. and M. Rothstein. 1980. Altered enolase in aged Turbatrix aceti results from conformational changes in the enzyme. Proc. Natl. Acad. Sci. USA 77: 5865–5868. Sharma, H. W., J. A. Sololoski, J. R. Perez, J. Y. Maltese, A. C. Sartorelli, C. A. Stein, G. Nichols, Z. Khaled, N. T. Telang and R. Narayanan. 1995b. Differentiation of immortal cells inhibits telomerase activity. Proc. Natl. Acad. Sci. USA 92: 12343–12346. Sharma, S. P., M. Sharma and R. Kakkar. 1995a. Methionine-induced alterations in the life span, antioxidant enzymes and peroxide levels in aging Zaprionus paravittiger (Diptera). Gerontology 41: 86–93. Sharom, J. R., D. S. Bellows and M. Tyers. 2004. From large networks to small molecules. Curr. Opin. Chem. Biol. 8: 81–90. Shay, J. 2005. Meeting report: the role of telomeres and telomerase in cancer. Cancer Res 65: 3513– 3515. Sheehy, M. R. J., J. G. Greenwood and D. R. Fielder. 1995. Lipofuscin as a record of “rate of living” in an aquatic poikilotherm. J. Gerontol. Biol. Sci. 50A: B322–B326. Shelton, D. N., E. Chang, P. S. Whittier, D. Choi and W. D. Funk. 1999. Microarray analysis of replicative senescence. Curr. Biol. 9: 939–945.

References

Shephard, R. A. 1982. Physiology and Bioichemistry of Exercise. Praeger, New York. Shepherd, J. C. W., U. Walldorf, P. Hug and W. J. Gehring. 1989. Fruit flies with additional expression of the elongation factor EF-1a live longer. Proc. Nat. Acad. Sci. USA 86: 7520–7521. Shigenaga, M. K., T. M. Hagen and B. N. Ames. 1994. Oxidative damage and mitochondrial decay in aging. Proc. Natl. Acad. Sci. USA 91: 10771– 10778. Shikama, N., R. Ackermann and C. Brack. 1994. Protein synthesis elongation factor EF-1 alpha expression and longevity in Drosophila melanogaster. Proc. Natl. Acad. Sci. USA 91: 4199–4203. Shimokawa, I., T. Fukuyama, K. Yanagihara-Outa, M. Tomita, T. Komatsu, Y. Higami, T. Tuchiya, T. Chiba and Y. Yamaza. 2003a. Effects of caloric restriction on gene expression in the arcuate nucleus. Neurobiol. Aging 24: 117–123. Shimokawa, I., Y. Higami, T. Tsuchiya, H. Otani, T. Komatsu, T. Chiba and H. Yamaza. 2003b. Lifespan extension by reduction of the growth homone-insulin-like signaling factor-1 axis: relation to caloric restriction. FASEB J. express article 10-.1096/fj.02-0819fje. Published online April 8, 2003. Shimokawa, I., B. P. Yu, Y. Higami, T. Ikeda and E. J. Masoro. 1993. Dietary restriction retards onset but not progression of leukemia in male F344 rats. J. Gerontol. Biol. Sci. 48: B68–B73. Shintani, T. and D. J. Klionsky. 2004. Autophagy in health and disease: a double-edged sword. Science 306: 990–995. Shock, N. W. 1972. Energy metabolism, caloric intake, and physical activity of the aging. Pp. 12–23 in L. A. Carlson (ed.), Nutrition in Old Age. Almqvist and Wiksell, Uppsala, Sweden. Shock, N. W. 1977. System regulation. Pp. 639–665 in C. E. Finch and E. L. Schneider (eds.), Handbook of the Biology of Aging. Van Nostrand Reinhold, New York. Shock, N. W. 1981. Indices of functional age. Pp. 270– 286 in D. Danon, N. W. Shock and M. Marois (eds.), Aging: A Challenge to Science and Society, vol. 1, Biology. Oxford University Press, New York. Shock, N. W. 1983. Aging of physiological systems. J. Chron. Dis. 36: 137–142. Shock, N. W. 1985. Longitudinal studies of aging in humans. Pp. 721–743 in C. E. Finch and E. L. Schneider (eds.), Handbook of the Biology of Aging, 2nd ed. Van Nostrand Reinhold, New York. Shock, N. W., R. Andres, A. H. Norris and J. D. Tobin. 1979. Patterns of longitudinal changes in renal function. Pp. 525–527 in H. Orimo, K. Shimada and D. Maede (eds.), Recent Advances in Geron-

579

tology. XI International Congress of Gerontology. Excerpta Medica, Amsterdam. Shock, N. W., R. C. Greulich, P. T. Costa, Jr., R. Andres, E. G. Lakatta, D. Arenberg and J. D. Tobin. 1984. Normal Human Aging: The Baltimore Longitudinal Study of Aging. NIH Publ. no. 84-2450, National Institutes of Health, Washington, DC. Shoffner, J. M., M. T. Lott, A. M. S. Leza, P. Seiber, S. W. Ballinger and D. C. Wallace. 1990. Myoclonic epilepsy and ragged red fiber disease (MERRF) is associated with a mitochondrial DNA tRNALYS mutation. Cell 61: 931–937. Shook, D. R., A. Brooks and T. E. Johnson. 1996. Mapping quantitative trait loci affecting life history traits in the nematode Caenorhabditis elegans. Genetics 142: 801–817. Short, R. A., D. D. Williams and D. M. Bowden. 1987. Cross-sectional evaluation of potential biological markers of aging in pigtailed macaques: Effects of age, sex and diet. J. Gerontol. 42: 644–654. Short, R. A., D. D. Williams and D. M. Bowden. 1994. Modeling biological aging in a nonhuman primate. Pp. 409–418 in A. K. Balin (ed.), Practical Handbook of Human Biologic Age Determination. CRC Press, Boca Raton, FL. Short, R. A., D. D. Williams and D. M. Bowden. 1997. Circulating antioxidants as determinants of the rate of biological aging in pigtailed macaques (Macaca nemestrina). J. Gerontol. Biol. Sci. 52A: B26–B38. Shoulson, I. 1992. Antioxidative therapeutic strategies for Parkinson’s disease. Ann. N.Y. Acad. Sci. 648: 37–41. Shrimpton, A. E. and A. Robertson. 1988. The isolation of polygenic factors controlling bristle score in Drosophila melanogaster. II. Distribution of third chromosome bristle effects within chromosome sections. Genetics 118: 445–459. Shukitt-Hale, B. 1999. The effects of aging and oxidative stress on psychomotor and cognitive behavior. Age 22: 9–17. Silverman, H. G. and R. S. Mazzeo. 1996. Hormonal responses to maximal and submaximal exercise in trained and untrained men of various ages. J. Gerontol. Biol. Sci. 51A: B30–B37. Simon, A. F., C. Shih, A. Mack and S. Benzer. 2003. Steroid control of longevity in Drosophila melanogaster. Science 299: 1407–1410. Sinclair, D. A. 2002. Paradigms and pitfalls of yeast longevity research. Mech. Ageing Dev. 123: 857– 867. Sinclair, D. A. and L. Guarente. 1997. Extrachromosomal rDNA circles: A cause of aging in yeast. Cell 91: 1033–1042. Sindzinski, J. A., L. J. Sweetlove and C. J. Leaver. 2002. A custom microarray analysis of gene ex-

580

References

pression during programmed cell death in Arabidopsis thaliana. Plant J. 30: 431–436. Singh, J. and A. K. Singh. 1979. Age-related changes in human thymus. Clin. Exp. Immunol. 37: 507– 511. Skulachev, V. P. 1996. Role of uncoupled and noncoupled oxidations in maintenance of safely low levels of oxygen and its one-electron reductants. Q. Rev. Biophys. 29: 169–202. Skulachev, V. P. 2001. Mitochondrial filaments and clusters as intracellular power-transmitting cables. Trends Biol. Sci. 26: 23–29. Slade, N. A. 1995. Failure to detect senescence in persistence of some grasslands rodents. Ecology 76: 863–870. Slonaker, J. R. 1912. The normal activity of the albino rat from birth to natural death, its rate of growth and duration of life. J. Anim. Behav. 2: 20–42. Small, W. C., R. D. Brodeur, A. Sandor, N. Fedorova, G. Li, R. A. Butow and P. A. Srere. 1995. Enzymatic and metabolic studies on retrograde regulation mutants of yeast. Biochemistry 34: 5569–5576. Smeal, T., J. Claus, B. Kennedy, F. Cole and L. Guarente. 1996. Loss of transcriptional silencing causes sterility in old mother cells of S. cerevisiae. Cell 84: 633– 642. Smith, D. E. W. 1993. Human Longevity. Oxford University Press, New York. Smith, D. E. W. and H. R. Warner. 1989. Does genotypic sex have a direct effect on longevity? Exp. Gerontol. 24: 277–288. Smith, D. E., J. Roberts, F. H. Gage and M. H Tuszynski. 1999. Age-associated neuronal atrophy occurs in the primate brain and is reversible by growth factor gene therapy. Proc. Natl. Acad. Sci. USA 96: 10893–10898. Smith, J. R. 1992. Inhibitors of DNA synthesis derived from senescent human diploid fibroblasts. Exp. Gerontol. 27: 409–412. Smith, J. R. 2003. Cell senescence: An evaluation of replicative senescence in culture as a model for aging in situ. J. Gerontol. Biol. Sci. 58A: 779–781. Smith, J. R. and O. M. Pereira-Smith. 1996. Replicative senescence: Implications for in vivo aging and tumor suppression. Science 273: 63–67. Smith, J. R., A. L. Spiering and O. Pereira-Smith. 1987. Is cellular senescence genetically programmed? Pp. 283–294 in A. D. Woodhead and K. H. Thompson (eds.), Evolution of Longevity in Animals: A Comparative Approach. Plenum Press, New York. Smith, J. R. and R. G. Whitney. 1980. Intraclonal variation in proliferative potential of human diploid fibroblasts: Stochastic mechanism for cellular aging. Science 207: 82–84. Smith, K. D. 1987. Paper presented to the 1987 Na-

tional Institutes of Health Conference on Gender and Longevity. Smith-Sonneborn, J. 1979. DNA repair and longevity assurance in Paramecium tetraurelia. Science 203: 1115–1117. Smith-Sonneborn, J. 1985. Aging in unicellular organisms. Pp. 79–104 in C. E. Finch and E. L. Schneider (eds.), Handbook of the Biology of Aging, 2nd ed. Van Nostrand Reinhold, New York. Smith-Sonneborn, J. 1990. Aging in protozoa. Pp. 24– 44 in E. L. Schneider and J. W. Rowe (eds.), Handbook of the Biology of Aging, 3rd ed. Academic Press, San Diego, CA. Snowden, D. A. 1997. Aging and Alzheimer’s Disease: Lessons from the nun study. Gerontologist 37: 150–156. Snowdon, D. A. 2003. Healthy aging and dementia: findings from the Nun Study. Ann. Intern. Med. 139: 450–454. Snowden, D. A., R. L. Dane, L. Beeson, G. L. Burke, M. Sprafka, J. Bitter, H. Iso, D. R. Jacobs and R. L. Phillips. 1989. Is early natural menopause a biologic marker of health and aging? Am. J. Public Health 79: 709–714. Sohal, R. S. 1983. Aging in insects. Pp. 595–632 in G. Kerkut and L. Gilbert (eds.), Comprehensive Insect Physiology, Biochemistry and Pharmacology. Pergamon Press, Oxford. Sohal, R. S. 1986. The rate of living theory: A contemporary interpretation. Pp. 23–44 in K. G. Collatz and R. S. Sohal (eds.), Insect Aging: Strategies and Mechanisms. Springer-Verlag, Berlin. Sohal, R. S., A. Agarwal and W. C. Orr. 1995a. Simultaneous overexpression of copper- and zinccontaining superoxide dismutase and catalase retards age-related oxidative damage and increases metabolic potential in Drosophila melanogaster. J. Biol. Chem. 270: 15671–15674. Sohal, R. S., S. Agarwal and B. H. Sohal. 1995b. Oxidative stress and aging in the Mongolian gerbil (Meriones unguiculatus). Mech. Ageing Dev. 81: 15–25. Sohal, R. S. and H. Donato. 1978. Effects of experimentally altered life span on the accumulation of fluorescent age pigment in the housefly, Musca domestica. Exp. Gerontol. 13: 335–341. Sohal, R. S., K. J. Farmer, R. G. Allen and N. R. Cohen. 1983. Effects of age on oxygen consumption, superoxide dismutase, catalase, glutathione, inorganic peroxides and chloroform-soluble antioxidants in the adult male housefly, Musca domestica. Mech. Ageing Dev. 24: 185–195. Sohal, R. S., P. L. Toy and R. G. Allen. 1986. Relationship between life expectancy, endogenous antioxidants and products of oxygen free radical reactions in the housefly, Musca domestica. Mech. Ageing Dev. 36: 71–77.

References

Sohal, R. S. and R. Weindruch. 1996. Oxidative stress, caloric restriction, and aging. Science 273: 59–63. Solomon, N. G. and J. G. Vandenbergh. 1994. Management, breeding, and reproductive performance of pine voles. Lab Anim Sci. 44: 613–617. Sonntag, W. E., K. L. Brunso-Bechtold and D. R. Riddle. 2001 Age-related decreases in growth hormone and insulin-like growth factor (IGF-1): Implications for brain aging. J. Anti-Aging Med. 4: 311–329. Sorlie, P., E. Rogot, R. Anderson, N. J. Johnson and E. Backlund. 1992. Black-white mortality differences by family income. Lancet 340: 346–350. Spence, A. P. 1988. Biology of Human Aging. PrenticeHall, Englewood Cliffs, NJ. Spence, J. C. 1921. Some observations on sugar tolerance with special reference to variations found at different ages. Q. J. Med. 4: 314–326. Spinage, C. A. 1972. African ungulate life tables. Ecology 53: 645–652. Spindler, S. 2003. Rapid identification of candidate CR mimetics using microarrays. Paper presented at 10th IABG meeting, Cambridge, UK. Available at http://www.gen.cam.ac.uk/iabg10/ppts/ Spindler.mp3. Sprott, R. 1991. Development of animal models of aging at the National Institute of Aging. Neurobiol. Aging 12: 635–638. Sprott, R. 1997. Mouse and rat genotype choices. Exp. Gerontol. 32: 79–86. Sprott, R. L. 1999. Biomarkers of aging. J. Gerontol. Biol. Sci. 54A: B464–B465. Sprung, C. N., L. Sabatier and J. P. Murname. 1996. Effect of telomere length on telomeric gene expression. Nucleic Acids Res. 24: 4336–4340. Squires, G. R., S. Okouneff, M. Ionescu and A. R. Poole. 2003. The pathobiology of focal lesion development in aging human articular cartilage and molecular matrix changes characteristic of osteoarthritis. Arthritis Rheum. 48: 1261–1270. Srgo, and L. Partridge. 1999. A delayed wave of death from reproduction in Drosophila. Science 286: 2521–2524. Srivastava, V. K., S. Miller, M. D. Schroeder, R. W. Hart and D. Busbee. 1993. Age-related changes in expression and activity of DNA polymerase alpha: Some effects of dietary restriction. Mutat. Res. 295: 265–280. St. George Hyslop, P. H. et al. 1987. The genetic defect causing familial Alzheimer’s disease maps on chromosome 21. Science 235: 885–890. Stahelin, H. B., M. Eichholzer, F. Gey and G. Brubacher. 1989. Antioxidant vitamins and mortality in the elderly: Results of the prospective Basle study. P. 201 in Proceedings of the XIV International Congress of Gerontology (abstract).

581

Stampfer, M. J., C. H. Hennekens, J. E. Manson, G. A. Colditz, B. Rosner and W. C. Willett. 1993. Vitamin E consumption and the risk of coronary disease in women. N. Engl. J. Med. 328: 1444– 1449. Starai, V. J., I. Celic, R. N. Cole, J. D. Boeke and J. C. Escalante-Semerena. 2002. Sir2-dependent activation of acetyl-CoA synthetase by deacetylation of active lysine. Science 298: 2390–2392. Starke-Reed, P. and C. N. Oliver. 1989. Protein oxidation and proteolysis during aging and oxidative stress. Arch. Biochem. Biophys. 275: 559–567. Stary, H. 1986. Evolution and progression of atherosclerosis in the coronary arteries of children and adults. Pp. 20–36 in S. R. Bates and E. C. Gangloff (eds.), Atherogenesis and Aging. Springer-Verlag, New York. Staveley, B. E., J. P. Philips and A. J. Hilliker. 1990. Phenotypic consequences of copper-zinc superoxide dismutase over-expression in Drosophila melanogaster. Genome 33: 867–872. Stearns, S. C. and M. Kaiser. 1993. The effects of enhanced expression of EF-1 alpha on life span in Drosophila melanogaster. IV. A summary of three experiments. Genetica 91: 167–182. Steger, R. W., et al. 1993. Premature ageing in transgenic mice expressing different growth hormone genes. J. Reprod. Fertil. (suppl. 46): 61–75. Steller, H. 1995. Mechanisms and genes of cellular suicide. Science 267: 1445–1449. Steingrimsson, E., N. G. Copeland and N. A. Jenkins. 2005. Melanocyte stem cell maintenance and hair graying. Cell 121: 9–12. Steinmetz, L. M., C. Scharfe, A. M. Deutschbauer, D. Mokranjac, Z. S. Herman, T. Jones, A. M. Chu, G. Giaever, H. Prokisch, P. J. Oefner and R. W. Davis. 2002. Systematic screen for human disease genes in yeast. Nat. Genet. 31: 400–404. Steller, H. 1995. Mechanisms and genes of cellular suicide. Science 267: 1445–1449. Stenmark, P., et al. 2001. A new member of the family of di-iron carboxylate proteins: Coq7 (clk-1), a membrane-bound hydrosylase involved in ubiquinone biosynthesis. J. Biol. Chem. xxx:xxx Sternberg, H. 1994. Aging of the immune system. Pp. 75–87 in P. S. Timiras (ed.), Physiological Basis of Aging and Geriatrics, 2nd edition. CRC Press, Boca Raton, FL. Stevens, J. J. Cai, E. R. Pamuk, D. F. Williamson, M. J. Thun and J. L. Wood. 1998. The effect of age on the association between body-mass index and mortality. N. Engl. J. Med. 338: 1–7. Stevnsner, T., T. Thorslund, N. C. de Souza-Pinto and V. A. Bohr. 2002. Mitochondrial repair of 8-oxoguanine and changes with aging. Exp. Gerontol. 37: 1189–1196. Steward, F. C. 1970. From cultured cells to whole

582

References

plants: The induction and control of their growth and morphogenesis. Proc. R. Soc. Loindon B 175: 1–30. Strehler, B. 1982. Time, Cells and Aging. Academic Press, New York. Strehler, B. 1986. Genetic instability as the primary cause of human aging. Exp. Gerontol. 21: 283– 319. Strother, S. 1988. The role of free radicals in leaf senescence. Gerontology 34: 151–156. Stuchlikova, E., M. Juricova-Herakiva and Z. Deyl. 1975. New aspects of the dietary effect of life prolongation in rodents. What is the role of density in aging? Exp. Gerontol. 10: 141–144. Subramanian, G., M. D. Adams, J. C. Venter and S. Broder. 2001. Implications of the human genome for understanding human biology and medicine. JAMA 286: 2296–2307. Sugawa, M., H. Coper, G. Schulze, I. Yamashina, F. Krause and N. A. Dencher. 1996. Impaired plasticity of neurons in aging: Biochemical, biophysical, and behavioral studies. Ann. N.Y. Acad. Sci. 786: 274–282. Sugita, K., N. Suzuki, K. Fujii and H. Nimi. 1995. Reduction of unscheduled DNA synthesis and plasminogen activator activity in HutchinsonGilford fibroblasts during passaging in vitro: Partial correction by interferon-beta. Mutat. Res. 316: 133–138. Sun, J., J. Molitor and J. Tower. 2004. Effects of simultaneous over-expression of CuZnSOD and MnSOD on Drosophila melanogaster life span. Mech Ageing Dev. 125: 341–349. Sun, J. and J. Tower. 1999. FLP recombinasemediated induction of CuZn-superoxide dismutase transgene expression can extend the life span of adult Drosophila melanogaster flies. Mol. Cell. Biol. 19: 216–228. Sun, J., D. Folk, T. J. Bradley and J. Tower. 2002. Induced over-expression of the mitochondrial Mnsuperoxide dismutase extends the life span of adult Drosophila melanogaster. Genetics 161: 661–672. Sun, J., S. P. Kale, A. M. Childress, C. Pinswadi and S. M. Jazwinski. 1994. Divergent roles of RAS1 and RAS2 in yeast longevity. J. Biol. Chem. 269: 18638–18645. Svanborg, A. and L. Selker. 1993. Postponement of aging-related disability. World Health Forum 14: 150–157. Sweeney, K. J. and A. S. Weiss. 1992. Hyaluronic acid in progeria and the aged phenotype? Gerontology 38: 139–152. Symphorien, S. and R. C. Woodruff. 2003. Effect of DNA repair on aging of transgenic Drosophila melanogaster: I. mei-41 locus. J. Gerontol. A. Biol. Sci. Med. Sci. 58: B782–787. Szakmary, A., S-M. Huang, D. T. Chang, P. A. Beachy

and M. Sander. 1996. Overexpression of a Rrp1 transgene reduces the somatic mutation and recombination frequency induced by oxidative DNA damage in Drosophila melanogaster. Proc. Natl. Acad. Sci. USA 93: 1607–1612. Szilard, L. 1959. On the nature of the aging process. Proc. Natl. Acad. Sci. USA 45: 30–45. Tabbarah, M., E. M. Crimmins and T. E. Seeman. 2002. The relationship between cognitive and physical performance: MacArthur studies of successful aging. J. Gerontol. Med. Sci. 57A: M228– M235. Takagi, Y., K. Isumi, H. Kinoshita, T. Yamada, K. Kaji and H. Tanabe. 1989. Identification of a gene that shortens clonal life span of Paramecium tetraurelia. Genetics 123: 749–754. Takahashi, N. and O. Smithies. 2004. Human genetics, animal models and computer simulations for studying hypertension. Trends Genet. 20: 136– 145. Takata, H., M. Suzuki, T. Ishii, S. Sekiguchi and H. Iri. 1987. Influence of major histocompatibility complex region genes on human longevity among Okinawan-Japanese centenarians and nonagenarians. Lancet 2(8563): 824–826. Takeda, T., M. Hosokawa, S. Takeshita, M. Irino, K. Higuchi, T. Matsushita, Y. Tomita, K. Yasuhira, H. Hamamoto, K. Shimizu, M. Ishii and T. Yamamuro. 1981. A new murine model of accelerated senescence. Mech. Ageing Dev. 17: 183–194. Takeoka, Y., S. Y. Chen, H. Yago, R. Boyd, S. Suehiro, L. D. Schultz, A. A. Ansari and M. E. Gershin. 1996. The murine thymic microenvironment changes with age. Int. Arch. Allergy Immunol. 111: 5–12. Talbert, G. B. 1977. Aging of the reproductive system. Pp. 318–356 in C. E. Finch and L. Hayflick (eds.), Handbook of the Biology of Aging. Van Nostrand Reinhold, New York. Tan, F-L., C. S. Moravec, J. Li, C. Apperson-Hansen, P. M. McCarthy, J. B. Young and M. Bond. 2002. The gene expression fingerprint of human heart failure. Proc. Natl. Acad. Sci. USA 99: 11387– 11392. Tanner, J. M. 1955. Growth at Adolescence. Blackwell, Oxford. Tanzi, R. E., J. F. Gusella, P. C. Watkins, G. A. Bruns, P. St. George-Hyslop, M. L. Van Keuren, D. Patterson, S. Pagan, D. M. Kurnit and R. L. Neve. 1987. Amyloid beta protein gene: cDNA, mRNA distribution and genetic linkage near the Alzheimer’s locus. Science 235: 880–884. Tappel, A. L., B. Fletcher and D. Deamer. 1973. Effect of antioxidants and nutrients on lipid peroxidation fluorescent products and ageing parameters in the mouse. J. Gerontol. 28: 415– 424.

References

Tatar, M. 2002. Regulation of aging by germline stem cells. Sci. Aging Knowl. Environ. http://sageke .sciencemag.org/cgi/content/full/sageke;2002/3/ pe2. Tatar, M. 2003. Unearthing loci that influence life span. Sci. Aging Knowl. Environ. http://sageke .sciencemag.org/cgi/content/full/sageke;2003/9/ pe5. Tatar, M., A. Bartke and A. Antebi. 2003. The endocrine regulation of aging by insulin-like signals. Science 299: 1346–1351. Tatar, M., A. Kopelman, D. Epstein, M. P. Tu, C. M. Yin and R. S. Garofalo. 2001. A mutant Drosophila insulin receptor homolog that extends life-span and impairs neuroendocrine function. Science 292: 107–110. Taube G. 2003. Insulin insults may spur Alzheimer’s disease. Science 301: 40–41. Teitelbaum, S. L. 2000. Bone resorption by osteoclasts. Science 289: 1504–1508. Tennyson, A. (Lord). 1842. Ulysses. Thomas, D. 1953. Do not go gentle into that good night. P. 262 in T. R. Cole and M. G. Winkler (eds.), The Oxford Book of Aging: Reflections on the Journey of Life. Oxford University Press, Oxford, 1994. Thomas, J. and D. Nyberg. 1988. Vitamin E supplementation and intense selection increase clonal life span Paramecium tetraurelia. Exp. Gerontol. 23: 501–512. Thompson, C. R. 1995. Apoptosis in the pathogenesis and treatment of disease. Science 267: 1456–1462. Thompson, J. N. 1975. Quantitative variation and gene number. Nature 258: 665–668. Tice, R. R. and R. B. Setlow. 1985. DNA repair and replication in aging organisms and cells. Pp. 173– 224 in C. E. Finch and E. L. Schneider (eds.), Handbook of the Biology of Aging, 2nd ed. Van Nostrand Reinhold, New York. Timiras, P. S. 1994. Aging of the adrenals and pituitary. Pp. 133–146 in P. S. Timiras (ed.), Physiological Basis of Aging and Geriatrics, 2nd ed. CRC Press, Boca Raton, FL. Timiras, P. S. 1995. Education, homeostasis and longevity. Exp. Gerontol. 30: 189–198. Tissenbaum, H. and L Guarente. 2001. Increased dosage of a sir-2 gene extends lifespan in Caenorhabditis elegans. Nature 410: 227–230. Tissenbaum, H. A., J. Hawdon, M. Perregaux, P. Hotez, L. Guarente and G. Ruvkun. 2001. A common muscarinic pathway for diapause recovery in the distantly related nematode species Caenorhabditis elegans and Ancylostoma caninum. Proc. Natl. Acad. Sci. USA 97: 460–465. Tobin, J. D. 1981. Physiological indices of aging. Pp. 286–294 in D. Danon, N. W. Shock and M. Marois (eds.), Aging: A Challenge to Science

583

and Society, vol. 1, Biology. Oxford University Press, New York. Tonna, E. A. 1977. Aging of skeletal-dental systems and supporting tissues. Pp. 470–495 in C. E. Finch and L. Hayflick (eds.), Handbook of the Biology of Aging. Van Nostrand Reinhold, New York. Torella, D., M. Rota, D. Nurzynnska, E. Musso, A. Monsen, I. Shiraishi, E. Zias, et al. 2004. Cardiac stem cell and myocyte aging, heart failure, and insulin-like growth factor-1 overexpression. Circul. Res. 94: 514–524. Torres, C., M. K. Francis, A. Lorenzini, M. Tresini and V. J. Cristofalo. 2003. Metabolic stabilization of MAP kinase phosphastase-2 in senesence of human fibroblasts. Exp. Cell Res. 290: 195–206. Torrey, B. B., K. Kinsella and C. M. Taeuber. 1987. An Aging World. International Population Reports Series P-95, no. 78. U.S. Government Printing Office, Washington, DC. Toussaint, O., C. Michiels, M. Raes and J. Remacle. 1995. Cellular aging and the importance of energetic factors. Exp. Gerontol. 30: 1–22. Tower et al., 2004 . Proc. Natl. Acad. Sci. USA article. Tower, J. 1996. Aging mechanisms in fruit flies. BioEssays 18: 799–807. Tower, J., J. Wheeler, R. Kurapati and E. Bieske. 1993. Novel promoter elements direct aging-specific transcriptional regulation of heat shock and other genes. P. 310 in Proceedings of the 34th Drosophila Research Conference (abstract). Tracy, R. P. 2003. Emerging relationships of inflammation, cardiovascular disease and chronic diseases of aging. Int. J. Obesity 27: 529–534. Tricerei, A., A. R. Errani, M. Vangeli, L. Guidi, I. Pavse, L. Antico and C. Bartoloni. 1995. Neuorimmunomodulation and psychoneuroendocrinology: Recent findings in adults and aged. Panmin. Med. 37: 77–83. Trifunovic, A., A. Wredenberg, M. Falkenberg, J. N. Spelbrink, A. T. Rovio, C. E. Bruder, M. BohlooyY, et al. 2004. Premature ageing in mice expressing defective mitochondrial DNA polymerase. Nature 429: 417–423. Trout, D. L. 1991. Vitamin C and cardiovascular risk factors. Am. J. Clin. Nutr. 53(suppl. 1): 322S– 325S. Tsuchiya, T., J. M. Dhahbi, X. Cui, P. L. Mote, A. Bartke and S. R. Spindler. 2004. Additive regulation of hepatic gene expression by dwarfism and caloric restriction. Physiol. Genomics 17: 307–315. Tu, M-P., C-M. Yeng and M. Tatar. 2002. Impaired ovarian ecdysone synthesis of Drosophila melanogaster insulin receptor mutants. Aging Cell 1: 158–160. Tucker, M. J. 1979. The effect of long-term food restriction on tumours in rodents. Int. J. Cancer. 23: 803–807.

584

References

Turgeon, J. L., D. P. McDonnell, K. A. Martin and P. M. Wise. 2004. Hormone therapy: physiological complexity belies therapeutic simplicity. Science 304: 1269–1273. Tyler, R. H., H. Brar, M. Singh, A. Latorre, J. L. Graves, L. D. Mueller, M. R. Rose and F. J. Ayala. 1993. The effect of superoxide dismutase alleles on aging in Drosophila. Genetica 91: 143–149. Tzarkoff, S. P. and A. H. Norris. 1978. Longitudinal changes in basal metabolism in man. J. Appl. Physiol. 45: 536–539. Udelsman, R., D. G. Li, C. A. Stagg and N. K. Holbrook. 1995. Aortic crosstransplantation between young and old rats: Effect upon the heat shock protein 70 stress response. J. Gerontol. Biol. Sci. 50A: B187–B192. Udelsman, R., M. J. Blake, C. A. Stagg, D. G. Li, D. J. Putney and N. Holbrook. 1993. Vascular heat shock protein expression in response to stress: Endocrine and autonomic regulation of the agedependent response. J. Clin. Invest. 91: 465–473. Ueno, L. M., Y. Yamashita, T. Moritani and E. Nakamura. 2003. Biomarkers of aging in women and the rate of longitudinal changes. J. Physiol. Anthropol. Appl. Hum. Sci. 22: 37–46. U.S. Bureau of the Census. 1965. Estimates of the Populations of the United States, by Single Years of Age, Color, and Sex, 1900–1959. Current Population Reports, Series P-25, no. 311. U.S. Government Printing Office, Washington, DC. U.S. Senate. 1987–1998. Committee on Aging. Aging America: Trends and Projections. Uthus, E. O. and H. M. Brown-Borg. 2003. Altered methionine metabolism in long living Ames dwarf mice. Exp. Gerontol. 38: 491–498. Utsuyama, M., M. Kasai, C. Kurashima and K. Hirokawa. 1991. Age influence on the thymic capacity to promote differentiation of T cells: Induction of different composition of T cell subsets by aging thymus. Mech. Ageing Devel. 58: 267– 277. Uvnas-Moberg, K. 1989. The gastrointestinal tract in growth and reproduction. Sci. Am. 261(1): 78–83. van Bockxmeer, F. M. 1994. ApoE and ACE genes: impact on human longevity. Nature Genet. 6: 4–5. van Dam, B., V. W. van Hinsbergh, C. D. Stehouwer, A. Versteilen, H. Dekker, R. Buytenhek, H. M. Princen and C. G. Schalkwijk. 2003. Vitamin E inhibits lipid peroxidation-induced adhesion molecule expression in endothelial cells and decreases soluble cell adhesion molecules in healthy subjects. Cardiovasc. Res. 57: 563–571. Van Den Biggelaar, A. H., A. J. De Craen, J. Gussekloo, T. W. Huizinga, B. T. Heijmans, M. Frolich, T. B. Kirkwood and R. G. Westendorp. 2004. Inflammation underlying cardiac mortality is a late con-

sequence of evolutionary programming. FASEB J. 18: 1022–1024. van Eekelen, J. A., N. Y. Rots, W. Sutanto, M. S. Oitzl and E. R. de Kloet. 1991. Brain corticosteroid receptor gene expression and neuroendocrine dynamics during aging. J. Steroid Biochem. Mol. Biol. 40: 679–683. Vanfleteren, J. R. and B. P. Braeckman. 1999. Mechanisms of life span determination in Caenorhabditis elegans. Neurobiol. Aging 20: 487–502. Vanfleteren, J. R. 1993. Oxidative stress and ageing in Ceanorhabditis elegans. Biochem. J. 292: 605– 608. Van Remmen, H. and A. Richardson. 2001. Oxidative damage to mitochondria and aging. Exp. Gerontol. 36: 957–968. Van Remmen, H., Y. Ikeno, M. Hamilton, M. Pahlavani, N. Wolf, S. R. Thorpe, N. L. Alderson, J. W. Baynes, C. J. Epstein, T. T. Huang, J. Nelson, R. Strong and A. Richardson. 2003. Life-long reduction in MnSOD activity results in increased DNA damage and higher incidence of cancer but does not accelerate aging. Physiol. Genomics 16: 29– 37. Van Remmen, H., W. F. Ward, R. V. Sabin and A. Richardson. 1995. Gene expression and protein degradation. Pp. 171–238 in E. J. Masoro (ed.), Handbook of Physiology, Section 11: Aging. Oxford University Press, New York. Van Voorhies, W. A., A. A. Khazaeli and J. W. Curtsinger. 2004. Testing the “rate of living” model: Further evidence that longevity and metabolic rate are not inversely correlated in Drosophila melanogaster. J Appl. Physiol. In press. Vasan, S., P. Roiles and H. Founds. 2003. Therapeutic potential of advanced glycation end productprotein crosslinks. Arch. Biochem. Biophys. 419: 89–96. Vaughn, D. W. 1977. Age-related deteriorations of pyramidal cell basal dendrites in rat auditory cortex. J. Comp. Neurol. 171: 501–516. Vaupel, J. W., J. R. Carey, K. Christensen, T. E. Johnson, A. I. Yashin, N. V. Holm, I. A. Lachine, V. Kannisto, A. A. Khazaeli, P. Liedo, V. D. Longo, Y. Zeng, K. G. Manton and J. W. Curtsinger. 1998. Biodemographic trajectories of longevity. Science 280: 855–860. Vaux, D. L. and A. Strasser. 1996. The molecular biology of apoptosis. Proc. Natl. Acad. Sci. USA 93: 2239–2244. VERIS. 1991. Safety of oral vitamin E. Vitamin E Research and Information Service, LaGrange, IL. VERIS. 1996. The role of antioxidants in prevention of coronary heart disease. Vitamin E Research and Information Service, LaGrange, IL. VERIS. 1998a. An overview of Vitamin E efficacy. Vitamin E Research & Information Service.

References

VERIS. 1998b. The protective role of antioxidants in the aging process. Online at http://www.cognis .com/veris/verisdefault.htm Vesselinovitch, D. 1986. Age-related changes in selected animal species. Pp. 154–175 in S. R. Bates and E. C. Gangloff (eds.), Atherogenesis and Aging. Springer-Verlag, New York. Vestal, R. E. 1978. Drug use in the elderly: A review of problems and special considerations. Drugs 16: 358–382. Vestal, R. E. and G. W. Dawson. 1985. Pharmacology and aging. Pp. 744–819 in C. E. Finch and E. L. Schneider (eds.), Handbook of the Biology of Aging. Van Nostrand Reinhold, New York. Via, S., R. Gomulkiewicz, G. De Jong, S. M. Scheiner, C. D. Schlichting and P. H. Van Tienderen. 1995. Adaptive phenotypic plasticity: Consensus and controversy. Trends Ecol. Evol. 10: 212–217. Viidik, A., H. M. Nielsen and M. Skalicky. 1996. Influence of physical exercise on aging rats. II. Lifelong exercise delays aging of tail tendon collagen. Mech. Ageing Dev. 88: 139–148. Vijg, J. and D. L. Knook. 1987. DNA repair in relation to the aging process. J. Am. Geriatr. Soc. 35: 532–541. Vijg, J., A. G. Vitterlinden, E. Mullaart, P. A. M. Lohman and D. L. Knook. 1985. Processing of DNA damage during aging: Induction of genetic alterations. Pp. 155–172 in R. S. Sohal, L. S. Birnbaum and R. G. Cutler (eds.), Molecular Biology of Aging: Gene Stability and Gene Expression. Raven Press, New York. Vito, P., E. Lacana and L. D’Adamio. 1996. Interfering with apoptosis: Ca2+-binding protein ALG-2 and Alzheimer’s disease gene ALG-3. Science 271: 521–525. Vlassara, H., M. Brownlee, K. R. Manogue, C. A. Dinarello and A. Pasagian. 1988. Cachectin/TNF and IL-1 induced by glucose-modified proteins: Role in normal tissue remodeling. Science 240: 1546–1548. Vogel, H., D-S. Lim, G. Karsenty, M. Finegold and P. Hasty. 1999. Deletion of Ku86 causes early onset of senescence in mice. Proc. Natl. Acad. Sci. USA 96: 10770–10775. von Zglinicki, T. 2002. Oxidative stress shortens telomeres. Trends Biochem. Sci. 27: 339–344. von Zglinicki, T. 2003. Replicative senescence and the art of counting. Exp. Gerontol. 38: 1259–1264. von Zglincki, T. 2003. Telomeric Damage in Aging. Pp. 121–129 in Aging at the Molecular Level (T. von Zglincki, ed.). Kluwer Academic Publishers, the Netherlands. von Zglincki, T., G. Saretzki, W. Docke and C. Lotze. 1995. Mild hyperoxia shortens telomeres and inhibits proliferation of fibroblasts: A model for senescence? Exp. Cell Res. 220: 186–193.

585

Wachter, K. W. and C. E. Finch (eds.). 1997. Between Zeus and the Salmon: The Biodemography of Longevity. National Academy Press, Washington, D.C. Waldron, I. 1987. Causes of the sex difference in longevity. J. Am. Geriatr. Soc. 35: 365–366 (letter to the editor). Walford, R. H. 1983. Supergenes: Histocompatibility; Immunologic and other parameters in aging. Pp. 53– 68 in W. Regelson and F. M. Sinex (eds.), Intervention in the Aging Process, Part B: Basic Research and Preclinical Screening. Alan R. Liss, New York. Walford, R. L. 1986. The 120 Year Diet: How to Double Your Vital Years. Pocket Books, New York. Walford, R. L., S. B. Harris and M. W. Gunion. 1992. The calorically rstricted low-fat nutriet-dense diet in Biosphere 2 significantly lowers blood glucose, total leukocyte count, cholesterol and blood pressure in humans. Proc. Natl. Acad. Sci. USA 89: 11533–11537. Walford, R. L., D. Mock, R. Verdery and T. MacCallum. 2002. Caloric restriction in Biosphere2: Alterations in physiologic, hematologic, hormonal, and biochemical parameters in humans restricted for a 2–year period. J. Gerontol. Biol. Sci. 57A: B211–B2224. Walker, G. A. and G. J. Lithgow. 2003. Lifespan extension in C. elegans by a molecular chaperone dependent upon insulin-like signals. Aging Cell 2: 131–139. Wallace, D. C. 1992. Mitochondrial genetics: A paradigm for aging and degenerative disease? Science 256: 628–632. Wallace, D. C. 1995. Mitochondrial DNA mutations in human disease and aging. Pp. 163–178 in K. Esser and G. M. Martin (eds.), Molecular Aspects of Aging. Wiley, Chichester, England. Wallace, D. C. 1999. Mitochondrial diseases in man and mouse. Science 283: 1482–1488. Wallace, D. C., V. A. Bohr, G. Cortopassi, B. Kadenbach, S. Linn, A. W. Linnane, C. Richter and J. W. Shay. 1995. Group report: The role of bioenergetics and mitochondrial DNA mutations in aging and age-related diseases. Pp. 199–226 in K. Esser and G. M. Martin (eds.), Molecular Aspects of Aging. Wiley, Chichester, England. Wallis, C. V., A. N. Sheerin, M. H. L. Green, C. J. Jones, D. Kipling and R. G. A. Faragher. 2004. Fibroblast clones from patients with HutchinsonGilford progeria can senesce despite the presence of telomerase. Exp. Gerontol. 39: 461–467. Walsh, K. and H. Perlman. 1997. Cell cycle exit upon myogenic differentiation. Curr. Opin. Genet. Dev. 7: 597–602. Wang, H. D., P. Kazemi-Esfarjani and S. Benzer. 2004. Multiple stress analysis for isolation of Drosophila longevity genes. Proc. Natl. Acad. Sci. USA 101: 12610–12615.

586

References

Wang, M. C., D. Bohmann and H. Jasper. 2003. JNK signaling confers tolerance to oxidative stress and extends lifespan in Drosophila. Develop. Cell. 5: 811–816. Wang, S. M., C. Nishigori, T. Yagi and H. Takebe. 1991. Reduced DNA repair in progeria cells and effects of gamma-ray irradiation on UV-induced unscheduled DNA synthesis in normal and progeria cells. Mutat. Res. 256: 59–66. Ward, W. F. 1988a. Enhancement by food restriction of liver protein synthesis in the aging Fischer 344 rat. J. Gerontol. Biol. Sci. 43: B50–B53. Ward, W. F. 1988b. Food restriction enhancement of the proteolytic capacity of aging rat liver. J. Gerontol. Biol. Sci. 43: B121–B124. Warner, H. R., G. Fernandes and E. Wang. 1995. A unifying hypothesis to explain the retardation of aging and tumorigenesis by caloric restriction. J. Gerontol. Biol. Sci. 50A: B107–B109. Warner, H. R. and F. Sierra. 2003. Models fo accelerated ageing can be informative about the molecular mechanisms of ageing and/or age-related pathology. Mech. Ageing Dev. 124: 581–587. Warner, H. R. and E. Wang. 1989. Control of cell proliferation in senescent cells—a synopsis. J. Gerontol. Biol. Sci. 44: B23–B25 (meeting report). Watkinson, A. R. and J. White. 1985. Some life-history consequences of modular construction in plants. Phil. Trans. R. Soc. Lond. B 313: 31–51. Weaver, J. K. and J. Chalmers. 1966. Cancellous bone: Its strength and changes with aging and an evaluation of some methods for measuring its mineral content. I. Age changes in cancellous bone. J. Bone Joint Surgery 48A: 289–308. Webster, C., L. Silberstein, A. P. Hays and H. M. Blair. 1988. Fast muscle fibers are preferentially affected in Duchenne muscular dystrophy. Cell 52: 503–513. Webster, G. C. 1988. Protein synthesis. Pp. 119–130 in F. A. Lints and M. H. Soliman (eds.), Drosophila as a Model Organism for Ageing Studies. Blackie, Glasgow. Webster, G. C. and S. L. Webster. 1979. Decreased protein synthesis by microsomes from aging Drosophila melanogaster. Exp. Gerontol. 14: 343–348. Webster, G. C. and S. L. Webster. 1983. Decline in synthesis of elongation factor one (EF-1) precedes the decreased synthesis of total protein in aging Drosophila melanogaster. Mech. Ageing Dev. 22: 121–128. Webster, G. C. and S. L. Webster. 1984. Specific disappearance of translatable messenger RNA for elongation factor one in aging Drosophila melanogaster. Mech. Ageing Dev. 24: 335–342. Wei, Y. H., S. H. Kao and H. C. Lee. 1996. Simultaneous increase of mitochondrial DNA deletions and lipid peroxidation in human aging. Ann. N.Y. Acad. Sci. 786: 24–43.

Weidemann, A., G. Konig, D. Bunke, P. Fischer, J. M. Salbaum, C. L. Masters and K. Bayreuther. 1989. Identification, biogenesis and localization of precursors of Alzheimer’s disease A4 amyloid protein. Cell 57: 115–126. Weindruch, R. 1995a. Animal models. Pp. 37–52 in E. J. Masoro (ed.), Handbook of Physiology, Section 11: Aging. Oxford University Press, New York. Weindruch, R. 1995b. Diet restriction. Pp. 276–279 in G. Maddox (ed.), The Encyclopedia of Aging, 2nd ed. Springer, New York. Weindruch, R. 1995c. Interventions based on the possibility that oxidative stress contributes to sarcopenia. J. Gerontol. Biol. Sci. 50A(special issue): 157–161. Weindruch, R., T. Kayo, C-K. Lee and T. A. Prolla. 2002. Gene expression profiling of aging using DNA microarrays. Mech. Ageing Dev. 123: 177–193. Weindruch, R. and T. Prolla. 2003. Gene expression profile of the aging brain. Arch. Neurol. 59: 1712– 1714. Weindruch, R. and R. L. Walford. 1982. Dietary restriction in mice beginning at 1 year of age: Effect on life-span and spontaneous cancer incidence. Science 215: 1415–1418. Weindruch, R. and R. L. Walford. 1988. The Retardation of Aging and Disease by Dietary Restriction. Charles Thomas, Springfield, IL. Weindruch, R., R. L. Walford, S. Fligiel and D. Guthrie. 1986. The retardation of aging in mice by dietary restriction: Longevity, cancer, immunity and lifetime energy intake. J. Nutr. 116: 641– 654. Weinkove, D. and S. J. Leevers. 2000. The genetic control of organ growth: insights from Drosophila. Curr. Opin. Genet. Dev. 10: 75–80. Weirach-Schwaiger, H., H. G. Weirich, B. Gruber, M. Schweiger and M. Hirsch-Kauffmann. 1994. Correlation between senescence and DNA repair in cells from young and old individuals and in premature aging syndromes. Mutat. Res. 316: 37–48. Weismann, A. 1891a. The continuity of the germ plasm as the foundation of a theory of heredity (1885). Pp. 163–256 in E. B. Poulton, S. Schonland and A. E. Shipley (eds.), Essays upon Heredity and Kindred Biological Problems, 2nd ed., vol. 1. Clarendon Press, Oxford. [Reprint: Baker Science, Oceanside, NY.] Weismann, A. 1891b. The duration of life (a paper presented in 1881). Pp. 1–66 in E. B. Poulton, S. Schonland and A. E. Shipley (eds.), Essays upon Heredity and Kindred Biological Problems, 2nd ed., vol. 1. Clarendon Press, Oxford. [Reprint: Baker Science, Oceanside, NY.] Weismann, A. 1891c. Life and death (a paper presented in 1883). Pp. 111–157 in E. B. Poulton,

References

S. Schonland and A. E. Shipley (eds.), Essays upon Heredity and Kindred Biological Problems. Clarendon Press, Oxford. [Reprint: Baker Science, Oceanside, NY.] Weitz, J.S. and H.B. Fraser, 2001. Explaining mortality rate plateaus. Proc. Natl. Acad. Sci. USA 98: 15383–15386. Weitzmann, M.N., C. Roggia, G. Toraldo, L. Weitzmann and R. Pacifici. 2002. Increased production IL-7 uncouples bone formation from bone resorption during estrogen deficiency. J. Clin. Invest. 110: 1643–1650. Welle, S., A. I. Brooks, J. M. Delehanty, N. Needler, K. Bhatt, B. Shah and C. A. Thornton. 2004. Skeletal muscle gene expression profiles in 20–29 year old and 65–71 year old women. Exp. Gerontol. 39: 369–378. Welle, S., A. I. Brooks, J. M. Delehanty, N. Needler and C. A. Thornton. 2003. Gene expression profile of aging in human muscle. Physiol. Genomics 14: 149–159. Werner-Washburne, M., E. L. Braun, M. E. Crawford and V. M. Peck. 1996. Stationary phase in Saccharomyces cerevisiae. Mol. Microbiol. 19: 1159–1166. Westendorp, R. G. J. and T. B. L. Kirkwood, 1998. Human longevity at the cost of reproductive success. Nature 396: 743–746. Westendorp, R. G. J. and T. B. L. Kirkwood, 1999. Human longevity and reproductive success: Response to Gavrilov and Gavrilova. J. Anti-Aging Medicine 2: 125–126. Westendorp, R. G. J. and T. B. L. Kirkwood, 2001. Maternal and paternal lines of familiar longevity. Population—An English Selection 13: 223. Westerterp, K. R. and E. P. Meijer, 2001. Physical activity and parameters of aging: A physiological perspective. J. Gerontol. 56A (special Issue II): 7–12. Westing, A. H. 1964. The longevity and aging of trees. Gerontologist 4: 10–15. Whalley, L. J. and I. J. Deary. 2001. Longitudinal cohort study of childhood IQ and survival up to age 76. Br. Med. J. 322: 819–823. Wharton W. 1984. Newborn human skin fibroblasts senesce in vitro without acquiring adult growth factor requirements. Exp. Cell Res. 154(1): 310– 314. Wheeler, K. T. and J. T. Lett. 1974. On the possibility that DNA repair is related to age in non-dividing cells. Proc. Natl. Acad. Sci. USA 71: 1862–1865. Whitbourne, S. K. 1985. The Aging Body: Physiological Changes and Psychological Consequences. Springer-Verlag, New York. White, H. J. 2003. The social security and medicare debate three years after the 2000 election. Public Policy and Aging Report 13(4): 15–19. White, K., M. E. Grether, J. M. Abrams, L. Young, K.

587

Farrell and H. Steller. 1994. Genetic control of programmed cell death in Drosophila. Science 264: 677–683. Whitehead, I. and T. A. Grigliatti. 1993. A correlation between DNA repair capacity and longevity in adult Drosophila melanogaster. J. Gerontol. Biol. Sci. 48: B124–B132. Whitfield, C. W., A-M. Cziko and G. E. Robinson. 2003. Gene expression profiles in the brain predict behavior in individual honey bees. Science 302: 296–299. Whitfield, K. E. 1994. The use of quantitative genetic methodology to gain insights into the origins of individual differences in later life (letter to the editor). Exp. Aging Res. 20: 135–143. Whitman, C. O. 1894. Evolution and epigenesis. Biol. Lect. (Woods Hole) 3: 205–224. Wickelgren, I. 1996. Is hippocampal cell death a myth? Science 271: 1229–1230. Wiens, E. and M. S. Grotewiel, 2002. Dissociation between functional senescence and oxidative stress resistance in Drosophila. Exp. Gerontol. 37: 1347– 1357. Wilkins, A. S. 1986. Genetic Analysis of Animal Development. Wiley-Interscience, New York. Willett, W. C. 1994. Diet and health: What should we eat? Science 264: 532–537. Willett, W. C. 2002. Balancing life-style and genomics research for disease prevention. Science 296: 695– 698. Williams, G. C. 1957. Pleiotropy, natural selection and the evolution of senescence. Evolution 11: 398– 411. Williams, J. R. 1983. Alteration in DNA/chromatin structure during aging. Pp. 145–153 in W. Regelson and F. M. Sinex (eds.), Intervention in the Aging Process. Part B. Basic Research and Preclinical Screening. Alan R. Liss, New York. Williams, M. D., H. Van Remmen, C. C. Conrad, T. T. Huang, C. J. Epstein and A. Richardson. 1998. Increased oxidative damage is correlated to altered mitochondrial function in heterozygous manganese superoxide dismutase knockout mice. J. Biol. Chem. 273: 28510–28515. Wilmoth, J. R., L. J. Deegan, H. Lundstrom and S. Horiuchi. 2000. Increase of maximum life-span in Sweden, 1861–1899. Science 289: 2366–2368. Wilmoth, J. R. 2000. Demography of longevity: past, present, and future trends. Exp. Gerontol. 35: 1111–1129. Wilson, A. C. 1991. From molecular evolution to body and brain evolution. Pp. 331–340 in Perspectives on Cellular Regulation: From Bacteria to Cancer. MBL Lectures in Biology, vol. 11. Wiley, New York. Wilson, D. 1994. The analysis of survival (mortality) data: Fitting Gompertz, Weibull and logistic functions. Mech. Ageing Dev. 74: 15–33.

588

References

Wilson, D. L. 1988. Aging hypotheses, aging markers and the concept of biological age. Exp. Gerontol. 23: 435–438. Wilson, F. H., A. Hariri, A. Farhi, H. Zhao, K. F. Petersen, H. R. Toka, C. Nelson-Williams, K. M. Raja, M. Kashgarian, G. I. Shulman, S. J. Scheinman and R. P. Lifton. 2004. A cluster of metabolic defects caused by mutation in a mitochondrial tRNA. Science 306: 1190–1194. Winston, M. L. 1987. The Biology of the Honey Bee. Harvard University Press, Cambridge, MA. Wise, P. M. 1986. Changes in the central nervous system and neuroendocrine control of reproduction in males and females. Pp. 81–96 in L. Mastroianni Jr. and C. A. Paulsen (eds.), Aging, Reproduction and the Climacteric. Plenum Press, New York. Wissler, R. W. and D. Vesselinovitch. 1986. The pathogenesis of atherosclerosis: Myths and established facts about its relationship to aging. Pp. 7– 19 in S. R. Bates and E. C. Gangloff (eds.), Atherogenesis and Aging. Springer-Verlag, New York. Witten, M. 1984. A return to time, cells, systems and aging. II. Relational and reliability theoretic aspects of senescence in mammalian systems. Mech. Ageing Dev. 27: 323–340. Witten, M. 1987. Information content of biological survival curves arising in aging experiments: Some further thoughts. Pp. 295–318 in A. Woodhead and K. H. Thompson (eds.), Evolution of Aging Processes in Animals. Plenum Press, New York. Witten, M. 1989. Re-examining the Gompertzian model of aging. Institute for Mathematics and Its Applications, University of Minnesota Reprint Series no. 483, St. Paul. Witten, M. 1992. The Frankenstein project: Building a man in the machine and the arrival of the computional physician. Int. J. Supercomputer Applic. 6: 245–319. Witten, M. 1994. Can stochasticity explain variation in clonal population survival curves. Mech. Ageing Dev. 73: 33–64. Wolf, A. M. and G. A. Colditz. 1996. Social and economic effects of body weight in the United States. Am. J. Clin. Nutr. 63(suppl. 3): 466S–469S. Wolf, D. S., M. Gearing, D. A. Snowdon, H. Mori, W. R. Markesbery and S. S. Mirra. 1999. Progression of regional neuropathology in Alzheimer disease and normal elderly: findings from the Nun study. Alzheimer Dis. Assoc. Disord. 13: 226– 231. Wolfe, J. 1998. Growth hormone: A physiological fountain of youth? J. Anti-Aging Med. 1: 9–26. Wolkow, C. A., K. D. Kimura, M. S. Lee and G. Ruvkun. 2000. Regulation of C. elegans life-span by insulin-like signaling in the nervous system. Science 290: 147–150.

Wong, A., P. Boutis and S. Hekimi. 1995. Mutations in the clk-1 gene of Caenorhabditis elegans affect developmental and behavioral timing. Genetics 139: 1247–1259. Wood, J. G., B. Rogina, S. Lavu, K. Howitz, S. G. Helfand, M. Tatar and D. Sinclair. 2004. Sirtuin activators mimic caloric restriction and delay ageing in metazoans. www.nature.com/nature/doc .10.1038/nature02789. Wood, S. M. and R. R. Watson. 1994. Antioxidants and cancer in the aged. Pp. 281–294 in R. Watson (ed.), Handbook of Nutrition in the Aged, 2nd ed. CRC Press, Boca Raton, FL. Woodruff, R. C. and A. G. Nikitin. 1995. P DNA element movement in somatic cells reduces lifespan in Drosophila melanogaster: Evidence in support of the somatic mutation theory of aging. Mutat. Res. 338: 35–42. Woodruff, R. C. and J. N. Thompson Jr. 2003. The role of somatic and germline mutations in aging and a mutation interaction model of aging. J. Anti Aging Med. 6: 29–39. Wright, R. M. and D. J. Cummings. 1983. Integration of mitochondrial gene sequences within the nuclear genome during senescence in a fungus. Nature 302: 86–88. Wright, W. E., D. Brasiskyte, M. A. Piatyszek and J. W. Shay. 1996. Experimental elongation of telomeres extends the lifespan of immortal ¥ normal cell hybrids. EMBO J. 15: 1734–1741. Wu, L., M. H. N. Ashraf, M. Facci, R. Wang, P. G. Paterson, A. Ferrie and B. H. J. Juurlink. 2004. Dietary approach to attenuate oxidative stress, hypertension, and inflammation in the cardiovascular system. Proc. Natl. Acad. Sci. USA 101: 7094–7099. Wurtman, R. J. 1985. Alzheimer’s disease. Sci. Am. 252(1): 62–74. Yakes, F. M. and B. Van Houten. 1997. Mitochondrial DNA damage is more extensive and persists longer than nuclear DNA damage in human cells followng oxidative stress. Proc. Natl. Acad. Sci. USA 94: 514–519. Yamaoka, M., K. Isobe, H. Shitara, H. Yonekawa, S. Miyabayashi and J. I. Hayashi. 2000. Complete repopulation of mouse mitochondrial DNA-less cells with rat mitochondrial DNA restores mitochondrial translation but not mitochondrial respiratory function. Genetics 155: 301–307. Yan, S. D., X. Chen, J. Fu, M. Chen, H. Zhu, A. Roher, T. Slattery, L. Shao, M. Nagashima, J. Morser, A. Migheli, P. Nawroth, D. Stern and A. M. Schmidt. 1996. RAGE and amyloid-b peptide neurotoxcicity in Alzheimer’s disease. Nature 382: 685–691. Yanase, S., K. Yasuda and N. Ishii. 2002. Adaptive responses to oxidative damage in three mutants of

References

Caenorhabditis elegans (age-1, mev-1 and daf-16) that affect life span. Mech. Ageing Develop. 123: 1579–1587. Yang, S. H., R. Liu, E. J. Perez, Y. Wen, S. M. Stevens Jr., T. Valencia, A. M. Brun-Zinkernagel, L. Prokai, Y. Will, J. Dykens, P. Koulen and J. W. Simpkins. 2004. Mitochondrial localization of estrogen receptor beta. Proc. Natl. Acad. Sci. USA 101: 4130–4135. Yashin, A. I., A. S. Begun, S. I. Boiko, S. V. Ukraintseva and J. Oeppen, 2002. New age patterns of survival improvement in Sweden: Do they characterize changes in individual aging? Mech. Ageing Dev. 123: 637–647. Yashin, A. I. and I. A. Iachine. 1995a. Genetic analysis of durations: Correlated frailty model applied to survival of Danish twins. Genet. Epidemiol. 12: 529–538. Yashin, A. I. and I. A. Iachine. 1995b. How long can humans live? Lower bound for biological limit of human longevity calculated from Danish twin data using correlated frailty model. Mech. Ageing Dev. 80: 147–169. Yasuda, K., H. Adachi, Y. Fujiwara and N. Ishii. 1999. Protein carbonyl accumulation in aging dauer formation-defective (daf) mutants of Caenorhabiditis elegans. J.Gerontol. Biol. Sci. 54A: B47– B51. Yeger-Lotem, E., S. Sattath, N. Kashtan, S. Itzkovitz, R. Milo, R. Y. Pinter, U. Alon and H. Margalit. 2004. Network motifs in integrated cellular networks of transcription-regulation and proteinprotein interaction. Proc. Natl. Acad. Sci. USA 101: 5934–5939. Yoon, S-O., C-H. Yun and A-S. Chung. 2002. Dose effect of oxidative stress on signal transduction in aging. Mech. Ageing Dev. 123: 1597–1604. Yoshioka, M., H. Tanaka, N. Shono, E. E. Snyder, M. Shindo and J. St-Amand. 2003. Serial analysis of gene expression in the skeletal muscle of endurance athletes compared to sedentary men. FASEB J. 17: 1812–1819. Yu, B. P. 1995. Putative interventions. Pp. 613–633 in E. J. Masoro (ed.), Handbook of Physiology. Section 11: Aging. Oxford University Press, New York. Yu, B. P., E. J. Masoro and C. A. McMahan. 1985. Nutritional influences on aging of Fischer 344 rats. I. Physical, metabolic and longevity characteristics. J. Gerontol. 40: 657–670. Yu, C. E., J. Oshima, Y. H. Fu, E. M. Wijsman, F. Hisama, R. Alisch, S. Matthews, J. Nakura, T. Miki, S. Ouais, G. M. Martin and G. D. Schellenberg. 1996a. Positional cloning of the Werner’s syndrome gene. Science 272: 258–262. Yu, C. E., J. Oshima, F. M. Hisama, S. Matthews, B. J. Trask and G. D. Schellenberg. 1996b. A YAC,

589

P1, and cosmid contig and 17 new polymorphic markers for the Werner syndrome region at 8p12– p21. Genomics 35: 431–440. Yu, C. E., J. Oshima, E. M. Wijsman, J. Nakura, T. Miki, C. Piussan, S. Matthews, Y. H. Fu, J. Mulligan, G. M. Martin, J. Mulligan and G. D. Schellenberg. 1997. Mutations in the consensus helicase domains of the Werner syndrome gene. Am. J. Hum. Genet. 60: 330–341. Yuan, H., T. Kaneko and M. Matsuo. 1996. Increased susceptibility of late passage human diploid fibroblasts to oxidative stress. Exp. Gerontol. 31: 465– 474. Yuan, I. C. 1932. The influence of heredity upon the duration of life in man based on a Chinese genealogy from 1365 to 1914. Hum. Biol. 4: 41–68. Yuh, K. C. M. and A. Gafni. 1987. Reversal of agerelated effects in rat muscle phosphoglycerate kinase. Proc. Natl. Acad. Sci. USA 84: 7458–7462. Yunis, E. J. and M. Salazar. 1993. Genetics of life span in mice. Genetica 91: 211–223. Zaina, S., L. Pettersson, A. N. Thomsen, C-M. Chai, Z. Qi, J. Thyberg and J. Nilsson. 2003. Shortened life span, bradycardia, and hypotension in mice with targeted expression of an Igf2 transgene in smooth muscle cells. Endocrinology 144: 2695– 2703. Zakeri, Z. and R. A. Lockshin. 1994. Physiological cell death during development and its relationship to aging. Ann. N.Y. Acad. Sci. 719: 212–229. Zakian, V. A. 1995. Telomeres: Beginning to understand the end. Science 270: 1601–1607. Zambon, A. C., E. L. McDearmon, N. Salomonis, K. M. Vranizan, K. L. Johansen, D. Adey, J. S. Takahashi, M. Schambelan and B. R. Conklin. 2003. Time- and exercise-dependent gene regulation in human skeletal muscle. Genome Biol. 4:R61. http://genomebiology.com/2003/4/10/ R61 Zaug, A. J. and T. R. Cech. 1986. The Tetrahymena intervening sequence ribonucleic acid enzyme is a phosphotransferase and an acid phosphatase. Biochemistry 25: 4478–4482. Zentgraf, U., K. Hinderhorer and D. Lolb. 2000. Specific association of a small protein with the telomeric DNA-protein complex during the onset of leaf senescence in Arabidopsis thaliana. Plant Mol. Biol. 42: 429–438. Zglincki, T. 2002. Oxidative stress shortens telomeres. Trends Biol. Sci. 27: 339–334. Zhang, C., A. Baumer, R. J. Maxwell, A. W. Linnane and P. Nagley. 1992. Multiple mitochondrial DNA deletions in an elderly human individual. FEBS Lett. 297: 34–38. Zhang, J., J. Asin-Cayuela, J. Fish, Y. Michikawa, M. Onafe, F. Olivieri, G. Passarino, G. De Benedictis, G. Franscheschi and G. Attardi. 2003.

590

References

Strikingly higher frequency in centenarians and twins of mrDNA mutation causing remodeling of replication origin in leukocytes. Proc. Natl. Acad. Sci. USA 100: 1116–1121. Zhu, H., X. Wang and J-Y. Zhu. 2003. Effect of aging on network structure. Phys. Rev. E 68, 056121-1.

Zimmerman, J. A., V. Malloy, R. Krajcik and N. Orentreich. 2003. Nutritional control of aging. Exp. Gerontol. 38: 47–52. Zwaan, B., R. Bijlsma and R. Hoekstra. 1995. Direct selection on lifespan in Drosophila melanogaster. Evolution 49: 649–659.

Index

accelerated aging, mice, 376 actuarial analysis creatinine clearance, 56–58 cross-sectional studies, 54–55 empirical longitudinal and cross-sectional comparisons, 56–62 growth curves of adolescents, 60, 61 longitudinal studies, 55–56 maximum oxygen uptake, 58–60, 61 weight, 56–57 adolescents, growth curves, 60, 61 age chronological changes, 56 relationship to mortality, 35 U.S. population by, 40 aged animals, as survivors, 20 age-dependent changes, architecture, 355–357 age-related changes. See also actuarial analysis; age-specific mortality. See also mortality definition, 41 Drosophila, 270, 275 in Sweden, 509 in United States, 42 age structure changes during 20th century, 507–510 developed and less developed countries, 508 aging. See also cellular aging; human aging; human aging, modulation; stochastic theories accelerating or retarding changes, 69–71 avian, 129–130 bacterial, 107–110 biological traits, 10, 14–15, 505 definitions, 9–13 deteriorative changes with time, 10 determinants, 494 environmental changes modulating, 68–69 evidence, 6–8 fungi and mitochondria in senescence, 118– 119 immunosenescence and, 466–470 inflammatory parameters, 471

insect, 123–126 invertebrate, 119–131 mammalian, 126–129, 296 models, 13 modular nature of organisms, 131 physiological variables, 75 plant, 130 plasticity, 21–23 postmaturational changes, 71–72 protistan, 96–97, 107–116 rates, 17, 72–90 relationship to disease, 62–63, 137–139,198–201 rotifer, 119–121 sources of information, 24–25 study of, 4, 5–9, 11, 51–52, 104 Tetrahymena, 112 theories of, 24, 105–107, 357–358 vertebrate, 126–130 Volvox, 116–118 yeast, 112–116 aging syndromes Down’s syndrome, 333–334 Hutchinson–Guilford syndrome, 334 progeria, 334 Werner’s syndrome, 334–335 algae, somatic and reproductive cells, 116–118 alpha-tocopherol, 384. See also vitamin E altered protein theory, 361, 366–367 Alzheimer’s disease abnormal apoptosis, 447 age-related pathology, 180–184 amyloid angiopathy, 182 genetic mutations, 183 inflammation, 183–184, 471–472 natural history, 181 risk factors, 180–181 American Association for Advancement of Science, website on aging, 24 amoebae. See also protists aging process, 110–111 amphibians, maximum life spans, 127

591

592

Index

anatomical changes cardiovascular system, 155–162 digestive system, 166–167 excretory system, 167–169 immune system, 184–188 metabolic and hormonal, 193–198 muscle tissue, 151–155 nervous system, 169–184 reproductive system, 188–193 respiratory system, 163–166 skeletal system, 144–151 skin and connective tissue, 141–144 angiotensin-converting enzyme gene, 337 animal models. See also models Caenorhabditis elegans, 253–266 Drosophila melanogaster, 267–289, 490 of Huntingdon’s disease, 353 Saccharomyces cerevisiae, 239–253 animal species behavioral biomarkers, 86–87 mortality rate doubling time, 46–48 antagonistic pleiotrophy theory, 98 anti-aging interventions antioxidant treatment, 217 arguments for and against, 516–522 caloric restriction, 202–211 choosing between scenarios, 523–524 continuation of present trends, 510–511 exercise, 215–217 extended health span, 511–515 genetic manipulations, 217–218 hormonal, 232–234 limits of medicine, 519–520 longevity phenotypes, 511–513 metabolic rate, 212–215 social aspects, 520–524 stem cells, 218 therapeutic cloning, 218 transplantation, 218 antiglycation agents, 212 antioxidant gene expression, Drosophila, 279 antioxidants administering to experimental animals, 387–389 capacity, 385–386 decreasing cell damage by aging, 212 laboratory animals, 217 mitochondria and oxidative stress, 404–405 supplementation in humans, 234–235 aorta, elasticity of, 64, 65 apolipoprotein E, genetic factors, 335–336 apoptosis, diseases of abnormal, 447 arteries age-related changes, 64–67 structure and function, 155–156 arteriosclerosis, age-related, 67 arthritis, bone pathology, 147, 150–151 ascorbic acid, 384–385

ataxia telangiectasia, genetic defects, 375 atherosclerosis cardiovascular pathology, 158–162 inflammation, 471–472 autophagy, 448 bacteria aging, 107–110 growth-arrested, 108, 109 Baltimore Longitudinal Study of Aging (BLSA) biomarkers, 75–77 comparison with cross-sectional studies, 56–62 physiological predictors, 340–343 basal metabolic rate, 193, 194 bees, environmental changes, 68–69 behavioral biomarkers, 86–87 benzo[a]pyrene conversion rate, 373 biological age chronological age vs., 83–84 profiles, biomarkers, 80, 81 biological clock, 10 biological gerontology, 347 biological markers. See biomarkers biological models. See animal models; models biological theories demographic transition, 498–499 developmental span, 500–501 diet and intelligence, 496 early network model, 486–488 epicellular interactions, 493–494 evolution of human life span, 496–498 hierarchical gene expression, 500 integrative models, 485–494, 499–501 intergenerational transfers, 496–498 linear model, 485–486 longevity, 494–499 MARS (mitochondria, aberrant proteins, radicals, and scavengers) network model of aging, 487, 488 protein interaction, 490, 493 scale-free network, 489 senescence, 483–494 biological traits, aging, 14–15, 505 biology, aging inquiry, 4 The Biology of Senescence, 4 biomarkers age-specific groups of, 77–78, 80, 81, 83–85 behavioral, 86–87 concept and criticisms, 73, 90–92 cross-sectional approach, 55 definition, 73–74 fraility syndrome, 85–86 gender-specific, 82–83 human, 75–90 ideal characteristics, 73 invertebrate, 89–90 mean correlation with age, 79

Index

mortality, 77, 79, 81–82, 85–90 physiological, 78–79 for primate aging, 87, 88 rodent, 88–89 segmental and nonsegmental, 74–77 bipolar disorder, age-related pathology, 184 birds aging patterns, 129–130 maximum life spans, 127 as models, 18–19 nonmitochondrial adaptations, 412 birth cohort, chronological changes, 56 birth rates, comparison of countries, 508–509 blood, maintaining glucose levels, 14 blood pressure, relation to age, 64–67 blood vessels. See also arteries structure and function, 155–156 body mass index gender-specific, 340, 341 risk of death in humans, 224, 226, 227 bone. See also skeletal system age-related changes, 146–147, 149–151 arthritis, 147, 150–151 formation and resorption, 148 gender and loss, 149–150 gonadal hormones, 147, 149 human aging, 145–151 normal structure and function, 145–146 osteoporosis, 147, 149–150 brain. See also nervous system aging and caloric restriction in mouse, 294–295 benefit of exercise, 216 structure and function, 169–171 bristlecone pine, life span, 68 budding life span method, yeasts, 112–114 budgies, animal model, 19 Caenorhabditis elegans, 253. See also nematodes endocrine regulation, 266 genomic stability, 263–264 heat shock factor, 352 longevity, 254–263 nuclear-mitochondrial interaction, 259–261 patterns of senescence, 266 reproductive effects, 264–266 calcium absorption, 167 caloric restriction, 101, 202–211 accumulation of advanced glycosylation end products, 208 compounds mimicking, 212 effects on age-related pathology, 203–205 effects on longevity, 79, 81–82, 202–203, 205, 207, 221, 226–228, 242, 465–466 evolutionary origins of, 209–211 hormonal changes, 210–211 inflammation and senescence, 470–471 influence on tissue structure, 204

593

learning performance, 208–209 mechanisms underlying effects, 206 metabolic control, 240–245, 267–273 modulation of gene expression, 293–302 mortality of rats, 43–44, 45 physiological responses to, 205–209 reproduction, 212 thymic function, 469 toxicity studies and, 211 in yeast, 251 cancer, abnormal apoptosis, 447 capillaries, structure and function, 155–156 carbohydrates, reactive oxygen species damage, 386 cardiac index, relationship to age, 159 cardiovascular diseases accelerating inflammatory cycle, 472 age-related changes, 67, 158–162 atherosclerosis, 158–162 development of atherosclerotic plaques, 160 lesions in coronary artery, 161 cardiovascular system age-related changes, 157–158, 294–295 interactions between aging and disease, 199 normal structure and function, 155–157 serum cholesterol levels, 162 workload during exercise, 160 carnitine, mitochondrial function, 413–414 cartilage. See also skeletal system age-related changes, 144–145 normal structure and function, 144 celebration, aging, 3–4 cell biology, mutations, 417 cell death. See also apoptosis autophagy, 448 controllable process, 445–448 red blood cell model, 443–445 cells. See also cellular aging in culture, 419–420, 422 cycle, 420–421 life span, 17–18 mechanisms of oxidative damage, 383 mitochondrial function and cellular energy demand, 402 model of transition, 354 senescence and function, 484–485 senescence in nondividing, 437–443 cell signaling pathways, 433–437 cell spiral model, yeast aging, 114–115 cellular aging cell cycle and, 420–421 cell proliferation and, 424, 435 cytogerontology, 420 Hayflick hypothesis, 425–426 immortality, 419–420 replicative phenotypes, 428 quiescent and senescent fibroblasts, 434–435 seven stages of cell differentiation, 427

594

Index

cellular aging (continued) signaling pathways, 433–437 of skin fibroblasts, 428 telomeres and, 428–433 in vitro, 421–428 cellular organelles, autophagy, 448 centenarians heritability of longevity, 331–332 mitochondrial mutation, 333 superlongevity, 330–331 central nervous system. See nervous system cerebral metabolism, changes, 175–176 chemical system at equilibrium, 13–14 cholesterol, gender-specific, 340, 341 chromatin-dependent gene regulation, 115–116 chronic disease, risk factors, 84 chronological age physiological age vs., 10 physiological variables, 74, 75 relationship with biological age, 83–84 Saccharomyces cerevisiae, 239 chronological changes, confusion and resolution, 56 cloning, stem cells, 218 Cockayne syndrome, genetic defects, 375 cohort life table, Drosophila, 33, 37 cohort mortality, age-specific, 509 Columbus livia, animal model, 19 connective tissue human aging, 141–144 structure and function, 141–142 coronary artery disease age-related change, 67 survival probabilities for men, 78 correlative evidence, aging, 6 corticosterone, 461–462 creatinine clearance, 57–58, 59 creature similarity, evolutionary theory, 106 cross-sectional studies creatinine clearance, 57–58, 59 drawbacks, 54–55 empirical longitudinal and, 56–62 height, 56 maximum oxygen uptake, 58–60, 61 point-of-time studies, 54 timing as measure of aging process, 60 weight, 56–57 cumulative, progressive, intrinsic and deleterious (CPID) age-related changes, 11, 12, 139–140 cytogerontology, cell culture, 420 Dall mountain sheep, survival curve, 32–33, 34, 36 death acceptance, 3 aging and probability of, 15–16 avoiding premature, 219–220 contributory causes in United States, 219

relative risk, 226, 227, 230, 231 vitality and aging rates, 17 dehydroepiandrosterone, 233, 465 delayers, centenarians, 332 demographic transition, longevity, 498–499 development, and senescence, 52 developmental change, age-related change vs., 9 developmental span, theory, 499–501 diabetes biological vs. chronological age, 84 hormonal age-related changes, 196–197 diet. See also nutrition human life span, 496–498 recommendations for healthy, 223 supplements, 234–235 dieting, human, 226–228 digestive system age-related changes, 167 normal structure and function, 166 diseases. See also specific diseases abnormal apoptosis, 447 aging and, 62–67, 69–71, 137–139 biomarkers of mortality and, 85–90 environmental changes, 68–69 interactions between aging and, 198–201 passage of time, 63–64 postmaturational changes, 71–72 disposable-soma theory, 104, 348 division of labor, human life span, 498 DNA damage theory level of damaged DNA, 372 reactive oxygen species damage, 386 senescent mechanism, 361 stochastically based, 367, 371–376 DNA repair, 375 benzo[a]pyrene conversion rate, 373 caloric restriction, 208 cells, 95–96 life span vs., 372 mitochondria and, 405–406 organ-specific age-related decreases, 374 in paramecia, 111 senescent mechanism, 361 stochastically based, 371–376 yeast cell mechanisms, 252 dormancy and diapause, 70 Down’s syndrome genetic defects, 375 premature aging, 333–334 Drosophila age-associated alterations, 124 age-specific mortality, 270, 275 animal model, 18–20, 267 behavioral and physiological trait loss, 269 biological time, 10 biomarkers, 90 caloric restriction, 211, 270, 272–273

Index

development vs. senescence, 52 ecdysone receptor protein, 288 expression of antioxidant genes, 279, 282–283 genetic stability, 284–286 insulinlike signaling pathway, 274, 303 life tables, 30, 32, 33, 37 longevity, 103–104, 268, 277–286, 289–291, 317, 416–418, 513 metabolic control and rate, 213, 267–277, 398 mitochondria and senescence, 277, 282, 283, 412–413, 415 model for endocrine circuits of aging, 287 protein interaction maps of, 490 rate-of-living theory, 396–398 reproductive effects, 286–289 somatic mutation theory, 370 stress resistance and extended longevity, 277– 284 superoxide dismutase gene, 280–281 survival curves, 32, 33, 35, 270 drug metabolism, age-related changes, 198 Duke First Longitudinal Study, 135 dysdifferentiation theory, 361, 376–378 ecology studies, life-history strategies, 104 Ecuador, superlongevity, 330–331 electron transport system, mitochondrial, 400 emphysema, 165–166 endocrine system caloric restriction, 208, 210–211 intercellular communication system, 179 link between nervous system and, 178 energy metabolism, 193–194 environmental changes, modulating aging, 68–69 environmental effects, 22, 54–55 enzymes, properties of altered, 122, 123 epicellular interactions, modeling aging, 493–494 error catastrophe theory, 361, 378–380 escapers, centenarians, 332 estradiol, age-related changes, 451–453 estrogen replacement, 233–234, 450, 453–455 ethics, anti-aging interventions, 515–522 ethnic groups, 327, 328–330 evidence correlative, 6 gain-of-function, 7–8 loss-of-function, 6–7 evolutionary models. See also models disposable-soma theory, 348 modern, 97–98 strengths and weaknesses, 105–107 evolutionary origins, caloric restriction, 209–211 evolutionary studies, life-history strategies, 104 excretory system age-related changes, 168–169 normal structure and function, 167–168 structure of nephron and blood supply, 168

595

exercise hormone induction via, 233 humans, 228–232 laboratory animals, 215–217 experimental organisms. See animal models; models extended health span anti-aging intervention, 511–515 scenario, 507, 524–525 extended senescence anti-aging intervention, 510–511 scenario, 507, 524 fecundity, longevity and, 98–104 female life expectancy, northern European, 512 female reproductive system age-related changes, 190–191 neuroendocrine senescence, 450–455 normal menstrual cycle, 190 normal structure and function, 188–189 fetal development, 70–71, 72 fibroblasts. See human skin fibroblasts fish, maximum life spans, 127 Floscularia, survival curve, 36 food. See diet; nutrition food restriction. See caloric restriction forced vital capacity, biomarkers, 75–77 frailty index, basis, 35 frailty syndrome, biomarkers, 85–86 Framingham study, biomarkers and aging, 75–77 free radicals cellular defenses against, 384 definition and reactions, 381–382 senescent mechanism, 361 fruit flies. See Drosophila fruits, antioxidant capacity, 385 fungi, mitochrondria in senescence, 118–119 future, pros and cons of controlling, 506 gain-of-function evidence, aging, 7–8 gender age predictions, 340, 341 bone loss, 149–150 life span differences, 323, 325–328 -specific biomarkers, 82–83 genealogical studies, 319–322 heritability of longevity, 320–322 parental age at death, 319 gene delivery, targeted, to mitochondria, 415–416 gene dysregulation, 115–116 gene expression modulation by caloric restriction, 293–302 neuroendocrine control, 455–460 gene expression networks, model, 500 gene interaction networks, aging, 489–493 genetic analysis, Saccharomyces cerevisiae, 239–240 genetic defects, human progeroid syndromes, 375 genetic instability, yeast life span, 115–116

596

Index

genetic manipulations, lab animals, 217–218 genetic mechanisms, senescence, 361 genetic responses, caloric restriction, 209 genetic stability Drosophila, 284–286 longevity process, 238 mice and mammals, 312–313 Saccharomyces cerevisiae, 250–252 genomic stability, Caenorhabditis elegans, 263–264 geography, 137, 331 gerbils, biochemical study, 388–389 germ cells, DNA repair functions, 95–96 gerontology life-span considerations, 4, 35–36 problems with aging models, 20–21 problems peculiar to, 20–21 role of theory, 347–349 view of aging and diseases, 62 Gerontology Society of America, 12 glucose, 240, 241, 365 blood levels, 14 intolerance, 222–223 glycoregulatory agents, cell damage, 212 Gompertz plots age-specific mortality and rate of aging, 43 initial mortality rate, 46 parameters, 46 survival curves, 43, 44, 50–51 Swedish females, 49 gonadal hormones, skeletal maturation, 147, 149 graying hair, phenomenon of, with age, 63–64 green algae, somatic and reproductive cells, 116–118 grip strength, biomarkers, 75, 76 growth curves adolescents, 60, 61 prudent selected organism, 100 growth hormone gene expression, 456 insulinlike signaling pathway, 302–306 intervention in humans, 232–233 supplementation, 306 growth retardation, caloric restriction, 206 guinea pigs, oxygen concentration, 380

Habrobracon, somatic mutation theory, 370 hair graying, phenomenon of, with age, 63–64 hamsters, oxygen concentration, 380 Hayflick hypothesis, 425–426 Hayflick limit, 110, 425 health, transition to senescence, 357–358 health span anti-aging intervention of extended, 511–515 delaying disease onset, 85 extended health span scenario, 507, 524–525 life span model, 358 longevity, 348

heart. See also cardiovascular system aging and caloric restriction, 294–295 structure and function, 156–157 hearing loss, 177 heat shock factor, 352 heat shock proteins Caenorhabditis elegans, 262 cytoprotective function, 391 mitochondrial function, 415 schematic of role in protein folding, 391 stress response, 390, 392 height regression, 55, 56 hematocrit, gender-specific ideal age, 340, 341 heritability, longevity, 321–322, 331–332 herring gull, survival curves, 36 hippocampus system, gene expression, 456–460 homeostatic process, steady state, 14 honeybee, 68–70 hormonal changes age-related, 195–198 caloric restriction, 210–211 induction via exercise, 233 human aging. See also senescence anatomical changes, 140–141 cardiovascular system changes, 155–162 cartilage, 144–145 connective tissue changes, 141–144 cumulative, progressive, intrinsic, and deleterious, 11, 12 digestive system changes, 166–167 ethnic and social differences, 328–330 excretory system changes, 167–169 history of mortality and longevity, 136 how-to-books, 9 immune system changes, 184–188 interactions between aging and disease, 137–139, 198–201 life expectancies by time and place, 137 metabolic and hormonal changes, 193–198 muscle tissue changes, 151–155 nervous system changes, 169–184 overview of theory, 139–140 perspective on, 135–136 plasticity of, 135–136, 140 relationship between aging and disease, reproductive system changes, 188–193 respiratory system changes, 163–166 sex differences and longevity, 325–328 skeletal system changes, 144–151 skin changes, 141–144 human aging, modulation arteries and blood pressure, 64–67 cardiovascular system, 157–158 cartilage, 144–145 developmental change vs., 9 digestive system, 167 diseases and time passage, 63–64

Index

excretory system, 168–169 female reproductive system, 190–191 fuel storage and utilization, 195 hormonal changes, 195–198 immune system, 185–188 male reproductive system, 192–193 measuring, 15–16 muscle tissue, 153–155 nervous system, 171–180 pros and cons of, 506 relationship between aging and diseases, 62–63 respiratory system, 164–165 skeletal system, 144–147 skin and connective tissue, 142–144 Human Genome Project, 339 human nerve growth factor, 459–460 humans. See also human aging animal model, 20 behavioral biomarkers, 86 effect of high oxygen concentration, 380 exercise, 228–232 fetal development, 71, 72 life span, 496-498 life table curves, 37, 38 mortality rates, 43, 45, 49–50 neuroendocrine status and aging rate, 464–465 physiological functions, 6, 7 potential panels of biomarkers, 75–90 struggle against fate, 3 survival data, 37–39 human segmental progeroid syndromes, 375 human skin fibroblasts cell differentiation, 427–428 cell doublings vs. donor age, 423 proliferative capacity vs. donor age, 424 Huntingdon’s disease, 338, 353 Hutchinson–Guilford syndrome, 334, 375 hypertension, 84, 337 hypophysectomy, senescence retardation, 466, 467 hypothalamic-pituitary-adrenal axis, 460–461 immune system age-related changes, 185–188 integration of neuroendocrine and, 473–475 intercellular communication system, 179 nervous system and, 179–180 normal structure and function, 185 senescence, 187, 466–470 thymus development and function, 186 immunological theories immunosenescence and aging, 466–470 inflammation and senescence, 470–473 senescence, 362 immunosenescence, aging and, 466–470 infant mortality, 41 inflammation, 470–473 information flow, genome to gene products, 379

597

information transfer, human life span, 496–498 insects, 123–126. See also Drosophila insulin mitochondrial function, 414–415 resistance, 216–217 insulinlike signaling pathway (ISP) cooperative effect on longevity, 352 evolutionary conservation, 131 metabolic control in Drosophila, 273–276 metabolic control of longevity, 302–306 summary of longevity pathways, 316–318 insulin sensitivity, 332–333 intelligence, 496 intercellular communication, schematic, 179 intercellular regulatory processes basic assumptions, 449–450 neuroendocrine-immune system, 473–475 signal transduction, 475–479 theories of senescence, 450–473 interspecific plasticity, aging, 22–23 interventions. See anti-aging interventions intracellular regulatory processes aging in dividing cells, 419–437 senescence in nondividing cells, 437–443 terminally differentiated cells, 443–448 intraspecific plasticity, aging, 21–22 invertebrates aging patterns, 119 biomarkers, 89–90 caloric restriction, 211 insects, 123–126 maximum recorded longevities, 120 mitochondria, 412–413 nematodes, 122–123 rotifers, 119–121 Kass, Leon, ethics of aging interventions, 516–519 kidney, aging, 168–169 kidney disease, time-related, 64 knockouts, loss-of-function evidence, 6–7 K-selected strategy, 98–100, 102 laboratory animals. See also animal models exercise, 215–217 genetic manipulations, 217–218 manipulations of metabolic rate, 212–215 transplantation of tissues and organs, 218 vitamin and antioxidant treatment, 217 lactation, nutrition in maternal mice, 70–71 Lansing effects, aging, 121 lapwings, survival curve, 29, 31 learning, caloric restriction in mice, 208–209 life expectancy genealogical studies, 319–322 human by place and time, 137 life tables and survival curves, 35 sex-specific differences, 326

598

Index

life history manipulating parasitic wasp, 100–101 strategies, 98–99, 104 life span across species, 127 bristlecone pine, 68 caloric restriction, 207 children with very long-lived parent, 323 concept of maximum, 53 correlation with DNA repair, 372 factors influencing human, 496–498 female, 323 health span and senescent span, 24 model update, 358 mortality, 41, 509 nematode, 259 oxidative damage and maximum, 411 philosophy of extending, 506 relation of cell doublings to, 426 relationship to reproduction, 98–104 term, 17–18 life-span clock, gerontology, 4 lifestyle factors and aging, 235–236 longevity and, 330 reducing chronic conditions, 86 life tables. See also survival curves cohort, 37 constructing, 37–41 curves for indirect, 37, 38 Dall mountain sheep, 34 hypothetical, 29 numerical relationships, 26–28 organisms with constant mortality rate, 27 organisms with constant number of deaths, 29, 30 period, 37 relationship between age and mortality, 35 survival statistics, 26 lifetime energy potential, value by species, 214– 215 lipids, reactive oxygen species damage, 386 lipofuscin, cellular aging, 443 lipoic acid, mitochondrial function, 413–414 liver, 297, 299, 300 liver disease, abnormal apoptosis, 447 longevity. See also life span analysis by Alexander Graham Bell, 319 caloric restriction, 202–203, 251, 256 comparison of mammalian and avian, 130 determinants, 494 estrogen exposure in mice, 454 extended, 261–263, 277–284 fecundity and, 98–104 gain-of-function evidence, 7–8 genealogical studies, 319–322 gene interaction diagram for mammals, 491

heritability, 331–332 human, 39, 136 implications of theory, 499 individual variability, 393–394 insulinlike signaling pathway, 254–259, 273–276 mathematical limits, 52–53 maximum, in invertebrates, 120 metabolic control, 259–261, 267–273, , 276–277, 416–418 pathways in Drosophila, 291 phenotypes, 21, 511–512, 513 population studies, 322–324 processes in model systems, 238 rodent biomarkers, 88–89 role of mitochondria, 409–412 selected, 495–496 sex differences and, 325–328 stress responses, 393–394 superlongevity, 330–331 twin studies, 324–325 longitudinal studies age-related changes with time, 55–56 creatinine clearance, 57–58, 59 cross-sectional comparisons, 56–62 height, 56 maximum oxygen uptake, 58–60, 61 physiological and other predictors, 340–343 practice effect, 55 timing as measure of aging process, 60 weight, 56–57 long-lived animals, mitochondria of, 409–412 loss-of-function evidence, aging, 6–7 Macaca mulatta, animal model, 19–20 macaques, biomarkers, 87, 88, 389 male reproductive system age-related changes, 192–193 normal structure and function, 191–192 serum testosterone levels, 192 mammals. See also mice; rats aging patterns, 126–129 changes for extended longevity, 416–418 life span and metabolic rate, 213–214 maximum life spans, 127 reproductive schedules and longevity, 100, 101 mating scheme, genetically heterogenous mice, 292 measurement age-related changes, 15–16 of aging, 11, 51–52, 239 medflies, mortality rates, 48 medical limits, anti-aging interventions, 519–520 medical model aging, 13 aging as disease, 62 evolutionary view, 105 Melopsittacus undulatus, animal model, 19

Index

memory formation, mitochondrial function, 414 mental abilities, changes, 176–180 metabolism basal metabolic rate, 193, 194 changes in cerebral, 175–176 fuel utilization and storage, 194–195 pharmacological age-related changes, 198 rate-of-living theory, 213, 396–398 metabolic control in Caenorhabditis elegans, 254–259 in Drosophila, 267–277 longevity process, 238 in Saccharomyces cerevisiae, 240–248 summary of longevity pathways, 316–318 yeast life span, 115 metabolic efficiency, caloric restriction, 208 metabolic mechanisms, senescence, 361 metabolic rate, manipulations in animals, 212–215 metabolic syndrome, age-related changes, 197–198 mice, 18–19, 290–293 caloric restriction, 205, 208–209, 297, 298, 293– 302 characteristics, 305 comparison with other organisms, 303 cumulative survival of wild-type and mutant, 309 genetic defects for accelerated aging, 376 genetic stability, 312–313 growth hormone, 303, 306 hair graying, 63 insulinlike growth factor, 302–306, 441, 514 longevity and estrogen exposure, 454 longevity predictor, 89 mating scheme for genetically heterogenous, 292 metabolic control, 306–307 neuroendocrine-immune system, 473–475 oxidative stress, 310 patterns of senescence, 314–315 phenotypic plasticity, 21 physical activity, 216 pregnancy and lactation nutrition, 70–71 reproductive effects, 313–314, 450–455 stress, 307–312 summary of longevity pathways, 318 micronutrient supplementation, thymus, 469 minerals, human nutrition, 224–226 mitochondria cellular energy demand, 402 electron transport system, 400 estrogen protection of, 329 interventions that affect function, 413–414 of long-lived animals, 409–412, 413 mutation in centenarians, 333 role in senescence for fungi, 118–119 senescence of invertebrate, 412–413 structure and function, 399–402 targeted gene delivery, 415–416

599

mitochondrial damage. See also heat shock proteins differential damage cascade, 403 DNA repair, 405–406 DNA and senescence, 402–404 genomes contributing to respiratory chain, 401 insulin, 414–415 memory formation, 414 oxidative damage and stress, 328, 398–399, 404– 405, 407, 410, 411 sarcopenia, 407–409 vitamin E, 413 mitogen-activated protein kinase cascade, 478– 479 models. See also animal models of aging, 13, 16–21 architecture of scale-free network, 489 of cell transition, 354 conceptual, for data on aging, 23–24 early network model, 486–488 epicellular interactions, 493–494 experimental organism, 18–20 gene interaction networks, 489–493 gerontology problems, 20–21 hierarchical networks, 488–494, 500 hippocampus and brain aging, 457–458 life span updated, 358 linear model to senescence and death, 485–486 MARS (mitochondria, aberrant proteins, radicals, and scavengers) network model, 487–488 modern evolutionary, 97–98 need for and costs, 237–239 organizational level, 16–18 red blood cell, 443–445 regulation of life span, 258 molecular-level studies, senescence, 352–355 molecular signature, aging, 11 monkeys, 380, 389, 459–460 Mormons, lifestyle, 330 mortality. See also age-specific mortality assessing changes and age dependence, 356 benefits by Leon Kass, 516–518 biomarkers, 77, 79, 81–82, 85–90 compression, 515 effects of procaine on rats, 43, 45 gender-specific, 327–328, 340, 341 human, 37–39, 41, 42, 136 late-life, 41–42 relationship to age, 35 slowing of rate, 41 Weibull function, 51 mortality kinetics comparisons, 43–48 late-life, 48–50, 53 senescence patterns, 46, 47 mortality rates, 48, 325 doubling time, 46, 77 mouflon sheep, survival curve, 29, 31

600

Index

mountain sheep, survival curves, 32–33, 34, 36 Muller’s ratchet, decline in fitness, 105 multicellular organisms, Volvox and, 116–118 muscle tissue age-related changes, 153–155 atrophy, 153–154 normal structure and function, 151, 153 organization, 152 sarcopenia, 153–154 Mus musculus, animal model, 18–19 mutation accumulation hypothesis, 98 National Institute on Aging, 19 National Library of Medicine, 24 National Research Council, 223 negligible senescence long-lived species, 23 mortality kinetics of organisms, 47 nematodes. See also Caenorhabditis elegans animal model, 18–19, 353 biomarkers, 89–90 caloric restriction, 211 diapause, 70 insulinlike signaling pathway, 254–259, 274, 303, 352 longevity and senescence, 122–123, 352 summary of longevity pathways, 316–317 nephron, structure and function, 168–169 nerve growth factor, 459–460 nervous system age-related pathologies, 180–184 brain and anatomy of central nervous system, 170 changes in cerebral metabolism, 175–176 changes in mental abilities, 176–180 changes in neurotransmitter function, 174–175 endocrine systems and, 178 Golgi-stained neurons, 173 gross structural changes, 171–172 immune system and, 179–180 intercellular communication, 179 microscopic structural changes, 172–174 normal structure and function, 169–171 thermoregulation defect, 178–179 neurodegenerative disorders, 447 neuroendocrine hormones, 465–466 neuroendocrine-immune system, 473–475 neuroendocrine system caloric restriction, 465–466 dehydroepiandrosterone, 465 factors modifying status and aging rate, 461– 465 gene expression and control, 455–460 hormonal phenomena, 465–466 hypophysectomy, 466, 467 senescence, 362, 450–455 stress responses, 460–461

Neurospora biological time, 10 mitochrondrial role in senescence, 118–119 neurotransmitters, change in function, 174–175 nitrogen metabolism, mice, 300 non-aging systems, 13–14 nondividing cells blood cells and cardiovascular senescence, 442 cardiac myocytes, 439–442 endothelial cells, 438–439 fibroblasts and skin aging, 437–438 insulinlike growth factor, 441 lipofuscin and waste accumulation theory, 443 senescence and differentiation in vivo, 437 stem cells and senescence in tissues, 442–443 nonsegmental biomarkers. See biomarkers nourishment, human life span, 496–498. See also nutrition nuclear-mitochondrial interaction Caenorhabditis elegans, 259–261 Drosophila, 276–277 mice and mammals, 306–307 Saccharomyces cerevisiae, 246–248 nucleic acids, reactive oxygen species damage, 386 Nun Study, 171, 180–181 nutrition human adult, 220–224 human fetal, 71, 72 pregnancy and lactation in mice, 70–71 organisms. See also animal models choosing experimental, 18–20 need and costs of model, 237–239 organizational level, aging models, 16–18 organ transplantation, 218 osteoporosis bone mineral density in women, 148 bone pathology, 147, 149–151 oxidative damage and stress. See also mitochondrial damage; reactive oxygen species antioxidant capacity of fruits and vegetables, 385 antioxidant strategy, 387–389, 390 Caenorhabditis elegans, 262 cellular defenses against free radicals, 384 differential damage cascade by, 403 Drosophila, 280 effects of high oxygen concentrations, 380 estrogen protection, 329 maximum life span, 411 mechanisms in hypothetical cell, 383 methionine, 388 mimetics of antioxidant enzymes, 388 network of antioxidant action, 386 reactive oxygen species, 381–383, 386 sex and, 328 signaling pathways, 311 stochastically based, 380–390

Index

vitamin C, 384–385 vitamin E, 384 oxygen radicals. See reactive oxygen species oxygen uptake, changes, 58–60, 61 Pakistan, superlongevity, 330–331 paramecia. See also protists aging process, 111 somatic and reproductive nuclei, 96 parasitic wasps, manipulating life history, 100–101 pathogens, inflammation, 472–473 pathologies caloric restriction effects on, 203–205 failure of maintenance, 138 Pauling, Linus, ascorbic acid, 224 perennial trees, environment and aging, 68 period, chronological changes, 56 pharmaceutical interventions, 514–515 pharmacological changes, age-related, 198 phenotypic plasticity, intraspecific, 21–22 4-phenylbutyrate, 285–286 philosophy extending life span, 506 finitude and ideas of Leon Kass, 516–519 objections to anti-aging interventions, 516–521 physiological age, chronological age vs., 10 physiological biomarkers, panels, 78–79 physiological functions, humans, 6, 7 physiological responses, caloric restriction, 205– 209 physiological variables aging studies, 75, 84 longitudinal studies, 340–343 pigeons, model, 19 plants, aging and senescence, 130 plasticity aging, 21–23,135–136, 140 interspecific, 22–23 intraspecific, 21–22 phenotypic, 21–22 Podospora, mitochrondria in senescence, 118–119 point-of-time studies, cross-sectional, 54 polycystic kidney disease, time-related, 64 polyglutamine, Huntingdon’s disease, 352–353, 354 polyunsaturated fatty acids in cells, 389–390 in long-lived animals, 409–411 population age distribution, 39, 40 cell doubling, 421–426 growth rate, 105–106 longevity studies, 322–324 median age of, 51–52 population-level studies large and small effects, 349–350 mutations altering blood pressure in humans, 350

601

pathogenesis of hypertension, 351 senescence, 349–352 positive arguments, anti-aging interventions, 521–522 post-translational protein modification theory, 361, 364–366 practice effect, longitudinal, 55 pregnancy, nutrition in maternal mice, 70–71 premature death humans avoiding, 219–220 mechanism in female flies, 289, 290 prenatal stress, rats, 462–463 primate biomarkers, 87, 88 primate longevity, hypothesis, 128 procaine, effect on rat mortality, 43, 45 prodigal strategy (r) distinction from prudent (K) strategy, 99–100 fecundity and longevity, 98–99 traits of life-history strategy, 99 programmed cell death, controllable, 445–448 protein kinase A pathway, Saccharomyces cerevisiae, 241, 245 proteins Drosophila interaction maps, 490 glucose-based cross-linking, 365 interaction network, 493 reactive oxygen species damage, 386 protists. See also unicellular organisms aging, 110–112 amoebae, 110–111 Hayflick limit, 110 paramecia, 111 Tetrahymena, 112 prudent strategy (K) distinction from prodigal (r) strategy, 99–100 fecundity and longevity, 98–99 growth curve, 100 longevity, 100, 102 traits of life-history strategies, 99 quality, animal model, 20 quantitative trait loci (QTL) mapping, 339 random predation, 28 rapid senescence mortality kinetics of organisms, 47 short-lived invertebrates, 22–23 rate of living theory, metabolism, 213, 396–398 rats caloric restriction and mortality, 43–44, 45 effect of high oxygen concentration, 380 effect of procaine on mortality, 43, 45 exercise, 215–217 influence of diet, 203–204 muscle fiber, 154 prenatal stress, 462–463 sarcopenia, 407–408 stress response, 460–461

602

Index

reactive oxygen species inflammation and senescence, 470–471 oxidative damage, 381–384, 386 red blood cell model, 443–445 replicative aging, Saccharomyces cerevisiae, 239 replicometer, telomere, 432 reproduction caloric restriction and, 212 relationship to life span, 98–104 reproductive effects Caenorhabditis elegans, 264–266 Drosophila, 286–289 longevity process, 238 mice and mammals, 313–314 Saccharomyces cerevisiae, 253 summary of longevity pathways, 316–318 reproductive system age-related changes of female, 190–191 age-related changes of male, 192–193 female normal function and structure, 188–189 male normal function and structure, 191–192 senescence, 450-455 reptiles, maximum life spans, 127 respiratory chain, genomes contributing to, 401 respiratory system age-related changes, 164–165 age-related pathologies, 165–166 normal structure and function, 163–164 pulmonary volumes and capacities, 165 structure of human, 163 retail approach, aging and disease, 139 rhesus monkeys animal model, 19–20 caloric restriction vs. ad libitum feeding, 210 human nerve growth factor and hippocampus, 459–460 potential panel of biomarkers, 87, 88 rodents, survival curves of wild-type, 43, 44. See also mice; rats Rothmund–Thomson syndrome, 375 rotifers power of theory, 119–121 r-selected strategy, 98–100 Saccharomyces cerevisae, 239–240. See also yeasts aging, 239 genetic stability, 250–252 caloric restriction, 240–245 metabolic control, 246–248 nuclear-mitochondrial communication, 248 reproductive effects, 253 stress resistance, 248–250 survival characteristics, 113 sarcopenia mitochondrial disorder, 407–409 muscle atrophy, 153–154 rat skeletal muscle, 408

schizophrenia, age-related pathology, 184 segmental biomarkers. See biomarkers segmental progerias, premature aging, 333–335 selective mortality, cross-sectional, 55 senescence. See also aging cell signaling path to replicative, 433–437 cellular, 421–428 comparison of mammalian and avian, 130 definitions, 11–13, 26 distinguishing between development and, 52 extended senescence scenario, 507, 524 gradual, 23, 33, 47 inflammation and, 470–473 insect, 124–125 integrated theory of aging and, 357–358 mechanisms, 8, 23–24, 131, 358–362 mitochondria and, 402–404, 412–413 negligible, 23 pathways leading to cell, 436 patterns, 46, 47, 266, 289, 291, 314–315 plant, 130 rapid, 22–23 scientific explanation, 95 summary of longevity pathways, 316–318 viewing as continuum, 22 senescence, genetics of age-dependent changes, 355–357 angiotensin-converting enzyme gene, 337 apolipoprotein E gene, 335–336 effect of insulinlike signaling, 352, 354 Human Genome Project, 339 human leukocyte system alleles, 337–338 in Huntingdon’s disease, 338, 353 hypertension, 331, 350, 351 large and small effects, 349–352 studies of, 338–340, 349–355 quantitative trait loci, 339, 355 somatic gene therapy, 339–340 senescent span, life span model, 358 senses, age-related changes, 176–178 serum cholesterol levels, age-related increase, 162 Seventh Day Adventists, 235–236, 330, 340 sex, U.S. population by, 40 sex differences, longevity, 325–328 sheep, survival curves, 29, 31, 32–33, 34, 36 signaling mechanisms, senescence, 362 signaling pathways, oxidative stress, 311 signal transduction activation, 476, 477, 478 cell proliferation and survival, 476 gene expression regulation, 475–479 mitogen-activated protein kinase cascade, 478–479 number of signaling molecules, 477 silent information regulator gene family, Saccharomyces cerevisiae, 241–245 skeletal muscle, aging and caloric restriction in mice, 294–295

Index

skeletal system bone, 145–151 cartilage, 144–145 changes with human aging, 144–151 interactions between aging and disease, 199–201 skin and connective tissue age-related changes, 142–144 human aging, 141–144 normal structure and function, 141–142 wrinkles, 143–144 skin cancer, environmental effect, 68 smell, age-related change, 177 social differences, aging, 328–330 social effects, anti-aging interventions, 522–524 social justice, anti-aging interventions, 520–521 somatic cells maintenance, 102–103 senescence, 95 somatic gene therapy, 339–340 somatic mutation theory senescent mechanism, 361 stochastically based, 367, 368–371 stationary phase method, yeasts, 113–114 steady-state system, non-aging, 14 stem cells, therapeutic cloning, 218 stochastic process, cell death, 443–445 stochastic theories accelerated aging in mice, 376 altered protein, 366–367 benzo[a]pyrene conversion rate, 373 cellular defenses against free radicals, 384 DNA damage and repair, 371–376 dysdifferentiation, 376–378 effects of oxygen concentrations on tissues, 380 epigenetic aspects of stress responses, 393–394 error catastrophe, 378–380 glucose-based cross-linking of protein, 365 heat shock proteins, 390–392 human segmental progeroid syndromes, 375 information from genome to gene products, 379 network view of antioxidant action, 386 origin of change, 359 oxidative damage, 380–390 post-translational protein modification, 364–366 somatic mutation, 367–376 wear-and-tear, 363–364 stress aging and longevity in humans, 465–466 DNA damage, 252 mortality rates vs. age in humans, 43, 45 responses, 251, 393–394, 460–461 stress resistance Caenorhabditis elegans, 261–263 Drosophila, 277–284 longevity process, 238 mice and mammals, 307–312 Saccharomyces cerevisiae, 248–250

603

summary of longevity pathways, 316–318 yeast life span, 115 superlongevity, myth or reality, 330–331 survival curves. See also life tables biological meaning of transformed, 43–48 British males, 38 Drosophila, 32, 33, 35, 270 caloric restriction, 203, 207 comparison of species, 36 compilation of types, 34, 35 hypothetical, 29, 30 influence of diet, 203 linear decrease with time, 28 organisms with constant mortality rate, 28 organisms with constant number of deaths, 30 population with high juvenile mortality, 32–33 random predation, 28 relationship to Gompertz mortality curves, 50–51 sheep, 29, 31–34 special transformations, 41–52 wild lapwings, 29, 31 wild-type rodent in captivity, 44 survivors, 20, 332 Sweden, age-specific mortality over life span, 509 synthetic oligopeptide analogues, thymus, 469–470 systemic theories changes for extended longevity, 416–418 mechanism, 348–349, 360 metabolism, 396–398 mitochondrial damage, 398–416 origin of change, 359 rate-of-living, 396–398 variability in aging, 395–396 targeted gene delivery, mitochondria, 415–416 taste, age-related change, 176–177 telomeres cellular senescence, 428–433 computer model of, loss, 432–433 description, 429, 430 hypothesis of cell aging, 429 replicometer, 432 Tetrahymena, aging process, 112 theory aging, 24, 95, 139–140 role in gerontology, 347–349 senescent mechanisms, 361–362 therapeutic cloning, stem cells, 218 thermoregulation, neuroendocrine control, 178– 179 thymus age-related changes, 185, 186, 468–470 time. See also actuarial analysis diseases and passage of, 63–64 human life expectancy, 137 independent variable, 10 measure of aging process, 60

604

Index

time-related alterations, aging, 10 tissue. See also skin and connective tissue influence of diet, 203–204 transplantation, 218 touch, age-related change, 177 toxicity, caloric restriction and, 211 transplantation, tissue and organ, 218 trichothiodystrophy, genetic defects, 375 tumors, influence of diet, 203–204 twin studies, life span, 324–325 unicellar organisms. See also bacteria; yeasts aging, 96–97, 107–116 classification scheme, 107 protists, 110–112 United States age distribution of population, 40 age-specific mortality, 42 vegetables, antioxidant capacity, 385 veins, structure and function, 155–156 vertebrates birds, 129–130 general patterns, 126 mammals, 126–129 maximum reported life spans, 127 plants, 130 viral infections, abnormal apoptosis, 447 vision, age-related change, 177–178 vital capacity, gender-specific ideal age, 340, 341 vitality, aging rates, 17 vitamin C, 384–385 vitamin D, 167 vitamin E, 384, 413

vitamin treatment human nutrition, 224–226, 234–235 laboratory animals, 217 Volvox carteri, 116–118 wasps, manipulating life history, 100–101 waste accumulation cellular aging, 443 senescent mechanism, 361 wear-and-tear theory senescent mechanism, 361 stochastically based, 363–364 Weibull function, mortality, 51 weight cross-sectional vs. longitudinal data, 55–57 control inhumans, 226–228 Werner’s syndrome cell doublings vs. donor age, 423 genetic defects, 375 premature aging, 334–335 wheel animalcules, rotifers, 119 wholesale approach, aging and disease, 139 women, osteoporosis, 147–150 worker bees, environmental changes, 69 wrinkles, skin aging, 143–144 yeasts. See also Saccharomyces cerevisiae budding life span method, 112–114 cell spiral model, 114–115 caloric restriction and stress response, 251 detecting and repairing DNA damage, 252 plant model, 18–19 stationary phase method, 113–114 stress-response pathway, 274, 303 summary of longevity pathways, 316