International Review of Cytology: A Survey of Cell Biology, Volume 26

I N T E R N A T I O N A L REVIEW OF CYTOLOGY VOLUME26 Contributors to Volume 26 P. BORST KONHADKECK A. M. KROON GI...

Author: G. H. and Danielli | J. F. Bourne

12 downloads 997 Views 18MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

I N T E R N A T I O N A L

REVIEW OF CYTOLOGY VOLUME26

Contributors to Volume 26

P. BORST KONHADKECK

A. M. KROON GILBERTN. LING

D. W. A. ROBERTS

EWALDR. WEIBEL LEONARDWEBS

INTERNATIONAL

Review of Cytology EDITED BY

G. H. BOURNE

J. F. DANIELLI

Yerkes Regional Primate Research Center Emory University Atlanta, Georgia

Center for Theoretical Biology State University of New York at Buffalo Buffalo, New York

ASSISTANT EDITOR K. W. JEON Center for Theoretical Biology State University of New York at Buffalo Buflalo, New York

VOLUME26

Prepared Under the Auspices of The international Society for Cell Biology

ACADEMIC PRESS New York and London 1969

COPYRIGHT @ 1969, BY ACADEMIC PRESS. I N C . A1.I. RIGHTS RIiSERVI:D. N O PARI' OF THIS ROOK MAY Rli REPRODIJCED I N A N Y FORM, BY PHOTOSTAT, MICROFILM, RE'I'RIEVAJ. SYSTEM, OR A N Y O T H E R MEANS, W I T H O U T WRITTEN PERMISSION FROM 'JH11 PUBLISHERS.

ACADEMIC P R E S S , INC. 1 1 1 Fifth A v e n u e , New York, New York 10003

l'witeu' K i ~ q d o mEdition published by

ACADEMIC

PRESS,

INC. (LONDON) LTD.

Berkeley Square House, London W 1

LIRRAHY 01:CONGRI~SS CA'I'ALOG CARDNUMBER: 52-5203

PRIN1F.D IN T H E UNI'IBD STATES 01. AMEKICA

List of Contributors P. BORST,Department of iMediral Enzymologj, Luboiatory University of Am rtejdam, A m iterdam, T h e Netheilaud KONRADKECK, Department TUCJ~O?~, Arizoiiu

of

of

Biochemiitr),

Biologii-ul Scieiire~, Uiiiz'es.rjty o j Arizoua,

A. M. KROON,Depurtmetit of Medirul E n z y m o l o g ~ ,Laboratory University of Am iterdam, A m s t e r d ~ ~ n The z , Netherluiidr

of

BiochemiJtrj,

GILBERT N. LING, Department of Molecular Biology, Division of Neurology, Pennsylvania Hospital, Philadelphia, Pennsylvania The Netherluiidr

D. W. A. ROBERTS,Rejearrh Stutioti, Cmudu Department Lethbridge, Albestu, Canada

of

Agsii-iiltzrre,

EWALDR. WEIBEL, Depnstmeizt of Aiiutomy, VliiverJity of Beru, Besrz, Swifzerland LEONARDWEISS, Department of Experimental Pathology, RoJwell Park M e m o rial Institute, Buffalo, N e w York

This Page Intentionally Left Blank

Contents LIST OF CONTRIBUTORS ...................................................

v

CONTENTSOF PREVIOUS VOLUMES ........................................

xi

A New Model for the Living Cell : A Summary of the Theory and Recent Experimental Evidence in Its Support GJLEERT N . LING Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1

I. The Membrane Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2

11. An Interesting Clue in the Search for it Better Model of the Living Cell . . . . . 111. The Association-Induction Hypothesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

10 11 58

The Cell Periphery LEONARD WEISS Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Lipid Bilayers . . . . . . . . . . . . . . . . . . . . . . . . . . ...................... III. Other Models . . . . . . . . . . . . . . . . . . . . . . . . . . . ...................... IV. Cell Surface Charge . . . . . . . . .................................. V. Enzyme Activity and the Cell ery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI. The Peripheries of Malignant .............................. References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

I.

11.

63

64 70 78

91 94 99

Mitochondrial DNA : Physicochemical Properties, Replication, and Genetic Function P. BORSTA N D A. M. KROON

I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. Base Composition of Mitochondrial DNA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111. Nearest-Neighbor Frequencies of Mitochondria1 DNA . . . . . . . . . . . . . . . . . . . vii

108 109 117

...

Vlll

CONTENTS

IV . Differences in Base Composition and Base Sequence of the Complementary Strands of Mitochondrial DNA's . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V . Size and Structure of Mitochondrial D N A from Animal Tiss VI . Size and Structure of Mitochondrial D N A from Plants and Unicellular Organ isms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VII . The Amount of Mitochondrial D N A per Mitochondrion and per Cell . . . . . . VIII . Replication of Mitochondrial D N A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IX . Effects on Ycast Mitochondrial D N A of Anaerobiosis GIu and Mutagenic Agents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . X . Recombination of Mitochondrial D N A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XI . Renaturatioii Studies with Mitochondrial D N A . . . . . . . . . . . . . . . . . . . . . . . . . XI1 . Evolution of Mitochondrial D N A and the Relation between Mitochondrial ....................................... and Nuclear D N A . rial DNA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XI11. Genetic Function of Mitocl ................................................ XIV . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

.

117 118 139 141 145 154 163 165 167 168 179 181

Metabolism of Enucleated Cells KONRADKECR 1 . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I 1. Initiation of the Anucleate State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I I I . Quantitation of mRNA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV . Decay of inRNA and Protein Synthesis in Anucleate Cells ............. V . Nature of mRNA Decay . . . . . . . . . . . . . . . . . . . . . . . . . . . . ............. VI . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...................... References . . . . . . . . . . . . . .......................................

191 192 196 208 222 225 225

Stereological Principles for Morphornetry in Electron Microscopic Cytology

I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I1 . Fundamental Stereological Principles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ill . Application of Stereological Methods to Electron Microscopic Cytology . . . . . IV . An Example of Morphometric Characterization of Organelles: The Liver Cell V . Cytnmorphometric Methods in Experimental Pathology . . . . . . . . . . . . . . . . . . VI . Problems Arising in Applying Stereological Methods to Anisotropic Systems VII . Appreciation of Present State and Outlook on Future Possibilities . . . . . . . . . . References . . . . . . . . . . . . . .......................................

235 238 261 286 293 294 298 299

ix

CONTENTS

Some Possible Roles for Isozymic Substitutions during Cold Hardening in Plants D . W . A . ROBERTS I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. Effects and Prevention of Ice Formation . . . . . . . . . . 111. The Effect of Low Temperature on Proteins . . . IV . Metabolic Imbalance . . . . . . . . . . . . . . . . . . . . . . . . V . The Hypothesis of lsozyiiiic Substitution . . . . . . . . VI . Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

................... ................... ................... .................. ................. ................... . .................

AUTHORINDEX S U B J f I C T INDEX

303 304 309 313 318

322 323

329 .........................

...................

348

This Page Intentionally Left Blank

Contents of Previous Volumes Aspects of Bacteria as Cells and as Organisms-STumT MUDDAND EDWARD Some Historical Features in Cell BiolD. DELAMATER O~Y-ARTHUR HUGHES Ion Secretion in Plants-J. F. SUTCLIFFE Nuclear Reproduction-C. LEONARDHuS- Multienzyme Sequences in Soluble ExKINS tracts-HENRY R. MAHLER Enzymic Capacities and Their Relation The Nature and Specificity of the Feulgen to Cell Nutrition in Animals-GEORGE Nucleal Reaction-M. A. LESSLER W. KIDDER Quantitative Histochemistry of PhosphaThe Application of Freezing and Drying taSeS-wILLIAM L. DOYLE Techniques in Cytology-L. G. E. BELL Alkaline Phosphatase of the NucleusEnzymatic Processes in Cell Membrane M. CHBVREMONT AND H. FIRKET Penetration-TH. ROSENBERGAND w. Gustatory and Olfactory Epithelia-A. F. WILBRANDT BARADI AND G. H. BOURNE Bacterial Cytology-K. A. BISSET Growth and Differentiation of Explanted Protoplast Surface Enzymes and AbsorpTissues-P. J. GAILLARD tion of Sugar-R. BROWN Electron Microscopy of Tissue SectionsReproduction of Bacteriophage-A. D. A. J. DALTON HERSHEY A Redox Pump for the Biological PerThe Folding and Unfolding of Protein formance of Osmotic Work, and Its Molecules as a Basis of Osmotic Work Relation to the Kinetics of Free Ion R. J. GOLDACRE Diffusion across Membranes-E. J. Nucleo-Cytoplasmic Relations in AmphibCONWAY ian Development-G. FANK-HAUSER A Critical Survey of Current Approaches Structural Agents in Mitosis-M. M. in Quantitative Histo- and CytochemSWANN istry-DAvID GLICK Factors Which Control the Staining of Nucleo-cytoplasmic Relationships in the Tissue Sections with Acid and Basic Development of Acetdularia-J. HAMDyes-MARCUS SINGER MERLING The Behavior of Spermatozoa in the Report of Conference of Tissue Culture Neighborhood of Eggs-Lorn ROTHSWorkers Held at Cooperstown, New Volume 1

CHILD

The Cytology of Mammalian Epidermis and Sebaceous Glands-wrLLrAM MON-

York-D.

C. HETHERINGTON

AIJTHOR INDEX-SUB

JECT INDEX

TAGNA Volume 3 The Electron-Microscopic Investigation of The Nutrition of Animal Cells-CHARITY Tissue Sections-L. H. BRETSCHNEIDER WAYMOUTH The Histochemistry of EsterasesCaryometric Studies of Tissue CulturesG. GOMORI OTTO BUCHER AUTHOR INDEX-SUBJECT INDEX The Properties of Urethan Considered in Volume 2 Relation to Its Action on MitosisIVORCORNMAN Quantitative Aspects of Nuclear Nucleoproteins-HEWSON SWIFT Composition and Structure of Giant Chromosomes-MAx ALFERT Ascorbic Acid and Its Intracellular Localization, with Special Reference to How Many Chromosomes in Mammalian Somatic Cells?-R. A. BEATTY Plants-J. CHAYEN

xi

XI1

LONTENTS 01: PREVIOUS VOLUMES

The Chemical Composition of the Bacterial Cell Wall-C. S. CUMMINS Theories of Enzyme Adaptation in Microorganisms-J. MANDELSTAM The Cytochondria of Cardiac and Skeletal MUSCI~--JOHN W. HARMAN The Mitochondria of the Neuron-WARREN ANDREW The Results of Cytophotometry in the Study of the Deoxyribonucleic Acid (DNA) Content of the NucleusR. VENDRELY AND C. VENDRELY Protoplasmic Contractility in Relation to Gel Structure : Temperature-Pressure Experiments on Cytokinesis and Amoeboid Movement-DOUGLAS MARSLAND Intracellular pH-PETCR C. CALDWELL The Activity of Enzymes in Metabolism and Transport in the Red Cell-T. A. J. PRANKERD AUTHOR INDEX-SUB J E C T INDEX Uptake and Transfer of Macromolecules Volume 4 by Cells with' Special Reference to Growth and Development-A. M. Cytochemical Micrurgy-M. J. KOPAC SCHECHTMAN Amoebocytes-L. E. WAGGE Cell Secretion: A Study of Pancreas and Problems of Fixation in Cytology, HistolSalivary Glands-L. C. U. JUNQUEIRA ogy, and Historhemistry-M. WOLMAN AND G. C. HIRSCH Bacterial CYtOlOgY-ALFRED MARSHAK The Acrosome Reaction-JEAN C . DAN Histochemistry of Bacteria-R. VENDRELY Cytology of Spermatogenesis-VrsHwA Recent Studies on Plant MitochondriaNATH DAVIDP. HACKETT The Ultrastructure of Cells, as Revealed The Structure of Chloroplastsby the Electron Microscope-FRITIor: K. M ~ ~ H L E T H A L E R S. SJOSTRAND Histochemistry of Nucleic Acids-N. B.

The Significance of Enzyme Studies on Isolated Cell Nuclei-ALEXANDER L. DOUNCE The Use of Differential Centrifugation in the Study of Tissue Enzymes-CHR. DE DUVEAND J. BERTHET Enzymatic Aspects of Embryonic Differentiation-TRYGGVE GUSTAFSON Azo Dye Methods in Enzyme Histochemistry-A. G. EVERSON PEARSE Microscopic Studies in Living Mammals with Transparent Chamber MethodsROYG . WILLIAMS The Mast Cell-G. ASBOE-HANSEN Elastic Tissue-EDWARD w. DEMPSEY AND ALBERTI. LANSING The Composition of the Nerve Cell Studied with New Methods-%ENOLOFBRA~TGARD AND HOLGER HYDEN

AUTHOR INDEX-SUB JECT INDEX KURNICK Structure and Chemistry of NucleoliVolume 6 W. S. VINCENT On Goblet Cells, Especially of the Intes- The Antigen System of Paramecium aurelia-G. H. BEALE tine of Some Mammalian SpeciesThe Chromosome Cytology of the Ascites HARALD MOE Tumors of Rats, with Special Reference Localization of Cholinesterases at Neuroto the Concept of the Stemline Cellmuscular Junctions-R. COUTEAUX SAJIRO MAKINO Evidence for a Redox Pump in the Active The Structure of the Golgi ApparatusTransport of Cations-E. J. CONWAY AND PRISCILLA ARTHURW. POLLISTER AUTHOR INDEX-SUB JECT INDEX F. POLLISTER An Analysis of the Process of Fertilization Volume 5 and Activation of the Egg-A. MONROY Histochemistry with Labeled AntibodyThe Role of the Electron Microscope in ALBERT H. COONS Virus Research-ROBLEY C. WILLIAMS

...

CONTENTS O F P R E V I O U S VOLUhlBS

The Histochemistry of PolysaccharidesARTHURJ. HALE The Dynamic Cytology of the Thyroid Gland-J. GROSS Recent Histochemical Results of Studies on Embryos of Some Birds and Mammals-Jho BORGHESE Carbohydrate Metabolism and Embryonic Determination-R. J. O’CONNOR Enzymatic and Metabolic Studies on Isolated Nuclei-G. SIEBERT AND R. M. S. SMELLIE Recent Approaches to the Cytochemical Study of Mammalian Tissues-GEORGE EDWARDL. KUFF,AND H. HOGEBOOM, WALTERC. SCHNEIDER The Kinetics of the Penetration of Nonelectrolytes into the Mammalian ErythTOCyte-FREDA BOWER AUTHOR INDEX-SUB

JECT INDEX

CUMULATIVE SUBJECT INDEX

(VOLUMES 1-5) Volume 7 Some Biological Aspects of Experimental Radiology: A Historical Review-F. G. SPEAR The Effect of Carcinogens, Hormones, and Vitamins on Organ CuhreS-ILSE LASNITZKI Recent Advances in the Study of the Kinetochore-A. LIMA-DE-FARIA Autoradiographic Studies with S35-Sulfate D. D. DZIEWIATKOWSKI The Structure of the Mammalian SperW, F A W C E ~ matozoon-DoN The Lymphocyte-0. A. TROWELL The Structure and Innervation of Lamellibranch Muscle-J. BOWDEN Hypothalamo-neurohypophysial Neurosecretion-J. C. SLOPER Cell Contact-PAUL WEISS The Ergastoplasm : Its History, Ultrastruca r e , and Biochemistry-FRANCOISE HAGUENAU Anatomy of Kidney Tubules-JoHANNEs RHODIN Structure and Innervation of the Inner

Xlll

Ear

Sensory Epithelia-HANS ENGWERSKLL The Isolation of Living Cells from Animal Tissues-L. M. J. RINALDINI STROM AND JAN

AUTHOR INDEX-SUBJECT

INDEX

Volume 8 The Structure of Cytoplasm-CHARLES OBERLING. Wall Organization in Plant Cells-R. D. PRESTON Submicroscopic Morphology of the Synapse-EDuARDo DE ROBERTIS The Cell Surface of Paramecium-C. F. EHRETAND E. L. POWERS The Mammalian Reticulocyte-LEAH MIRIAM LOWENSTEIN The Physiology of Chromatophores-MILTON FINGERMAN The Fibrous Components of Connective Tissue with Special Reference to the Elastic Fiber-DAVID A. HALL Experimental Heterotopic OssificationJ. B. BRIDGES A Survey of Metabolic Studies on Isolated Mammalian Nuclei-D. B. ROODYN Trace Elements in Cellular FunctionBERTL. VALLEE AND FREDERIC L. HOCH Osmotic Properties of Living C e l l s D. A. T. DICK Sodium and Potassium Movements in Nerve, Muscle, and Red Cells-I. M. GLYNN Pinocytosis-H. HOLTER AUTHOR INDEX-SUB

JECT INDEX

Volume 9 The Influence of Cultural Conditions on Bacterial Cytology-J. F. WILKINSON AND J. P. DUGUID Organizational Patterns within Chromosomes-BERWIND P. KAUFMANN, HELEN R. MCDONALD GAY, AND MARGARET Enzymic Processes in Ceh-JAY BOYD BEST The Adhesion of CellS-LEONARD WEISS Physiological and Pathological Changes

XIV

CONTENTS O F PREVIOUS VOLUMES

in Mitochondrial Morphology-CH. ROUILLER The Study of Drug Effects at the Cytological Level-G. B. WILSON Histochemistry of Lipids in OogenesisVISHWANATH Cyto-Embryology of Echinoderms and Amphibia-KATsuMA DAN The Cytochemistry of Non-Enzyme Proteins-RONALD R. COWDEN

Histochemistry of Ossification-RoMULO L. CABRINI Cinematography, Indispensable Tool for Cytology-C. M. POMERAT AUTHOR INDEX-SUBJECT

INDEX

Volume 12

Sex Chromatin and Human Chromosomes JOHN L. HAMERTON Chromosomal Evolution in Cell PopulaAUTHOR INDEX-SUBJECT INDEX tions-T. C. Hsu Chromosome Structure with Special ReferVolume 10 ence to the Role of Metal Ions-DALE M. STEFFENSEN The Chemistry of Schiff's Reagent-FREDElectron Microscopy of Human White ERICK H. KASTEN Blood Cells and Their Stem CellsSpontaneous and Chemically Induced BESSISAND JEAN-PAUL THIERY Chromosome Breaks-ARuN KUMAR MARCEL In Vivo Implantation as a Technique in SHARMA AND ARCHANASHARMA Skeletal Biology-WILLIAM J. L. FELTS The Ultrastructure of the Nucleus and The Nature and Stability of Nerve Myelin Nucleocytoplasmic Relations-SAUL J. B. FINEAN WISCHNITZER The Mechanics and Mechanism of Cleav- Fertilization of Mammalian Eggs in Vitro C. R. AUSTIN age-LswIs WOLPERT The Growth of the Liver with Special Physiology of Fertilization in Fish EggsTOKI-oYAMAMOTO Reference to Mammals-F. DOLJANSKI Cytological Studies on the Affinity of the AUTHOR INDEX-SUB JECT INDEX Carcinogenic Azo Dyes for Cytoplasmic Volume 13 Components-YosHIMI NAGATANI Epidermal Cells in Culture-A. GEDEON The Coding Hypothesis-MARTYNAs Y t A s Chromosome Reproduction-J. HERBERT MATOLTSY AUTHOR INDEX-SUB

JECT INDEX

TAYLOR

Sequential Gene Action, Protein Synthesis, and Cellular Differentiation-REED A. (VOLUMES 1-9) FLICKINGER Volume 11 The Composition of the Mitochondrial Membrane in Relation to Its Structure Electron Microscopic Analysis of the Seand Function-ERIC G. BALL AND cretion Mechanism-K. KUROSUMI CLIFFED. JOEL The Fine Structure of Insect Sense Organs Pathways of Metabolism in Nucleate and ELEANORH. SLIFER Cytology of the Developing E Y ~ A L F R E D Anucleate Erythrocytes-H. A. SCHWEIGER J. COULOMBRE The Photoreceptor Structures-J. J. WOL- Some Recent Developments in the Field of Alkali Cation Transport-W. WILK EN BRANDT Use of Inhibiting Agents in Studies on Fertilization Mechanisms-CHARLES B. Chromosome Aberrations Induced by Ionizing Radiations-H. J. EVANS METZ The Growth-Duplication Cycle of the Cell Cytochemistry of Protozoa, with Particular Reference to the Golgi Apparatus D. M. PRESCOTT CUMULATIVE SUBJECT INDEX

CONTENTS O F PREVIOUS V O L I J M E S

xv

and the Mitochondria-VrsHwA NATH Regeneration of Mammalian LiverNANCYL. R. BUCHER G. P. DUTTA Collagen Formation and Fibrogenesis Cell Renewal-FELIX BERTALANFFY AND with Special Reference to the Role of CHOSENLAU Ascorbic Acid-BERNARD S. GOULD AUTHOR INDEX-SUB-JECT INDEX The Behavior of Mast Cells in AnaphyVolume 14 laxis-IVAN MOTA Inhibition of Cell Division: A Critical Lipid Absorption-ROBERT M. WOTTON and Experimental Analysis-SEYMOUR AUTHOR INDEX-SUBJECT INDEX GELFANT Electron Microscopy of Plant Protoplasm Volume 16 R. BUVAT Ribosomal Functions Related to Protein Cytophysiology and Cytochemistry of the Synthesis-TORE HULTIN Organ of Corti: A Cytochemical The- Physiology and Cytology of Chloroplast ory of Hearing-J. A. VINNIKOV AND Formation and “Loss” in EuglenaL. K. TITOVA M. GRENSON Connective Tissue and Serum ProteinsCell Structures and Their Significance for R. E. MANCINI Ameboid Movement-K. E. WOHLThe Biology and Chemistry of the Cell FARTH-BOTTERMANN Walls of Higher Plants, Algae, and Microbeam and Partial Cell IrradiationFungi-D. H. NORTHCOTE C. L. SMITH Development of Drug Resistance by Nuclear-Cytoplasmic Interaction with IonStaphylococci in Vitro and in Viuoizing Radiation-M. A. LESSLER MARYBARBER In V i m Studies of Myelinated Nerve Cytological and Cytochemical Effects of Fibers-CARL CASKEY SPEIDEL Agents Implicated in Various PathologRespiratory Tissue : Structure, Histophysiical Conditions: The Effect of Viruses ology, Cytodynamics. Part I. Review and of Cigarette Smoke on the Cell and and Basic Cytomorphology-FELIX D. Its Nucleic Acid-CEcIm LEUCHTENBERTALANFFY BERGER AND RUDOLF LEUCHTENBERGER AUTHOR INDEX-SUB JECT INDEX The Tissue Mast Wall-DOUGLAS E. SMITH Volume 17 AND

AUTHOR INDEX-SUB

JECT INDEX

The Growth of Plant Cell Walls-K. WILSON The Nature of Lampbrush Chromosomes Reproduction and Heredity in Trypanosomes: A Critical Review Dealing H. G. CALLAN Mainly with the African Species in the The Intracellular Transfer of Genetic InMammalian Host-P. J. WALKER formation-J. L. SIRLIN Mechanisms of Gametic Approach in The Blood Platelet: Electron Microscopic Studies-J. F. DAVID-FERREIRA Plants-LEONARD MACHLISAND ERIKA The Histochemistry of MucopolysacchaRAWITSCHER-KUNKEL rides-ROBERT C. CURRAN The Cellular Basis of Morphogenesis and Sea Urchin Development-T. GUSTAF- Respiratory Tissue Structure, Histophysiology, Cytodynamics. Part 11. New ApSON AND L. WOLPERT proaches and Interpretations-FELIX D. Plant Tissue Culture in Relation to DeBERTALANFFY velopmental Cytology--CARL R. PARTANEN The Cells of the Adenohypophysis and Volume 15

xvi Their Functional HERLANT AUTHOR INDEX-SUB

CONTENTS OF PREVIOUS VOLUMES

Significance-MARC JECT INDEX

Phosphorus Metabolism in Plants-K. ROWAN AUTHOR INDEX-SUB

S.

JECT INDEX

Volume 20

Volume 18

S. BREATH- The Chemical Organization of the Plasma Membrane of Animal Cells-A. H. MADDY The Structure of the Mammalian EggSubunits of Chloroplast Structure and ROBERT HADEK Quantum Conversion in Photosynthesis Cytoplasmic Inclusions in OogenesisRODERICB. PARK M. D. L. SRIVASTAVA Control of Chloroplast Structure by Light The Classification and Partial Tabulation LFSTI-R PACKI‘R A N D PAUL-ANDRI? SIEof Enzyme Studies on Subcellular Frac-

The Cell of Langerhans-A. NACH

tions Isolated by Differential Centrifuging-D. B. ROODYN Histochemical Localization of Enzyme Activities by Substrate Film Methods: Ribonucleases, Deoxyribonucleases, Proteases, Amylase, and HyaluronidaseR. DAOUST Cytoplasmic Deoxyribonucleic AcidP. B. GAHANAND J. CHAYEN Malignant Transformation of Cells in Vitro-KATHERINE K. SANFORD Deuterium Isotope Effects in CytologyE. FLAUMENHAFT, S. BOSE,H. L. CRESPI, AND J. J. KATZ The Use of Heavy Metal Salts as Electron Stains-C. RICHARD ZOBELAND MICHAEL BEER AUTHOR INDEX-SUBJECT

INDEX

Volume 19

GENTHALER

The Role of Potassium and Sodium Ions as Studied in Mammalian Brain-H. HILLMAN Triggering of Ovulation by Coitus in the Rat-CLAUDE ARON, GITTAASCH, AND JACQUELINE Roos Cytology and Cytophysiology of NonMelanophore Pigment Cells-JOSEPH T. BAGNARA The Fine Structure and Histochemistry of Prostatic Glands in Relation to Sex Hormones-Davm BRANDES Cerebellar Enzymology-LucrE ARVY AUTHOR INDEX-SUB SECT INDEX Volume 21 Histochemistry of Lysosomes-P. B GAHAN Physiological Clocks-R. L. BRAHMACHARY

“Metabolic” DNA : A Cytochemical Study H. ROELS The Significance of the Sex ChromatinMURRAY L. BARR Some Functions of the Nucleus-J. M. MITCHISON Synaptic Morphology on the Normal and Degenerating Nervous System-E. G. GRAYAND R. W. GUILLERY Neurosecretion-W. BARGMANN Some Aspects of Muscle RegenerationE. H. BETZ,H. FIRKET,AND REZNIK The Gibberellins as Hormones-P. W. BRIAN Phototaxis in Plants-WOLFGANG HAUPT

Ciliary Movement and Coordination in CihteS-BELA PARDUCA Electromyography : Its Structural and Neural Basis-JOHN V. BASMAJIAN Cytochemical Studies with Acridine Orange and the Influence of Dye Contaminants in the Staining Nucleic Acids FREDERICK H. KASTEN Experimental Cytology of the Shoot Apical Cells during Vegetative Growth and Flowering-A. NOUGAREDE Nature and Origin of Perisynaptic Cells of the Motor End Plate-T. R. SHANTHAVEERAPPA AND G. H. BOURNE AUTHOR INDEX-SUB

JECT INDEX

C O N T E N T S OF PREVIOUS VOLUMES

xvii

Volume 22

Volume 24

Current Techniques in Biomedical Electron Microscopy-SAUL WISCHNITZER The Cellular Morphology of Tissue Repair-R. M. H . MCMINN Structural Organization and Embryonic Differentiation-GA JANAN V. SHERBET AND M. S. LAKSHMI The Dynamism of Cell Division during Early Cleavage Stages of the EggN . FAUTREZ-FIRLEFYN AND J. FAUTREZ Lymphopoiesis in the Thymus and Other Tissues: Functional Implications-N. B. EVERETTAND RUTH w. TYLER(CAF-

Synchronous Cell DifferentiationGEORGEM. PADILLAA N D IVAN L. CAMERON Mast Cells in the Nervous SystemYNGVEOLSON Developmental Phases in Intermitosis and the Preparation for Mitosis of Mammalian Cells in VitYO-BLAGOJE A. NEJKOVIC Antimitotic Substances-Guy DEYSSON The Form and Function of the Sieve Tube: A Problem in ReconciliationP. E. WEATHERLEY AND R. P. C. JOHN-

FREY)

SON

Analysis of Antibody Staining Patterns Structure and Organization of the MyoObtained with Striated Myofibrils in neural Junction-C. COERS Fluorescence Microscopy and Electron The Ecdysial Glands of ArthropodsMicroscopy-FRANK A. PEPE WILLIAM S. HERMAN Cytology of Intestinal Epithelial CellsCytokinins in Plants-B. I. SAHAISRIVASPETERG. TONER TAVA Liquid Junction Potentials and Their AUTHOR INDEX-SUB JECT INDEX Effects on Potential Measurements in CUMULATIVE SUBJECT INDEX Biology Systems-P. C. CALDWELL (VOLUMES 1-2 1 ) AUTHOR INDEX-SUBJECT INDEX Volume 23

Volume 25

Transformationlike Phenomena in Somatic Cells-J. M. OLENOV Recent Developments in the Theory of Control and Regulation of Cellular Processes-ROBERT ROSEN Contractile Properties of Protein Threads from Sea Urchin Eggs in Relation to Cell Division-HIKoIcHI SAKAI Electron Microscopic Morphology of Oogenesis-ARNE N@RREVANG Dynamic Aspects of Phospholipids during Protein Secretion-LOWELL E. HOKIN The Golgi Apparatus: Structure and Function-H. W. BEAMSAND R. G. KESSEL The Chromosomal Basis of Sex Determination-KENNETH R. LEWIS AND BERNARD JOHN AUTHOR INDEX-SUB

JECT INDEX

Cytoplasmic Control over the Nuclear Events of Cell Reproduction-NoET. DE TERRA Coordination o f the Rhythm o f Beat in Some Ciliary Systems-M. A. SLEIGH The Significance o f the Structural and Functional Similarities of Bacteria and Mitochondria-SYLVAN NASS The Effects of Steroid Hormones o n Macrophage Activity-B. VERNONROBERTS

The Fine Structure o f Malaria Parasites ~ ~ A K A. I A RUDZINSKA The Growth o f Liver Parenchymal Nuclei and Its Endocrine Regulation--RITA CARRIERE Strandedness o f Chromosomes-SHErDoN WOLFF

xviii

CONTENTS OF PREVIOUS VOLUMES

Isozymes : Classification, Frequency, and S i g n i f i c a n c e - C ~ ~ R. ~ ~SHAW ~s

Protein Metabolism in Nerve Cells-B. DROZ

The Enzymes of the Embryonic Nephron LUCIEARVY

AUTHOR INDEX-SUBJECT

Freeze-Etching-HANs INDEX

A New Model for the Living Cell: A Summary of the Theory and Recent Experimental Evidence in Its Support GILBERT N. LING Department of Molecular Biology, Dii'ision of Neurology, Pennsylvania Hospilal, Philadelphia, Pennsylvania

............................ ............................ ............................ B. The Energy Requirement of the Necessary Pumps . . . . . C. The Physical State of Water in the Living Cell D. Is the Cell Membrane a Universal Rate-Limitin to the Intracellular-Ex.trarellular Traffic of Water and All Solutes? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. An Interesting Clue in the Search for a Better Model of the Living Cell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111. The Association-Induction Hypothesis . . . . . . . . . . . . . . . . . . A. The Molecular Mechanism for Solute Distribution in Living Cells: Theoretical Aspects . . . . . . . . . . . . . . . . . . B. The Molecular Mechanism for Solute Distribution in Living Cells: Experimental Evidence . . . . . . . . . . . . . . . . C. Answers to Fundamental Criticisms of Ionic Adsorption in Living Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D. Molecular Mechanisms in the Integrative Function of Protoplasm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . ..........

1 2 2

5 7

7 10 11 12

20

31

36 58

Introduction It is generally acknowledged that living organisms, although vastly complex, can eventually be understood in terms of basic principles derived from studies of the simple inanimate world. The primary function of a biologist is thus not so much to deduce basic principles-that belongs to physics-but to understand how these principles apply in a unique complex system. The most logical time to begin fundamental biological research, therefore, would be when men have achieved complete understanding of the physical world. Then one could be certain that the foundations of his reasoning would be sound. Since this has not been the case-investigations of living matter are as old as the study of the inanimate world-the student of biology must be doubly cautious in accepting assumptions presented as facts. Thus, for example, in the writing of textbooks, what could be is often transformed into what is. For this reason the student must constantly reexamine the major premises no matter how popular or venerable, using both up-to-date knowledge of the physical sciences and advanced tech1

2

GILBERT N. LING

nology that was not available to investigators of previous days. The present review begins with such a reevaluation of the basic assumptions of an old and venerable concept, the membrane theory. After discussing some of the difficulties in this concept, it goes on to summarize both theoretical aspects of and experimental evidence for an alternative model of the living cell.

I. The Membrane Theory A.

HISTORY

About 90 years ago, H. Pfeffer, struck by the similarity between the osmotic behavior of living cells and that of an aqueous solution enclosed in a semipermeable membrane, propounded the membrane theory (Pfeffer, 1921). According to this view, the outermost layer of any living cell consists of a membrane which is a universal rate-limiting barrier to the traffic of water and all solutes between the cell interior and the external environment. A second implicit assumption was that there is no significant interaction between the cell proteins which constitute 15-25% of the cell weight and the cell water which makes u p almost all the rest (75-85%) (Ling, 1962). Thus the water in the cell was postulated to be essentially the same as in any dilute salt solution. O n the basis of these assumptions, the problems of water and solute distribution as well as the maintenance of cell volume can be expressed in terms of one all-encompassing parameter, the meinhrzlm pevmeubility. Thus the degree of cell swelling in a medium containing a particular solute reflects the permeability or impermeability of the cell membrane to this solute (De Vries, 1885; Hamburger, 1889). In the following six decades much effort was devoted to discovering the properties of this cell membrane that allowed it to determine not only the rates of entry and exit of water and solutes into and out of the cell but also their steady levels within the cell. At the turn of the century, Overton, using osmotic methods such as cell swelling, studied the relative permeability of many solutes (Overton, 1899, 1907). A parallel between the permeability of many nonelectrolytes and their oil/water distribution coefficients led him to suggest that the cell membrane consists of a continuous lipid layer. He further suggested that the transport of nonlipid-soluble solutes such as sugars, amino acids, and ions was promoted by “adenoid” or secretory activity. In 1933, Collander and Batlund confirmed Overton’s findings by measuring solute permeation into plant cell sap. They found, however, that the high permeability of water was not compatible with its low oil/watrr distribution coefficient, leading them to postulate the existence of small aqueous channels in the lipid membrane that permit water (and other small molecules) to pass through (mosaic membrane theory) (Collander and Barluncl, 1933). The discovery, also in the 1930’s, that the surface tension of cells is far less than that at an oil/water interface led to the suggestion of a

A NEW MODEL FOR THE LIVING CELL

3

protein covering for the lipid layer (Cole, 1932; Davson and Danielli, 1952). Ruhland and Hoffman (1925) first suggested that the cell membrane may be sievelike, selectively admitting some ions but not others (Ruhland and Hoffman, 1925; see also Mond and Netter, 1930). The sieve theory reached the height of its development in the version presented in 1941 (Boyle and Conway, 1 9 4 1 ) : The critical pore size was postulated to be such that the permeabilities of both cations and anions could be explained. Small cations like K + ion and H+ ion were considered permeable, while the large cations, Na+, C a + + , and M g + + were not. Small anions like C1- and OH- were permeable but not larger anions like ATP, CrP, and hexose phosphates. Not only did this theory present a unified interpretation for the distribution of all types of ions, it also provided the molecular basis essential to Bernstein’s membrane theory of cellular electric potential. If one considered the small pores of Boyle and Conway the same as those postulated by Collander and Barlund one could envisage an internally consistent interpretation for the permeability of nearly all solutes and water. This state of apparent harmony between the membrane theory and the experimental facts, however, did not long survive the publication of Boyle and Conway’s theory ( 1941) . About the same time Heppel (1940) and Steinbach (1940) demonstrated that muscle cells are in fact permeable to Na ion. This was construed by most as a refutation of Boyle and Conway’s specific theory. In order to account for the low steady level of Na-+-ion in the cell in spite of its constant inward diffusion, the remedial N a + -ion pump hypothesis was advanced (Dean, 1941 ; Krogh, 1946; Hodgkin, 1951). T h e reviewer feels, however, that the findings of Heppel and of Steinbach have greater significance than is usually recognized: ( 1 ) Thus, increasing the Na+-ion concentration of the external medium leads to cell shrinkage; therefore, this ion satisfies the criterion on the basis of the membrane theory, of an impermeant solute. T h e demonstration that it is permeable shows such swelling and shrinking effects cannot be used as a gauge of permeability. This invalidates the basic assumption involved in the widespread use of osmotic behavior as a measure of membrane permeability. ( 2 ) If the N a + ion enters the cell via the aqueous channels (that the bulk of Na-+ ion entering muscle cells does not show competition supports this; see Fig. 21; Ling and Ochsenfeld, 1965), these channels must also be wide enough to permit the passage of nonelectrolytes like glycerol, erythritol, and so forth. Such compounds are far ICSS soluble in lipids than in water. Thus they cannot be assumed to enter the cell via the lipid layer rather than the aqueous channels. If they do enter via the aqueous channels, the observed linear correlation between the oil/water distribution coefficient and the permeability of nonelectrolytes (Overton, 1907 : Collander, 1940) cannot be based on permeation through a lipid membrane. ( 3 ) According to the theory of Boyle and Conway, hydrated N a 1 , Ca-1 +-, Mg-1 -1 ions, as well

4

GILBERT N. L I N G

as ATP, CrP, and hexose phosphates are all impermeable because they are larger than the pores. However, N a + , C a + + , and Mg+ + ions have all been shown to be permeable (Ling, 1962). What then prevents ATP, CrP, and hexose phosphates from diffusing out? Since according to the membrane theory these are the impermeant anions in a Donnan equilibrium, their outward diffusion also leads to a collapse of the resting potential (Boyle and Conway, 1941). These are some of the problems raised by the demonstration of Na+-ion permeability; they cannot be remedied merely by the postulation of a N a + pump. Even more difficulty is raised by the demonstration that cells are permeable to much larger molecules than Na+ ions: There is increasing evidence that both proteins and other macromoIecuIes can enter and exit from cells (Avery el al., 1944; Dawson, 1966; McLaren et al., 1960; Zierler, 1958; Ryser, 1968). Thus, Dawson (1966) has shown that enzymes diffuse from isolated intact chicken muscles. Ryser studied the entry of 1’31-labeled serum albumin and many other proteins including ply-L-lysine and Poly-D-lySine showing that all these molecules enter cells with rates generally proportional to their molecular weights1 Thus the membrane theory, in its present state, is beset with internal inconsistencies. There are, however, more fundamental questions that must be raised concerning the general validity of the membrane theory: (1) Does the resting cell produce energy at a rate sufficient to provide for all the necessary “pumps” ? ( 2 ) Is there truly no significant interaction between cell proteins and cell water ? ( 3 ) Is the cell membrane a universal rate-limiting barrier to the traffic of water and other solutes between the cell interior and its external medium? By making use of technology not available until very recently, critical experi1 One mechanism offered for the entry of proteins into cells is based on the concept of pinocytosis; i.e., the cell actually phagocytizes droplets of its external medium containing macromolecules (Bennett, 1956; Holter and Holtzer, 1959). There are a number of difficulties associated with this concept: (1) According to current concepts of pinocytosis, the macromolecules phagocytized, although within the cells, are actually outside of the cytoplasm because there is a continuous plasma membrane surrounding the vesicle. Unless there is entry of the substance into the cell from this vesicle, however, it is difficult to understand how macromolecules can have any physiological action on the cell, such as, for example, the effect of D N A on capsule formation in Pneumococrus (Avery et a]., 1944). If the substances enter the cells by a “melting away” of the plasma membrane, the consequence of pinocytosis would be the creation of a nondiscriminatory route for the entry of all solutes into the cell. If this is the case, it is difficult to explain why poly-D-lysine enters more rapidly than poly-1.-lysine (Ryser, 1968) into living cells, as well as the correlation between the rate of entry of nonelectrolytes and their oil-water partition coefficients. ( 3 ) Metabolic poisons such as iodoacetate, NaF, 2,4-dinitrophenol, and cyanide, which inhibit phagocytosis, do not inhibit the entry of P”-labeled serum albumin into living cells (Ryser, 1968). ( 5 ) Last, in what way can the cell prevent continued pinocytosis from ingesting and depositing an ever-increasing amount of external proteins in the cell? Must we postulate protein pumps?

A NEW MODEL FOR THE LIVING CELL

5

ments have been carried out that have provided answers to these questions. One may add that, had it been technically feasible, some of these experiments should have been carried out shortly after Pfeffer presented the membrane theory some 90 years ago. B. THEENERGYREQUIREMENT OF

THE

NECESSARY PUMPS

Shortly after the Na+ pump was postulated, no less than four sets of experiments (excluding the reviewer's) were published, all comparing the minimum energy need of the Na+ pump in frog sartorius muscles with the maximum energy available (Conway, 1946; Levi and Ussing, 1948; Harris and Burn, 1949; Keynes and Maisel, 1945). Conway (1946), as well as Levi and Ussing, regarded the energy consumption as being too high for the cell to cope with. However, Conway's Na+-ion efflux data were indirectly deduced and might thus be subject to question. The data of Levi and Ussing (1948), Harris and Burn (1949), and Keynes and Maisel (1945) are more or less consistent with one another and show a minimum requirement of about 20% of the total energy output. These figures were derived on the assumption that all the energy from glucose oxidation is converted to a form suitable for consumption by the pump at 100% efficiency and that the pump itself is also 100% efficient. Since neither assumption is likely to be true, these figures themselves indicate a failing in the pump concept (Ling, 1955). Because of the crucial importance of the issue, I reexamined the energy problem in two ways: (1) The energy rryuiremeiit of the Naf pump at 0°C. itz mriscles poisoned with iodoaretate atid pnre nitrogelz. Arrest of oxidation and glycolysis does not significantly alter the steady level of Na+ and K+ ion in frog muscle cells (0°C.) for as long as 7 hours (Ling, 1962, p. 200). During this time the efflux of N a + ion continues at a rate not slower than that of the unpoisoned control muscle (Ling, 1962, p. 198). Without oxidation and glycolysis the energy sources of the muscle cells are limited to its store of ATP and creatine phosphate. Comparing the maximum energy available from the hydrolysis of these compounds with the minimum energy need calculated for pumping on the basis of the measured resting potential, the intracellular Na+-ion concentration, and the Na+-ion efflux rate,2 I reached the conclusion that the minimum energy need is 2 I have been asked a number of times whether the Na+-ion efflux value used in the computation might be an overestimation. Thus for normal muscles at O"C, I gave an efflux rate of 1.76 x 10-11 mole/cm.~/sec. (in contrast to the value of Harris of 4.7 X mole/cm.2/sec., 123). The rate of efflux from poisoned muscle was higher (3.9-8.73 X 10-11 mole/cm.2/sec.). Harris' value was based on an efflux curve not significantly different from mine (compare Harris, 1950, Fig. 2 with Fig. 11.29 in reference Ling, 1962). However, he discounted

6

GILBERT N. LING

1500-3500C/o of the maximum available energy, again assuming 100u/, efficiency (Ling, 1962, p. 211). ( 2 ) T h e energy regiiiveineiil of the Nu,+, Ca++, and Mg++ pumps in re.rtzizg frog nzusrde cells. At the time N a + ion was demonstrated to be permeable, it was considered an exception and the N a + pump was postulated. However, it was not long before it was discovered that no two ions distribute themselves between the inside and the outside of the cell with the same Donnan ratio, and all are permeable (Ling, 1 9 5 5 ) . To explain this phenomenon on the basis of the membrane theory, more pumps must be postulated with, however, the same maximum energy source (already overstretched for the N a + pump alone). Using the data available in the literature I calculated that the N a + , C a + + , and M g + + pumps alone would consume no less than 350% of the total maximum available energy of a resting frog muscle (Ling, 1965b). To this must be added pumps to maintain the levels of all the other solutes (HC0,-, C k , amino acids, and so forth) which are not distributed between the cell and its surrounding medium according to thermodynamic equilibrium.

the early fast fraction (15-20 minutes) as being attributable to efflux from the extracellular space. O u r subsequent study of Na+-ion efflux from isolated single muscle fibers, however, shows that this fraction cannot be in the extracellular space. Thus it takes less than 1 second to wash away the adhering solution in this preparation-but these fibers still possess ii similar fast fraction (see Fig. 11.4, Ling, 1962) (Curves obtained by Horowicz and Hodgkin (Horowicz and Hodgkin, 1957) from single muscle fibers were apparently exponential; however. the setup used by these authors did not allow points before the first 10 minutes of washing; after that they could obtain readings at only about 10-minute intervals. Irnder such conditions the curvature is lost). In all the estimations of energy balance quoted above, it had been taken for granted that the slow flat portion of the efflux curve represents the N a + ion pump. However, the evidence quoted in Section 111, B,4 gives us considerable reason to equate the slow fraction in the Na+-ion efflux with exrhange of the adsorbed fraction. If this is the case (for additional evidence, see Ling, 1962), the fast fraction actually represents the rate of efflux of the free N a + ion into the environment and should be used for the rate of pumping (Ling, 1962, Chapt. 1 1 ; Ling, 1966a). In conclusion, 1 might point out that the method I used to derive the rate of efflux was chosen to give a conservative estimate. A small bundle of muscle fibers was dipped for a time interval (about 3 minutes) in an isotope-labeled solution. The tissue was then quickly mounted on the washout apparatus and an efflux curve obtained for the succeeding 100-200 minutes by which time the curve had become exponential. This exponential part of the curve was then extrapolated to the ordinate to give an estimate (in actuality too small) of the amount of Na+ ion that entered the cell during the 3-minute incubation. Since there was no change in the total Na-+-ion concentration during this time, the influx rate and efflux rate must have been equal. The rate of pumping was then derived from this figure.

A NEW MODEL FOR THE LIVING CELL

C. THEPHYSICAL STATEO F WATERI N

THE

7

LIVINGCELL

Evidence has been gathering at a rapid rate which shows that the cell water is in a different state than the water in a dilute salt solution. Since this subject will be discussed at length in a following section, we shall only point out here that there are two lines of evidence pointing to the above conclusion: ( 1 ) the abnormal freezing pattern of water in the living cell (Chambers and Hale, 1932; Rapatz and Luyet, 1958; Ling, 1 9 6 7 ~ )when compared to that of normal water or the water in a dilute salt solution (Ling, 1966b) ; and ( 2 ) T h e abnormal nuclear magnetic resonance ( N M R ) spectra (Chapman and McLaughlan, 1967; Fritz and Swift, 1967) obtained from the water in living cells.

D. Is

THE

CELL MEMHRANE A UNIVERSAL RATE-LIMITING BARRIER TO

THE

INTRACELLULAR-EXTRACELLULAR TRAFFIC O F WATER A N D ALL SOLUTES? Recently the reviewer has presented a technique, influx profile analysis, that provides a means of determining the rate-limiting step in the traffic of water or solutes between the cell and its environment (Ling, 1966a; see Fig. 1 ) . In brief, the fractional uptake of a labeled material t seconds after the introduction of a cell into a solution containing the isotope is plotted against the square root of t. The profile has specific features depending on the rate-limiting step. Thus, if the rate-limiting step is in the cell membrane, the curve is sigmoid in shape (Fig. l A ) . O n the other hand, if the solute diffuses with a more-or-less uniform rate throughout the entire cell including the cell membrane (bulk-phase-limited diffusion), the initial part of the curve is essentially a straight line (Fig. I B ) . This technique can be most usefully applied to single cells. Figure 2 shows the influx profile for the entry of tritium hydroxide-labeled water into a single frog ovarian egg (Ling et a/., 1967). The solid line, passing through the points has been theoretically calculated to represent bulk-phase-limited diffusion. This shows that the cell membrane is no more resistant to water movement than the cytoplasm. The overall diffusion coefficient ranges from one-half to onethird of the diffusion coefficient of tritium hydroxide in a 0.1 N salt solution. This set of experiments disproves one of the basic tenets of the membrane theory with respect to water. However, water is by no means the only substance whose movement is not limited by the membrane. Fenichel and Horowitz have demonstrated that the efflux of many nonelectrolytes from frog muscles is also bulk-phase limited (Fig. 3 ; Fenichel and Horowitz, 1963). This study included many of the same nonelectrolytes investigated by Overton and by Collander. Thus, it appears that the “membrane” permeability investigated by these authors was, in some cases a t least, the permeability of the bulk of the protoplasm. Summarizing, we can now state that (1) the resting cell does not command enough energy to operate all the pumps necessary in terms of the membrane

A

B

0.8

0.8

M t 0.6

0.6

0.4

OA

M,

0.2

0

I

2

3

I

4

Jt

2

3

0

I I

2

3

J7

FIG. 1. The time course of influx of a labeled substance into model systems with rate-limiting steps as indicated on each chart. The “influx profiles” are theoretically calmlated. The ordinate represents the uptake, M,, of the labeled material at time t as a fraction of the final amount of the material in the system ( M w ) . The abscissa represents the square root of t (Ling, 1966a, by permission of

Amah

of

the N e w YovR Arademy

of Srie?ice.r).

9

A N E W MODEL FOR T H E LIVING CELL

theory to maintain the observed assymetrical solute distribution ; ( 2 ) the cell water is not normal as postulated by the membrane theory, and ( 3 ) the cell membrane is not a universal rate-limiting barrier to the traffic of water and solutes between the cell and its environment. 105

-+

c _

08 /M, Mm

04

0

converted into an influx time course using the “inversion method.” T h e curve passing through the points is theoretically computed on the basis of simple bulk-phase-limited diffusion (Ling et ul,, 1967, by permission of The Journal of General Phyrirology.)

I 0

I

2000 t (seconds)

I

I

1

4000

FIG. 3. T h e time course of labeled thiourea efflux from a frog sartorius muscle. Efflux of C14-labeled thiourea was assayed by agitating muscles previously equilibrated with thiourea, in different portions of nonlabeled Ringer solution, the activity of which was then assayed. Curve A is theoretically calculated for memhrane-limited diffusion and does not fit the data. Curve B, which fits the data nearly perfectly, was calculated theoretically on the basis of bulk-phase-limited diffusion. C, is the concentration of labeled thiourea at time t in cells; C , is that at t equals 0 (Fenichel and Horowitz, 1767, by permission of Artu Physiologic-a Srundinauira) .

10

GILBERT N. LING

Taken together, this evidence is very strongly against the membrane pump theory. Therefore, we have little choice other than to seek a new model of the living cell both to interpret the vast amount of data already accumulated and to guide future research. 11. An Interesting Clue in the Search for a Better Model of the Living Cell

Figure 4 shows an electron micrograph by Starr and Williams of a flagellum from the Congo diphtheroid bacillus (Starr and Williams, 1952). The dry matter of such flagella is virtually pure protein (Weibull, 1960). Isolated flagella

FIG. 4. Electron micrographs exhibiting the helical fine structure of flagellar material from the Congo diphtheroid bacterium. The structure is that of a left-handed, triplestranded helix with a diameter of 19 nip and an axial periodicity of 50 m p 100,000 X (Starr and Williams, 1952, by permission of Journal of Bacleriology).

A NEW MODEL I:OR

THE LIVING C E L L

11

can be reversibly precipitated by ammonium sulfate and behave in many ways like homogeneous protein molecules. There is no membrane cover. T h e isolated flagellum apparently possesses neither ATPase nor any other enzyme. I t is directly connected to the protoplasm at a subcellular structure called the basement granule (Weibull, 1960). In spite of its structural simplicity, this protein-water system is capable of undergoing spiraling movement, thus providing the driving force for the mobility of the bacterium (Holwill and Burge, 1963). T h e bacterial flagellurn is an illuminating example of the fundamental capabilities of protoplasm, with control and energization occurring at the basement granule away from the body of flagellum itself (Astbury, 1951). Thus, this system illustrates the transmission of information and energy for long distances along a protein-water system. It is, according to the association-induction hypothesis, this fundamental ability of organized protein-water-ion systems to undergo reversible changes between metastable equilibrium states that distinguishes this and many other types of protoplasm from the inanimate world. T h e cell owes both its functional coherence and its discontinuity from the external environment not to a lipid membrane but to the unique properties of the protein-water system, just as the naked flagellum, a permanent organelle in an aqi~eousenvironment, is functionally coherent and discontinuous with its environment. 111.

The Association-I;iduction Hypothesis

T h e association-induction hypothesis considers the maintenance of the pattern of solute distribution to reflect the properties of the entire protoplasm (Ling, 1962, 1964b, 196ja,b, see also Butschli, 1894). It is well known that the water content of a living cell is more or less constant. It is also generally accepted that water distribution represents an equilibrium state, which means that the free energy of water within the cell is equal to the free energy of water outside the cell. Therefore, within a unit time interval, the number of water molecules entering the cells exactly equals the number of water molecules leaving the cell. To maintain this steady level of water, the cell does not expend energy. The association-induction hypothesis maintains that the steady levels of ull the solutes in the living cell also represent equilibrium states, or rather metastable equilibrium states. A metdstable equilibrium state is a true equilibrium state, only its maintenance is somewhat precarious much like the case of a narrow block of wood standing on its edge. In the following review, I shall deal specifically with K + and N a + ion, with

12

GILBERT N. LING

the understanding that the mechanisms involved in their distribution and control are basically all similar for other solutes.

A. THEMOLECULAR MECHANISM FOR SOLUTEDISTRIBUTION IN LIVINGCELLS:THEORETICAL ASPECTS According to the association-induction hypothesis (see also Fischer and Moore, 1908; Troschin, 1958), intracellular solutes exist in two states: (1) solution in the cell water and ( 2 ) adsorption onto cell proteins. Since the amount of solute in the first state depends on the state of the water in the cell, an important part of our discussion of solute distribution will be a consideration of the state of water in the living cell. Following this we will go on to consider the effect of the state of water on the distribution of solutes, and finaIIy we will deal with the specific adsorption sites for the solute. W e will first consider the theoretical aspects of these problems followed by a discussion of the experimental evidence in support of the model. 1. T h e Effect of t h e Protein on the State of Water in Living

Cells

Water molecules possess a strong permanent dipole moment (1.83 x lo-'* e.s.u.) as well as a high polarizability (1.44 x 1 0 - 2 4 cm.), hence a great propensity to form strong induced dipoles (Ling, 1962, p. 65). A simple calculation shows that the electrostatic interaction of water molecules with an electrical charge carried, for example, by an ion extends beyond the first layer of water molecules surrounding the ion to a number of additional layers. Polar compounds like titanium dioxide also interact with water. Experimentally, Harkins has shown that the heat of desorption of the first layer of water molecules from the surface of titanium dioxide is 6550 cal./mole higher than from quartz. For the second layer, it is 1380 cal./mole higher; for the third layer, 220 cal./mole higher; and for the fourth layer, 71 cal./mole higher (Harkins, 1945). Protein, similar to titanium dioxide, bears an abundance of polar groups. In muscle cells the average chain-to-chain distance between protein molecules is only 16.9 A., less than the thickness of seven layers of water molecules. All or nearly all of the water in a typical resting cell may thus be under the polarizing influence of the ionic and hydrogen-bonding groups of the proteins and exist as polarized multilayers (Ling, 1962, Chapt. 2, 1965a; 196613; 1 9 6 7 ~ ) .Further, this polarization must orient the water molecules in directions that are determined by the structure and orientation of the proteins. In such a system, the freedom of motion of the individual water molecules is more constrained than in normal water, this restriction being most prominent in rotational motion. The degree of restriction falls off gradually with the distance from the protein surfaces as illustrated in the diagram shown in Fig. 5, in which the length of the curved arrows indicates the degree of rotational freedom.

A NEW MODEL FOR T HE LIVING CELL

2.

13

The Enlropic Exclusio?z of Multiatomic or de Facto Multiatomic Solutes from Polarized Water

In an aqueous medium, N a + ion acquires at least one layer of water of hydration; i.e., water molecules that are strongly polarized under the influence of the electric charge of the ion. This hydrated ion behaves as a single unit, hence it is de farto multiatomic. Such a molecule possesses many modes of rotational mo-

FIG. 5 . Diagrammatic representation of the adsorption of water molecules as polarized multilayers o n proteins. O n entering such a system, the hydrated ion shown to the left suffers severe rotational restriction. A simple model of this effect is shown on the right where the restricted orientation of small nails in the field of horseshoe magnet filled with iron filings is shown.

tion. In fact, rotational entropy, which is a measure of the rotational freedom of the molecule, constitutes the major part of the entropy of such an ion in an aqueous medium. When a hydrated ion, such as Na+ ion, is introduced into the cell where the bulk of the water is polarized into multilayers, it suffers a restriction of its rotational movement in a manner analogous to the loss of freedom in the orientation of small nails (Fig. 5 ) introduced into the iron filing-filled space of a horseshoe magnet. The result of this rotational restriction is a lowered entropy. At equilibrium the distribution of a solute between a system containing oriented

14

GILBERT N. LING

water and a normal aqueous medium is determined by the difference in the standard free energy, of the solute in the two media. The AF", in turn, is the sum of an energy term3 and an entropy term. The energy does not differ very much between the two systems. A lowered entropy of the N a + ion in the cell water means a lowered AF" and thus a lower equilibrium concentration of N a + ion in the polarized water. 3 . The M o l e r z h v Merhanim of Ionic Adsoiptioiz

As far back as 1908, Fischer and Moore suggested that selective K+-ion accumulation might result from adsorption on cell colloids (Fischer and Moore, 1908). In 1951 and 1952, the reviewer suggested that the p- and y-carboxyl groups carried by the aspartic and glutamic acid side chains could offer anionic sites for the adsorption of K + as well as Na+ ion (Ling, 1951, 1952). The hydrated diameter of the K + ion is considerably smaller than that of the hydrated N a + ion. Following Coulomb's law, the electrostatic interaction of K + ion with the negatively charged carboxyl groups would be greater than that of larger hydrated N a + ion. By taking into account the profound reduction of dielectric constant in the immediate neighborhood of an ion (the dielectric saturatioi? phenomenon), a selectivity of K + ion over N a + ion of the order of 10 to 1 can be theoretically calculated. In support of this hypothesis, the reviewer drew an analogy between the newly developed ion exchange resins and the living cell. By introducing anionic groups and fixing them on a three-dimensional network, selective accumulation of K+ over N a + ion was achieved in the resin. In the years following, additional knowledge was gained in the field of ionexchange resin technology. It was demonstrated that such resins do not always selectively accumulate K + ion over N a + ion. Thus, resins bearing strongly acidic groups (low pK) prefer K + ion over N a + ion. However, for resins bearing weakly acidic groups (high p K ) , the reverse is the case (Bregman, 1953). This fact and the theoretical and experimental work of Eisenman, Rudin, and Casby on glass electrodes (Eisenman ef a/., 1957) led to a complete revision of the earlier model (Ling, 1960). It became apparent that the pK value primarily reflects the electron density of the acidic group. To put this concept in manipulatable form, the c-value was introduced. This parameter is rigorously defined elsewhere (Ling, 1962, p. 5 7 ). For simplicity, it may be mentioned that a high r-value (i.e., approximately - 1 A , ) corresponds to a high p K value (e.g., acetic acid, pK = 4.75). A low c-value (i.e., -5 A . ) on the other hand, corresponds to a low pK value (e.g., trichloroacetic acid, pK < 1.0). 3 More correctly this term should refer t o enthalpy or heat content H which is related to the energy U by the relation: H U PV, where P is the pressure and 1.' is the volume of the system. Since in a liquid system, volume changes are small and only changes of H and U are significant in our discussion, the more familiar rnevyy is used here.

+

15

A NEW M O D E L FOR 'THE LIVING CB1.L

With the r.-value defined, it becomes possible to calculate the total interaction energy between a specific cation (e.g., K + or N a + ion) and an oxyacid group (such as a carboxyl group) of a certain c-value, when the cation is separated from the oxyacid group by zero, one, two, or three water molecules (see Fig. 6 ) .

Configuration 0

Configuration I

Configuration

II

Configuration UI

FIG. 6. T h e linear inodel. l h e shaded cml: on the left i n each configuration represents the negatively charged oxygen atom of an oxyacid (e.g.. carboxyl) and the shaded circle o n the right represents its counter-action. Open circles represent water and the various letters denote distances used in the computations. Reprinted by permission of the publisher from: Gilbert N. Ling. "A Physical Theory of the Living State" (Waltham, Massachusetts: Blaisckll Publishing Company, 21 Division of Ginn and Company, 1962) p . 61.

From these results, the dissociation energy of different alkali-metal ions can be determined as a function of the r-value (as well as the polarizability) of the anionic group. Figure 7 shows the results of such a calculation. As the c-value increases, the order of preference of the anionic group for the five alkali-metal cations goes through 11 permutations. At the lowest c-value the sequential order is Cs > Rb > K > N a > Li while at the highest c-value the order is completely reversed. These theoretically calculated sequential order changes are similar to sequential order changes observed experimentally by Eisenman (Eisenman, 1961; Fig. 8 ) in the relative preference of glass electrodes of varying composition for the alkali-metal ions. One significant conclusion to be drawn from the results of these calculations

16

GILBERT N. LING

is that a small change of the c-value can significantly alter the relative preference of an acidic group for K+ over N a + ion. When the c-value change is large enough, the order of preference can actually be reversed. This point will be discussed again in Section 111, D,7.

FIG. 7. The computed dissociation energy of various cations as a function of the cvalue. A polarizability of 0.87 x 10-24 cm.3 has been assumed for the fixed anionic group. Reprinted by permission of the publisher, from Gilbert N. Ling, "A Physical Theory of the Living State (Waltham, Massachusetts: Blaisdell Publishing Company, A Division of Ginn and Company, 1962) p. 75.

4. The Equation f o r Solzdte Distribzition in Living Cells According to the Assoi-intion-lizdzictio?2 Hypothr r i ~ Figure 9 shows a diagram of a portion of a living cell in contact with its external medium. Within the cell ions exist in two states, free and adsorbed. The model cell shown possesses three types of protein sites which adsorb alkalimetal cations. Two of these types of sites prefer K+ ion over Na+ ion; one prefers Na+ ion over K + ion. Based on this model, the concentrations of the intracellular ions can be expressed by the following equations : a"a+Illlt+ and

"a+]:,+

"a+]:,+

"a+]"

(1)

17

A N E W MODEL FOR THE LIVING CELL

where [ N a + I i nand [ K + I i n are the total intracellular concentrations of N a + ion and K + ion, respectively, a is the percentage of water in the cell. [Na+Iint and [K+],,, are the concentrations of interstitial K + and N a + ion, respectively. [ N a + ] i d and [K+]:,, are adsorbed N a + and K + ion on the type I sites, "a+]:: and [K+]:', on the type I1 sites, and so forth. Intracellular and

kcal/mole

s

-

-6918

-

-4.612

-

-2306-

.k

0 II .li.-

a

2.306 -

i

4612 6.9121

4.612

2 306

0

- 2 306

AFhak=-RT InKhak= FAE'

FIG. 8. Ionic specificity in ionic glass electrode potentials at neutral pH. Each vertical row of data points corresponds to the observed selectivity properties of a particular material (Eisenman, 1961 ) .

adsorbed ion concentrations are in units of moles per kilogram of fresh cells. Interstitial ion, on the other hand, is in moles per liter of cell water. Putting Eqs. ( 1 ) and ( 2 ) in a more general form:

18

G I L B E R T N. L I N G

and

c N

IK’

111,

+

=a(K+J,,,,

IK’yl

(4)

I> 1 ~

,:I

Here [ N a t and [K+]>l;,refer to the concentrations of adsorbed N a + and K + ion on the Lth type of sites. In the case shown in the diagram of Fig. 9 , there are a total of three types of sites, therefore N z 3 .

Rti. 9. Dingrammatic illustrations of a living cell. Stiplrcl area represents space filled with water in polarized inultilayers.

a. The Egicafioiz f o r the liitevstitlai I o n . Let us first consider the distribution of interstitial ions. According to Henry’s law, the ratio of the Concentrations o f a solute distributed between two solvents at equilibrium with each other is a constant over a considerable concentration range (see also Troschin, 1958). Thus, in the case of the distribution of N a + ion between the cell water and the external aqueous medium, the ratio of internal to external N a + ion concentrations is a constant ysa called the distribution coefficient. Thus

where [ N a + ] ,,s is the external Na+ -ion concentration. Similarly

where [ K + is the external K+-ion concentration and ye is the equilibrium distribution coefficient of K + ion. Rearranging Eqs. ( 5 ) and (6) : I N a + I , l , t = ( j ~ , l [Na-‘I..

(7)

19

A NEW MODEL FOR THE LIVING CELL

and [K+I,rlt = qrc [K+],Y (8) According to these equations, a plot of the interstitial Na+- (or K+-) ion concentration as a function of the external Na+- (or K + - ) ion concentrations should yield a straight line having a slope equal to qNu (or q K ) .Further, there is no competition among ions in the interstitial water, i.e., the same concentration of Na+ ion is found in the cell water whether or not K + ion is also present. 6. T h e Equatioti for the Adsorbed I o m . If the concentration of the type I adsorption sites shown in Fig. 9 is [f ] I and i f each site can be occupied by one ion at a time, the total number ( J f Na-1- ions adsorbed on the type I sites can be described by the Langmuir adsorption isotherm (Langmuir, 1917).

where Ki,u and K:( are the adsorption constants of Na+ and K+ ion, respectively, on this type of site in iM-1. Similarly, for K+-ion adsorption on type I sites :

F+1nd =

Ifl'K":<

[K+l<,x

(10) 1 I?:,, "a+] ('I K":; [ K + ] A plot of [Na+],,,,against INa-t-],.,is a hyperbola (see, for example Figs. 2G and 30) : That is, at low "a+- I(,x values the sites arc largely empty and there is a proportionate increase of adsorhed Na-I- icin with increasing "a+] k.x. A t higher [Na-+] ,.P, the sites become more and more occupied. T h e increment of adsorbed Na + ion with a unit increment in I N a + If.X diminishes steadily until the total adsorbed N a + ion approaches a maximum value equal to the total number o f sites, J [ . Equation 3 shows that the adsorbed Na+ ion can be decreased by increasing the concentration of the competing K-t- ion. At a sufficiently high Ki--ion concentration, the adsorbed Na+ ion can he reduced to a n insignificant level. Equation 9 can be written in reciprocal form:

+

+

111

If

1

"a+

-

1a,,

1

(1

[fl'Qu

+ qi[ K f l e x )

1 "a+

1

ft'S

1 (11)

[fl'

If one plots l/[Na+];,,] as a function of l / l N a + ] , L x at constant [KfJ,,, one obtains a straight line. T h e intercept on the ordinate (i.e., at 1,"Na+Icr z 0 ) is the reciprocal of the total concentration o f type I sites. T h e slope of the lines T h e reciprocal form of Eq. ( 1 0 ) for is a function of [K+-.le., I?:, and K+-ion adsorption is

20

GILBERT N. LING

If Eqs. (7) and ( 9 ) are substituted into Eq. (3) and Eqs. (8) and (10) are substituted into Eq. (4), one obtains the explicit equations for the total Na+and K+-ion concentrations in living cells.

and

zia

Here and gi are the adsorption constants of N a + and K + ion on the Lth adsorption site. Eqs. (13) and (14) are basically similar to an equation first introduced by Troschin for sugar distribution in cells (Troschin, 1958). They will be henceforth referred to as the Troschin equations.

B. THEMOLECULAR MECHANISM FOR SOLUTEDISTRIBUTION IN LIVINGCELLS:EXPERIMENTAL EVIDENCE 1. The Physical State of Watei’ i j z Living Cells and Model Systems

The adsorption of molecules in polarized multilayers on solid surfaces is described by an equation of Bradley (Bradley, 1936; see also de Boer and Zwikker, 1929) Po = K1 K,’ log,, K, (15)

P

+

where a is the amount of gas (water in the present case) adsorbed, at vapor pressure p ; Po is the vapor pressure at satpration. K1, K , and K , are constants. As Fig. 10 shows, this equation is followed by the sorption of water onto an isolated protein, sheep’s wool (Ling, 1 9 6 5 ~ ) The . entire range of experimental data from Bull (1944) fit Eq. (15). Similar results are obtained for the sorption of water onto collagen (Ling, 1 9 6 5 ~ )and other proteins and macromolecules (Mellon and Hoover, 1950). These experiments indicate that proteins, in z h o , have the capability of organizing the water with which they are in contact into polarized multilayers. One would expect that cellulu proteins, iiz z~iuo,would have the same property and, indeed, there are two lines of evidence indicating that the water in cells ii in a different state from the water in a dilute salt solution.

A N E W MODEL FOR THE LIVING CELL

21

i. Intracellular freezing pattern. Normal water or a dilute salt solution, when cooled to below O"C., forms tridymite ice in hexagonal patterns (Hallett, 1965). In overall shape, normal ice is featherlike with branches at regular intervals (Fig. 11). Ice formed in supercooled living cells by seeding with an ice-tipped micropipette is abnormal in structure (Chambers and Hale, 1932; Rapatz and Bradley isotherm

log log ( % / P I +2

FIG 10. Sorption of water vapor on sheep's wool, plotted according to the Bradley multilayer adsorption isotherm. Data of Bull (1944) (Ling, 1966a, by permission of Annals of the New York Academy of Sciences).

Luyet, 1958). In skeletal muscle, for example, longitudinal spikes following the direction of muscle protein filaments are formed (Chambers and Hale, 1932; Rapatz and Luyet, 1958) (Fig. 1 2 ) . Such spikes, which have no branches, indicate that the intracellular water is not normal. That this abnormality is intimately associated with the structure and orientation of the cell proteins (Ling, 1965a, 1966b, 1967a) is indicated by the demonstration that when the myofilaments are twisted, the spikes also become twisted (Chambers and Hale, 1932). ii. Nuclear magnetic resoname ( N M R ). The recently developed technique of nuclear magnetic resonance spectroscopy is being used with increasing effectiveness to study the physical state of water in cells. From the widening of the NMR signal of the water in rabbit sciatic nerve and its dependence on the orientation of the magnetic field, Chapman and McLauchlen concluded that the bulk of this intracellular water is not normal but is in a partially oriented state (Chapman and McLauchlan, 1967) (Fig. 13). Fritz and Swift (1967) reached a similar conclusion working with frog nerves. [For criticism of the earlier conclusion of Bratton et al. (1965) that only part of the muscle cell water suffers restricted rotation, also see Fritz and Swift (1967)l. 2. Ionic Exclusion f r o m the Water of Sheep's W o o l ,

Frog Muscle,

atzd in Actomyosin Gel Figure 14 shows the equilibrium ionic uptake of sheep's wool. In the presence of high concentrations of a competing ion, a plot of the accumulated labeled

22

GILBERT N. LING

FIG. 11. Ice crystal growth in supercooled pure water initiated by the insertion of single crystal at -3.5"C. (Hallett, 1965, by permission of Federation Proceedings).

FIG. 12. Development of an ice spear in a single muscle fiber at - 2 . 5 " and Luyet, 1c)>8, by permission of Eiodynamicu).

3

C . (Rapatz

A NEW MODEL FOR THE LIVING C E L L

23

100 cps

FIG. 13. N M R spectra of water i n rabbit sciatic iierve: (top) with nerve axis parallel to the applied field; (center) axis perpendicular to the field: (bottom) nerve axis at approximately S t " to the direction of the applied field (Chapman and McLauchlan, 1967, by permission of Na/ur.r). 160

I

-0

:

lwool O

0

I

I

I

No nonlobeled R b 0 5 M nonlobelecl Rb

120

L

D

[Rb+l,,

(mM)

FIG. 1.4. Equilibrium labeled Rb+-ion distribution in sheep's wool in the presence of varying concentrations of competing nonlabeled Rb+ ion. Inc-reasing the nonlabeled R b + ion concentration from 0 to 0 . 5 M reduced the labeled Rb+ c-oncentration. Further increase from 0 . 5 to 0.7 M , however, produced no further decrease of the labeled Rb+-ion concentration in the wool and the distribution curve is linear. The wool contains 30% water. Assuming all labelecl Rb+ ion in the presence of 0 . 5 or 0.7 M nonlabeled Rbt- ion to be in this water. the distribution coethcient of the labeled R b + ion is about 0.29.

24

GILBERT N. LING

Rb+ ion as a function of the external Rb+-ion concentration approACh es a straight line (lower line). In the absence of competing ion, there is an additional fraction of Rb+ ion in the wool which is not linearly related to the external Rb+-ion concentration. In fact, subtracting the lower curve from the upper curve gives a hyperbola. These data indicate that ionic accumulation in sheep’s wool follows an equation similar to Eq. (13) or ( 1 4 ) (Ling, 1965b). Considering only the linear fraction, we note that the slope of this line is only 0.12, indicating that in this system which constitutes 30% protein and 70% water, Rb+ ion is accommodated to only 29% of its concentration in the external solution. Thus, the effect of the polarization of the water into multilayers as described above is the partial exclusion of solutes from this water. Figure 15 shows the equilibrium Na+-ion uptake of living frog sartorius

-k

300

5 0 mM 10 0mM

0, Ln

0

r

20

40

60

80

100

120

C No+lex( mM)

FIG. 15. Equilibrium distribution of N a + ion in frog sartorius muscle in the presence of varying external K+-ion concentrations. T h e external K+-ion concentrations are 2.5 mM, 5.0 mM, and 10.0 m M. T h e data were calculated on the basis of a 10% extracellular space. The slope of the straight line going through the points at the higher K+-ion concentrations is 0.14. T h e muscle cells contain 78v0 water. If all N a f ion in the cell at the higher K+-ion concentrations is assumed to be in the cell water, the equilibrium distribution coefficient of N a + ion between the cell water and the external medium is 0.18.

muscle (Ling, 1965a,b, 1966b; Troschin, 1958). As in sheep’s wool, increasing the concentration of competing K + ion lowers the intracellular Na+-ion concentration. Again, however, the intracellular Na+-ion concentration is not reduced to zero at high K+ ion, Instead, there is a level beyond which no further increase of K+-ion concentration has any effect on the N a + ion in the cell. This remaining fraction (B) of intracellular Na+-ion concentration is also linearly related to the external Na+-ion concentration with a slope of about 0.14. As in the case of

25

A NEW MODEL FOR THE LIVING CELL

sheep’s wool, at lower concentrations of competing K + ion, there is an additional fraction of N a + ion (A fraction) which is roughly hyperbolic in its relation to the external Na+-ion concentration (see Section 111, D,lO). Figure 16 shows the equilibrium Rb+ -ion distribution in a protein-water

160

-

30 -

,-

40

0.5

10

15

20

[Rb+],,(mM)

FIG. 16. The equilibrium distribution of labeled Rb+ ion in isolated actomyosin gel at neutral ( 6 . 7 ) and acidic pH (4.3).

system (actomyosin) taken from the living cell (Ling and Ochsenfeld, 1968a). Here the Rb+-ion distribution follows the same pattern as seen in sheep’s wool and in living muscle. In the presence of a strongly competing cation (H+ ion, pH 4.4),the concentration of Rbi- ion in the gel is linearly related to the external Rb+-ion concentration (B fraction). At lower H+-ion concentration, an additional fraction ( A ) , having the typical characteristics of a Langmuir adsorption isotherm, is superimposed on the B fraction. The actomyosin gel is perfectly homogeneous and contains no anatomical structures, yet the Rb+-ion concentration in the gel water at low pH is only a fraction (50-70%) of that of the external solution. From these data we reach the conclusion that the water in this gel again has abnormal solubilities for alkali-metal ions. The actomyosin gel used in the investigation shown in Pig. 16 is very dilute (3-576 protein versus 95-97% water). Together with other proteins, actomyosin exists as a much more concentrated gel in the living muscle (205% protein versus 80% water). The ionexclusion property of the water of isolated actomyosin gel is a compelling reason to believe that the ions of the B fraction of Na+ ion in frog muscle (Fig. 1 5 ) represent the fraction in the muscle cell water and that the water in muscle

26

G I L H I X I ’ N. LING

cells similar to that in sheep’s wool accommodates less alkali-metal ion than the external aqueous solution. 3. T h e Ei2tiopit B~isjso f Ioii Excliision f r o m the Water

At

Actomyosiii Gel

Ling and Ochsenfeld studied the temperature coefficient of the exclusion of Rb+ and other alkali-metal ions from the water in isolated actomyosin gel (Ling and Ochsenfeld, 196821). From these studies, the conclusion was drawn that the exclusion was primarily attributable to the unfavorable entropy of the hydrated alkali-metal ion in the gel water.4

4. The Adsorbed Frmtjoiz o f Na+ I o n W e have shown above (Fig. 1 5 ) that increasing the K+-ion concentration to a certain level brings about a reduction in the level of intracellular N a + ion in frog sartorius muscle. This suggests that the K-i- ion is competing with an adsorbed fraction of N a + ion. There is now considerable independent evidence that such a fraction does indeed exist: ( I ) Lewis and Saroff (1957) showed N a + ion binding onto isolated actomyosin. ( 2 ) Hinke (1959) used an intracellular Na+-ion-sensitive microelectrode to study the activity of N a + ion in frog muscle and squid axon and concluded that about two-thirds of the intracellular Na+ ion is bound.6 ( 3 ) Recently Cope (1967) has used the techniques of N M R to study the N a + ion in muscle and has concluded that about 70%) of this ion is complexed. A similar conclusion was reached by Rotunno et al. (1967) in regard to N a + ion in frog skin cells. Thus, under normal conditions the intracellular N a + ion is made up of an adsorbed fraction as well as the interstitial fraction discussed above. 5.

K 4-- 1 ~ nArri/mi/lation

a. Interpietatioil Accordiir
Varying the external nonlabeled K+-ion concentration produces a family of straight lines converging at the same locus on the ordinate. Such a plot is characteristic of a Langmuir adsorption isotherm [ Eq. ( 1 1 ) I and in terms of the association-induction hypothesis indicates that the intracellular K + ion is all in 4 N o / r Added in P w u f : Recently Gary-Bobo and Solomon [Gary-Bobo, C. M. and Solomon, A . K. (1968). /. Gerr. Phyriol. 52, 8251 published results of their studies on the distribution of K + ion in hemoglobin solutions. At low pH, an exclusion of K + ion from this protein solution was also observed and was attributed by these authors to a Donnan effect. Whether this is a better explanation applicable to our in rlitro actomyosin data ( o r vice versa) remains t~ be determined. 5 Nore Addrd / o Proof: For comments on the use of ion-sensitive electrode to assay ionic activity in living cells see Ling [Ling, G. (1969). Nature 221, 3861.

A NEW' MODEL FOR THE LIVING CELL

27

the adsorbed state. The concentration of K + ion is many times higher in the muscle cell (90 m M ) than in the frog plasma ( 2 . 5 m M ) . Thus the interstitial K+ ion can amount to only a fraction of the concentration in the external solution and hence less than 1-2Cjc which is the limit of experimental error. From

0

0.I ([K+],,)-'

0.2

0.3

0.4

(rnrnole/liter)-'

FIG. 17. lntracellular labeled K+-ion concentration plotted reciprocally against the external labeled K+-ion concentration. with which i t is in equilibrium, in the presence of 0, 20, and 5 0 mmole/liter of nonlabeled potassium acetate. Labeled K + was also in the form of acetate salt. Each point represents the labeled K+-ion concentration in a single frog sartorius muscle; lines obtained by the method of least squares. On the lowest curve within the arc3 from [K+ICS-I = 0 to 0.05 (mmole/liter)-1 and from [ K + l i , , - l = 0 to 0.01 (mmoie/kg.)--l, a total of 23 points was determined; they fall so close to one another that only a few could be represented. All the others would be superimposed on these. Data from two series of experiments (Ling and Ochsenfeld, 1966, by permission of The Jourtzal of General Physiology).

these data we obtain an association constant for K + ion of 665 M-1 and a concentration of anionic sites I f ] equal to 143 mmoles per kilogram fresh cells. From the known protein content of muscle and the percentage of acidic side chains on these proteins the maximum number of anionic sites in the cell is 260 mmoles/kg. which is more than enough to account for this number of K+-ion adsorbing sites (Ling and Ochsenfeld, 1966) .I; (i

Note Added iiz Proof: For additional NMR evidence that the bulk of intracellular

K+ ion is in an adsorbed state see Ling and Cope [Ling, G. N., and Cope, F. W., Srierlce 163, 1335 (1969)l.

28

GlLBBRT N. LING

b. Alternative Interpretatioizs. i. Donnan equilibrium. While competition is usually considered a distinguishing feature of the Langmuir adsorption isotherm, it may also follow from the fact that in a Donnan equilibrium macroscopic electroneutrality must be maintained (Ling and Ochsenfeld, 1966). However, a Donnan system can be differentiated from a system in which the K + ion is adsorbed by studying the effect of a second ion on the equilibrium distribution of labeled K+ ion and comparing the effect of this second ion with that of nonlabeled K + ion. In the case of a Donnan equilibrium, the difference between the two effects depends on the difference between the respective activity coefficients. In Fig. 18 we have compared the effect of similar concentrations of Cs+ ion and nonlabeled K + ion on the equilibrium accumulation of K+ ion. The experimentally observed difference is far larger than can possibly be accounted for by the difference in the activity coefficients (less than 2 % ) . ii. The carrier model. According to a recent version of the membrane theory, the majority of K + ion enter the cell by combining with certain hypothetical “carrier” molecules which then ferry the K + ion across the lipid part of the plasma membrane (Osterhout, 1936; Jacques, 1936; Epstein and Hagen, 1952). Once the carrier-K+-ion complex reaches the inner surface of the cell membrane, the K+ ion dissociates, entering the cell water. This idea is similar to Overton’s postulation, at the turn of the twentieth century, of adenoid or secretary activity. The best evidence that can be cited in favor of such a carrier hypothesis comes from studies of the initial rate of entry of K+ ion. Thus this rate of entry shows saturability (i.e., as the external K + ion increases, the rate of entry per unit increment of external K+ -ion concentration steadily decreases to approach zero) and competition (i.e., similar alkali-metal ions compete with K+ ion and reduce its rate of entry) (Fig. 19; Ling and Ochsenfeld, 1965). 1;zmdamental difficulties of the carrier model. (1) Historically the entry of K + ion was always considered to involve diffusion of the free ion through the cell membrane. Bernstein’s original version of the membrane theory (Bernstein, 1902) of the resting potential was based on this proposition. Postulation of the Na+ pump did not involve a revision of this concept. Thus the Hodgkin-KatzGoldman equation [Eq. (16)] was derived on the basis that K + ion migrates through the cell membrane as a postively charged ion under the influence of the constant electric field. If the K + ion traverses the cell membrane in combination with a carrier, there are two possibilities: Either the carrier molecule bears an anionic charge, or it is a neutral molecule. If K + ion combines with a carrier bearing a negative charge, the foundation of the Hodgkin-Katz-Goldman equation is removed: The constant electric field would have no effect on the direction of motion of a neutral complex. If, on the other hand, one considers the carrier itself to be uncharged, the carrier-K+-ion complex would be a monovalent cation. This would overcome the above-mentioned difficulty. Another difficulty now arises, however, because of

A N E W MODEL FOR T H E LIVING CELL

29

the enormous gain of free energy when a charged ion is brought into a lipid phase. To achieve this, the carrier must possess both a strong affinity for K + ion and qualities that allow itself to stay permanently in the lipid membrane. It is

([CS'I e x )

-' (mmole/liter)-'

FIG. 18. Equilibrium labeled Cs +-ion concentration in muscle cells plotted reciprocally against the external Cs+-ion concentration with which it is in equilibrium. Competing K+-ion concentrations are 0 , 20, and 50 mmoleiliter respectively. Both Cs+ and K + were in the form of acetates (24'C.). Each point represents a single determination on one frog sartorius muscle; lines obtained by the method of least squares. The effect of K + ion on the accumulation of labeled K + ion (dotted lines) taken from Fig. 1 for comparison (Ling and Ochsenfeld, 1966, by permission of T h e Journal of General Physiology).

30

GILBERT N. LING

not easy to construct a molecule with these conflicting attributes. Perhaps this is why, in spite of the long history of the carrier, there is no experimental model that demonstrates the behavior observed, e.g., saturability and competition. It hardly needs to be pointed out that saturability and competition are evidence only that entry is associated with a limited number of sites. Thus, the rate of entry of K + ion into a sheet of ion-exchange resin shows entirely similar saturability and competition (Ling and Ochsenfeld, 1965) (see also Fig. 2 0 ) . Here such a resin is nothing but a three-dimensional anion-bearing matrix possessing neither carriers nor membranes. Recently, Ling and Ochsenfeld have succeeded in demonstrating similar saturability and competition in the rate of entry of Rb+ ion into layers of an isolated cytoplasmic protein-water system (actomyosin gel). In such a system the rate of entry is not surface- but bulkphase-limited (Ling and Ochsenfeld, 1968b). This data shows that even in the absence of a cell membrane, ion entry follows the kind of kinetics shown in Fig. 19. Thus saturability and competition result from the presence of a limited number of adsorption sites throughout the entire protein-water system. ( 2 ) Figure 21 shows the reciprocal of the initial rate of entry of N a + ion plotted as a function of the reciprocal of the external N a + ion concentration in the presence of varying concentrations of K + ion. Although there is a fraction of N a + ion entry whose rate is decreased by increasing the K+-ion concentration

0.I

0.2

0.3

0.4

0.5

([K +IcJ-', (mmole/liter)-' FIG. 19. Inhibitory effect of 2 5 mmole/liter of Rb+, Csf, and nonlabeled K + ion on the initial rate of entry of labeled K + ion into frog sartorius muscles. Lowest, nonlabeled curve represents the rate of K+-ion entry with no added competing ion. Muscles were soaked for 30 minutes at 24°C.. followed by 10 minutes of washing at 0°C. Each point represents a single determination on two sartorius muscles (Ling and Ochsenfeld, 1965, by permission of Biophy.ricd Journal).

A N E W M O D E L FOR THE LIVING C E L L

I

31

Pc

1AC

0. I

0.I

0

0.2

I 0.3

([cs+I,,~)-', (mmole/liter)-' Effects of CsCI. KCI, NaCI, and LiCl on the initial rate of entry of labeled Cs+ ion into ion-exchange resin sheets. Nallilm-1 strips soaked for 2 minutes at 5°C. i n an experimental solution containing (approximately) 2.5 mmolr tris buffer at pH 7.0, the labeled entrant ion and nonlaheled ion ( 4 )mmole/liter) as indicated in the figure. Strips washed for 10 seconds in cold distilled water ( 0 ° C . ) before counting (Ling and Ochsenfeld. 1965, by permission of Biuphyriiul J O M W J I ) . T'iG. 20.

from 0 to 30 m M ;another fraction, shown by the upper line, is unaffected by K+-ion concentrations as high as 100 miM. This upper line goes through the origin indicating that there is no saturation in the entry of this fraction (i.e., an inflnite number of "adsorption sites"). Thus there is a fraction of N a + ion which shows neither competition nor saturation and thus cannot be conceived to enter via carriers, but must enter via aqueous channels. If such aqueous channels, large enough for the hydrated N a + ion, exist, it is hard to understand how they could exclude the smaller hydrated K + ion.

c.

ANSWERSTO

F U N D A M E N T A L C R I T I C I S M S O F IONI(:

A D S O R P T I O N IN

LIVINGC E L L S

There are a number of lines of experimental evidence that seem, superficially at least, to contradict the adsorption model for ionic accumulation in living cells. Let us describe this evidence first and then discuss some recent advances in our understanding o f adsorption that serve to change this picture.

32

GILBERT N. LING

40

$ 3.0 \

s

\

0, 0

%

2.0

v

-

6

0

P

v

10

0

0.I

([No+],,)

0.2

0.3

0.4

-I (rn mole / liter)

FIG. 2 1 . Effect of various concentrations of K + ion on the initial rate of entry of N a + ion into frog sartorius muscles. Increasing the Kf -ion concentration from 30 mmole/liter to 100 mmole/liter causes no apparent effect on the rate of Na+-ion entry, while reduction from 30 mmole/liter to 2.5 mmole/liter causes an increased rate of Naf-ion entry. Fifteen minutes of soaking at 25°C. were followed by washing for 10 minutes at 0'. Each point represents average of three individual determinations (Ling and Ochsenfeld, 1965, by permission of Biophysical Journal).

(1) Living cells are isotonic with an approximately 0.1 M aqueous NaCl solution. This demands that the osmotic activity within the cells be equal to that of this solution. Since the total ionic concentration in the cell is approximately 0.1 M and K + ion constitutes the bulk of the cations, it follows that all or nearly all of this K+ ion as well as the intracellular anions must be in a free state such as is found in 0.1 M NaCl (Hill, 1930). (2) If a pair of electrodes are placed on the surface of an intact nerve or muscle fiber, the total resistance measured is, relatively speaking, little affected by the distance between the electrodes. This was interpreted as being the result of high membrane resistance and low cytoplasmic resistance (the core-conductor theory of Hermann (Hermann, 1879). The low cytoplasmic resistance was interpreted as indicating complete or nearly complete dissociation of the intracellular K + ion. (3) The mobility of labeled K+ ion in squid axons averages 1.5 x 10-5 cm.2/sec. This value is not too far from the diffusion coefficient of K + ion in an aqueous KC1 solution of a concentration comparable to that of squid cytoplasm (2.14 x 10-5 cm.2/sec.). Thus, the conclusion was drawn that the bulk of K+ ion inside an axon exists as free ions (Hodgkin and Keynes, 1953).

A N I X MODEL FOR THE LIVING CELL

33

(4) Bernstein’s membrane theory of cellular electrical potential (Bernstein, 1902) and its modified version introduced by Hodgkin and Katz (Hodgkin and

Katz, 1949) predicts the correct magnitude of the resting potential. Since this theory is based on the assumption of complete dissociation of intracellular K + ion, it follows that the bulk of the intracellular K + ion must be in the free state. Considered together, this evidence is too compelling to ignore. It is not surprising that many cell biologists have so far considered the membrane theory a better choice. However, with a fuller understanding of both the state of water in living cells and the nature of adsorption, it becomes possible to resolve the conflict. In the following discussion, we will note point-by-point answers to the criticisms listed above. (1) The postulate that the bulk of the intracellular ions must be free in order that the cell be iso-osmotic with a 0.1 M NaCl solution is valid only in so far as there is no interaction between the cell water and the cell proteins as postulated by the membrane theory (see above). There is now overwhelming evidence that the bulk of the water in the cell is adsorbed as polarized multilayers on the proteins. Thus the lowering of the activity of water within the cell (expressed as osmotic pressure) is explained by the interaction of water with the proteins. With this degree of interaction with and lowering of water activity by proteins, the fact that the total osmotic pressure does not exceed that of a 0.1 M NaCl solution, suggests intracellular ionic adsorption. Arguments ( 2 ) and ( 3 ) can be answered together. It is well known that in nerve and muscle cells the major protein components are oriented in a longitudinal direction (Chambers and Kao, 1952; Huxley, 1957). Thus one might anticipate a longitudinal orientation of the anionic sites as shown diagrammatically in Fig. 22. In this diagram concentric circles surrounding each anionic site represent equipotential lines. The closer they are to the anionic site, the lower the potential. As the diagram shows, these lines overlap. In seeking a minimum energy path, an adsorbed K + ion on one site can, therefore, pass from one site to another by following the overlapping regions of low energy. The activation energy for this migration is therefore low. Thus one may say that by a sequential orderly arrangement of anionic sites, a “conduction band” is provided for the K + ion (for further details, see Ling, 1962, Chap. 11). On the other hand, the distance between the myofilaments is such that a greater distance separates sites on neighboring protein chains. In consequence, a much higher activation energy has to be overcome before a radial migration of the K + ion can be achieved. If this hypothesis is correct, we might anticipate the longitudinal migration of K + ion in the cell to be equal to and perhaps even faster than K+-ion migration in a dilute salt solution. On the other hand, radial migration of K + ion would be slow. This can then explain both the core conduction behavior of muscle and nerve cells and the longitudinal K+-ion mobilities in squid axons. Two lines of in-

34

GILBERT N. LING

dependent experimental evidence suggest that this explanation is the proper one: (a) It has been known for a long time that K+-ion migration on the surface of glass that bears anionic sites is faster than K + - i o n migration in dilute solution (McBain and Peaker, 1930; Mysels and McBain, 1948; Nielsen et nl., 1952). (13) Schwindewolf (1953) and Heckmann (1953) have shown that the elec-

FIG. 2 2 . Diagrammatic illustration of an effective “conduction band” for adsorbed ions along a longitudinal array of anionic sites. Transversely, the anionic sites are too far apart to produce a conduction band and low conductance results. T h e concentric circles represent equipotential lines with lower potentials closer to the anions. T h e line with the arrows indicates a probable path for an adsorbed ion moving along the “conduction band”; it has to overcome relatively low activation energies because of the overlapping of the elec-tric fields. Much larger a d v a t i o n energy has to be overcome for transverse migration.

trical conductance of a solution containing linear anionic polymers (i.e., polyphosphate, thymonucleic acid, or solubilized silk protein) and flowing in a long tube is anisotropic. That is, the conductance is higher in the direction pzrallel to the flow of the solute than transversely. This anisotropic conductance can also be understood in terms of the interpretation offered in Fig. 2 2 if one recalls that flow produces an alignment effect on the long-chain anions. Such an alignment in living muscle and nerve cells is created by the pattern of cell growth. ( 4 ) There is now a large collection of self-consistent data in support of part of the Bernstein or the Hodgkin-Katz-Goldman equation for the cellular potential :

KT

+ +

PI< [K+]iii 11, = __ In __-1;

P K t t “a+ P , I K + I,.~ P,,, IN+ ;

liii

+ P,, +

I<.~

~

(

[C1-]ex 7

1

ICI-

(16)

lill

where P,,, PN,l,and are the K - t - , N a + - and Clk-ion permeability constants “a+ 1 ,,), and [Cl- I ill their intracellular concentrations. R and 1: and I K + I are the gas and Varaday constants, respectively. The dependence of the potential on the absolute temperature T and its logarithmic relation to the external K +

A N E W MODEL F O R T H E LIVING C E L L

35

and Na+-ion concentrations have been repeatedly verified (McDonald, 1900; Cowan, 1934; Curtis and Cole, 1942; Ling and Woodbury, 1949; Ling, 1967b, 1962, Chapt. 1 0 ) . However, attempts to demonstrate a dependence of the potential on the intracellular K + - and Na+-ion concentrations have met with contradictory results. Thus injection of highly concentrated salt solutions into squid axons and muscle fibers (Grundfest et a/., 1945; Falk and Gerard, 1954), or leaching of K + and N a + ion from muscle cells, produces none of the changes anticipated on the basis of the membrane theory (Tobias, 1950; Koketsu and Kimura, 1960). N o r does changing the external chloride concentration produce a permanent change in the potential the way changing the external K+-ion concentration does (Hodgkin and Horowicz, 1959). Other experiments performed both on living muscle cells as well as those model systems that played a major part in the development of the membrane theories of cellular electrical potential (e.g., glass electrodes, collodion electrodes) (Ling, 1960, 196717, 1962, Chapt. l o ) led the reviewer to the conclusion that this potential is in fact not a membrane potential but rather a phase-boundary potential (i.e., a potential arising at the interface of the external solution and the protein-water fixedcharge system; see also Beutner, 1920). As such, this potential is determined by the density and nature of the anionic groups on the proteins of the cell surface. In simplified form the equation f o r the potential is thus 111

-

RT = constant - __ In ( K , ; [ K + I, I;

-

+ K,,

( N a + I,.,)

(17)

where K , and K,, are the adsorption constants of the K + and N a + ions on the surface anionic sites. Although this equation has a different basis from the Hodgkin-Katz-Goldman equation, it is formally identical with that part of it which has been experimentally verified. It does not contain those terms relating to the intracellular ionic concentrations or the external Cl--ion concentrat'ion whose relation to the potential has not been verified. T h e adsorption constants

-

-

K,, replace the permeability constants in Eq. (16). According to the associatim-induction hypothesis, these constants change during excitation because of an all-or-none cooperative change in the r-value of the anionic sites (see Fig.

K,; and

-

7 ) . Thus at rest, the r-value is such that K ,

-

> > Ksi,. In consequence, Eq.

(17)

reduces to (18)

where i l l r is the potential of the resting cell. This relation has been confirmed over and over again in a large variety of tissues since it was first discovered by McDonAd (McDonald, 1 9 0 0 ) .

36

GILBERT N . LING

During excitation, the r-value shifts to a value such that KNa> K g and Eq. ( 1i ) now approaches

This relation between the magnitude of the action potential, 1 1 ) and ~ ~ log “a+] was discovered and extensively investigated by Hodgkin and his co-workers (Hodgkin, 1951; Hodgkin and Katz, 1949). Anticipating work to be presented in a following section, I would like to point out that there is theoretical reason to believe that a local chmge in the physical state of the water may accompany this c-value shift. Experimental NMR studies by Fritz and Swift (1967) have demonstrated a change in the state of water during depolarization of nerve. Such a change, accompanied by an increase in the N a t - i o n preference of the cell surface sites would contribute to the creation of the action potential profile. In summary, the failure to demonstrate a consistent correlation between the cellular potential and the intracellular K + and Na+ ion indicates that the cellular electrical potential cannot be construed to be a proof of the membrane theory. In fact, this failure adds very important evidence against it.

D. MOLECULAR MECHANISMS I N THE INTEGRATIVE FUNCTION OF PROTOPLASM In the preceding sections, we have examined the nature of the protoplasm in terms of the association-induction hypothesis. W e have shown that the equilibrium properties of the protein-water fixed-charge system are capable of accounting for most experimental observations concerning solute distribution in living cells. However, living protoplasm is vastly different from sheep’s wool or even isolated actomyosin gel. To illustrate this difference let us again consider the bacterial flagellum. This pure protein-water system as it exists on the bacteria is active, a wool fiber is not. On closer examination we find that this activity has the following characteristics : I t is the result of a reversible alteration between two states (short and long), the change of state is controlled by a signal received at a distance [from the basal granule, Astbury’s signal box (Astbury, 1951)] and the restoration of the flagellum to its original state is energized, also at the basal granule. In the following discussion, we shall examine a theoretical model capable of performing these functions. 1 . Two

Simple ModelJ

Let us consider a very simple mudel first. If we join soft iron nails end to end with pieces of string as shown in Fig. 23A, they would distribute themselves in a

A NEW MODEL FOR THE LIVING CELL

37

random manner and not interact with iron filings strewn around them. If a strong magnet is then brought near the nail at one end, a chain reaction of magnetization occurs ending with all the nails magnetized and the iron filings strongly oriented around the nails. In this process we have demonstrated both energy and information transfer over a distance made possible by the magnetic

FIG. 23. T w o simple moles demonstrating information and energy transfer over distances due to propagated short-range interactions. ( A ) A chain of soft iron nails joined end to end with pieces of string is randomly arrayed arid does not interact with the surrounding iron filings. The approach of a magnet causes propagated alignment of the nails and interaction with the iron filings. ( B ) Electrons in a series of insulators are uniformly distributed before the approach of the electrified rod, R. Approach of the rod displaces the electrons by induction such that the insulator becomes polarized with regions of low electron density and regions of high electron density.

susceptibilities of soft iron. (If we had used wooden nails, for example, nothing would have happened). This magnetic model, of course, has its electrical analog (Fig. 2 3 B ) . An electrically charged rod when brought near a series of closely placed insulators will produce, by induction, alternating negatively and positively charged poles on the insulators. When the electrified rod is removed, the insulators again lose their electrical polarization. Here once more, we have energy and information transfer made possible by a series of closely placed polarizable materials. Depending on the polarizing influence of the electrified rod, there is a redistribution of electrons in the chain o f insulators such that localized electron-poor and electron-rich regions ,ire created. These examples illustrate the mechanism of indziction.

38

GILBERT N. LING

I have suggested as a part of the association-induction hypothesis that the ability of protoplasm to function coherently relies on a fundamentally similar induction mechanism. T h e polarizable material is, in fact, the polypeptide chain m d its appendages, the side chains. 2. The Basic iMech~uii~rt~-TheImhctive Effect

Acetic acid (CH,,COOH) is a weak acid. This means that its carboxyl group interacts strongly with a proton in aqueous solution and the fraction of carboxyl groups in the dissociated state is relatively small. Trichloroacetic acid (CC1,C O O H ) , on the other hand, is a very strong acid. This means that its carboxyl group interacts only weakly with a proton; the fraction of carboxyl groups in the dissociated state is large. Trichloracetic acid is derived by substituting chlorine atoms for the three hydrogen atoms in the methyl group. Since chlorine atoms are more electronegative than hydrogen atoms (i.e., the chlorine atom has a greater propensity to draw electrons toward itself than does the hydrogen atom), this substitution produces a decrease in the electron density of the carboxyl group (i.e., a c-value decrease). Since the energy of interaction between the carboxyl group and the proton is primarily electrostatic in nature, a decrease of c-value leads to a weakening of the interaction energy and more H + ion is found in the dissociated state. Thus the substitution of chlorine atoms for hydrogen atoms produces an inductive change in the dissociation constant of an acid group on a different portion of the molecule. This inductive effect is a general property of organic compounds and its consequences are not limited to a chmge of acid dissociation constants. Taft, for example, showed that a wide variety of equilibrium and kinetic properties of organic compounds are influenced in a predictable manner by the inductive effects of a long list of substituents including hydrogen and chlorine atoms (Taft, 1960; Taft and Lewis, 1958; Hammett, 1940; Ling, 1964a,b). Among these properties is the strength of the hydrogen bonds formed by many of these compounds (Ling, 1964a,b, 1962, Chapt. 7 ) . Thus substituting a hydrogen atom for a chlorine atom on diethyl ether reduces the electron density of the ether oxygen, and thus lowers its proton-accepting power (Gordy and Stanford, 1941). 3. H o w Far Cali the I?idr/rtiue Effect Be Trammitted?

T h e dissociation constants of the tx-amino m d a-carboxyl groups of amino acids YH” (R-C-COOH)

vary because of the varying “electronegativity” of the side chain ( R ) . When

39

A NEW MODEL FOR THE LIVING CELL

peptides are formed by joining amino acids, and most of the a-carboxyl groups and a-amino groups are transformed into proton-accepting C=O and proton-donating NH groups. However, the (1-carboxyl group and (1-amino group at the ends of the peptide remain as such. T h e pK value of these terminal groups in a series of glycine peptides can be studied as an indication of just how far the inductive effect produced by substituting a hydrogen atom by a glycyl group (NH,CH,CO) can be transmitted. Table I shows that this substitution has an effect on the uTABLE 1 Carboxpl group

Amino group

I M NaCl

Water N H ,CH,COO H NH,CH,CONHCH,COOH NH,(CH,CONH),CH,COOH NH,(CH,CONH),,CH,COOH NH.,(CH,CONH),CH,COOH NH,(CH,CONH),CH,COOH

9.70 R.20 8.00

7.75 7.70 7.60

Water

9.60

9:lc)

2:42

8.13 7.91 7.75

8.07 7.83 7.93

3.13

7.70 7.60

3.00 3.05 3.cr5 3.05

~

~-

1M NaCl 2.3.4 3.06 1.26 3.05

1.05 3.05

3.02

3.33 3.39 3.50 ~

-

carboxyl oxygen even though it is separated from the substituent by one peptide amide group (-CONH-), two saturated carbon atoms, one nitrogen atom, and one carboxyl carbon atom. This example shows that prctein chains are unusually polarizable. O n this basis, one may anticipate that changes of the “electronegativity” of a side chain, created by the dissociation of a proton for example, may exercise significant influence on its two neighboring C O N H groups:

-

Y,e

I

-

-N-C-C-N-CI H

It

O

? ,?I -N-C-C-N-C-+

I I1 H O

I H

II O

I H

II

H+

O

Conversely the r-value of the cnrboxyl group will be changed if the N-C I I1 H O

groups change their H-bonding partners to others of different hydrogen-bonding (or polarizing) strength.

4. The I n t e s p l q of Euer;q), mid Entropy The models of magnetized iron nails and electrified insulators deal with macroscopic objects; here energy alone plays n significant role. Phenomena such

40

GILBERT N. LING

as ion adsorption and exclusion are microscopic events. As such, they depend on entropy as well as on energy. A simple example is the sublimation of ice at below-freezing temperatures. Thus both a housewife drying laundry in winter and a biochemist freeze-drying an enzyme depends on the large gain of translational and rotational entropy of the water when the ice vaporizes. From the standpoint of energy alone, such a step is highly unfavorable. Let us now consider a segment of a polypeptide chain containing m peptides. It can wind itself into, for example, an a-helix or it can become fully extended (a so-called “random coil”), In the helical form all the NHCO groups of the backbone have formed hydrogen bonds with other C O N H groups on the same polypeptide chain. These groups are thus internally saturated and shielded from further interaction. In the extended conformation, the NHCO groups are free; in an aqueous medium, they have little choice but to form hydrogen bonds with surrounding water molecules. Let us say that each NHCO group effectively interacts with ?z water molecules. If we ignore all other components of the chain (i.e., side chains), whether this peptide exists in the helical or the extended form will depend on the total energy and entropy of the entire system. In the helical state the energy and entropy terms to be considered are: (1) The energy of a pair of peptide H-bonds; ( 2 ) the entropy of the helical peptide; ( 3 ) the total energy of the free water involved with each NHCO group; (4) the total entropy of the free water involved with each NHCO group. Of these four items (2), ( 3 ) and (4) are constant for all helical proteins; (1) on the other hand, varies with the nature of the protein molecule. In the extended state we must consider: ( 5 ) The energy of the peptide (water), complex; (6) the entropy of the peptide (water), complex. However ( 5 ) , (6), and the value of n are mutually dependent. Thus if the energy of one peptide(H20), complex is known, its entropy as well as 12 can be defined because the properties of water molecules are unchanging. In brief, whether the polypeptide exists in the helical state or the extended form essentially depends on the relative magnitudes of the energies of the peptide-peptide bonds and the peptide (water), bonds. If these bonds had the same energy in all proteins, of course, one would find them either all in the helical form or all in the extended form. This clearly is not the case. The usual explanation offered for the variability in the conformation of different proteins is the interaction among the side chains and the disruptive influence of proline and hydroxyproline on the formation of the a-helix. Among the side-chain tertiary interactions are: ( 1) disulfide (S-S) formation; (2) hydrophobic bonds between nonpolar side chains; ( 3 ) ionic bonds between charged groups (salt linkages) ; (4) hydrogen bonds; ( 5 ) electrostatic attraction between oppositely charged side chains; (6) electrostatic repulsion between similarly charged side chains.

A NEW MODEL FOR THE LIVING CELL

41

Of these all except (6) and sometimes ( 5 ) favor the formation of the ahelical structure. The electrostatic repulsion can be effectively eliminated if the protein is at its isoelectric point (IEP). Thus if the conventional interpretation is entirely correct, a comparison of the helical stability at the IEP of different proteins should show that those with a large array of functional groups form the strongest helix. Those that are not capable of forming tertiary structure would form the least stable helix. An examination of the properties of a special poly-L-alanine polymer will show that this is not at all the case. 5 . Evideiice fov the Direct Inductive Inflzteizce o/

Streiigth of the N H

*

*

the Side Chain on the

- OC Boud in the a-Hciiccll Structure

It is well known that the majority of proteins contain a large number of side chains capable of forming helix-stabilizing tertiary structure. Yet of these proteins some, oxidized ribonuclease, for example, do not form the helical structure in an aqueous medium (Harrington and Schellman, 1956). Many others that do form a helical structure lose it in the presence of 8 M urea, 5 M quanidineHCl, or 0.1 M dodecyl sodium sulfate. Thus if tertiary interactions are the only factors in stabilizing the a-helix, one would expect that a polypeptide that cannot form any helix-stabilizing tertiary structure would be entirely in the extended conformation. The findings of Doty and Gratzer proved otherwise (Doty and Gratzer, 1962). These authors found that a poly-L-alanine polymer, made water soluble by being connected to two block polymers of poly-D, Lglutamic acid on either end, exists entirely in the form of an a-helix. The stability of this helix is such that it resists all common denaturing agents including 10 M urea, 4 M guanidine-HC1, and 0.1 M dodecyl sodium sulfate. Yet, as Doty and Gratzer point out, this stability cannot be the result of sidechain interaction because the methyl side chains are too short to interact with the nearest neighboring methyl side chains. Since the polypeptide backbone of all proteins and polypeptides is the same, the unusual strength of the poly-Lalanine helix can only be the result of some attribute derived from the properties of the CH, side chain. W e are therefore compelled to find a new mechanism by means of which the CH, group can strengthen the helical structure. The methyl side chain is an electron-donating substituent (compare the pK values of HCOOH, 3.8; CH,COOH, 4.75; and CH,CH,COOH, 4.87). W e have shown earlier that an inductive effect can be transmitted from a side chain at least as far as the two NHCO groups flanking it. From this we must conclude that the proton-donating power of the NH group and the proton-accepting power of the CO group immediately adjacent to the side chain are acted on to different degrees by the inductive effect emanating from the side chain and that the consequence of this electron-donating side chain is to strengthen the helical NH-OC bond.

42

GILBERT

N. LING

Let us consider this problem in a greater detail

The methyl side chain releases electrons toward both the immediately neighboring C=O and NH groups thereby both strengthening the proton-accepting power of the CEO group and weakening the proton-donating power of the N H group. If these actions were exactly equal in magnitude, it would be hard to understand how the helix could be strengthened because the strength of the CO-HN bond must depend on the proton-accepting power of the CO and the proton-donating power of the NH group. T h e fact that the helix is strengthened has a dual implication: ( 1 ) the proton-accepting power of the CO group is increased more than the proton-donating power of the NH group is decreased and ( 2 ) the increase of proton-accepting power of the CO group increases the free energy of the helical CO-HN bond more than that of the CO-(H,O),, bond. l o n e might recall that in Section 111, A,3 we showed how a similar increase in the electron density (i.e., r-value increase) of the parent group of the CO group, the -COO- group, also has a differential effect, producing a greater increase in the free energy of association of N a + ion than in that of K + ion.] 6. The Genesal Model of Coopesatizle Adsorptioii .with Emphasis on cl i\.loleridar~ iklechanisni of Coiitrol riild Eiies
The first statistical-mechanical treatment of a cooperative transition was made by Bragg and Williams (Bragg and Williams, 1934) for the order-disorder transition of (3-brass with increasing temperature. More recently, it has been recognized that the denaturation o f proteins and D N A also represents a cooperative phenomenon (Schellman, 1955; Zimrn and Bragg, 1958; Gibbs and DiMargio, 1958). In all these cases, the nearest-neighbor interaction energy was considered primarily entropic in nature, referring to the increase in entropy as the native helical structure is broken. Our treatment of cooperative phenomena in living cells uses similar statistical-mechanical methods; it differs primarily in that the transition is no longer considered to be between an ordered state and a disordered state but between two alternate states of adsorption which can be ordered or disordered and that the nearest-neighbor interaction has an energy component as well as an entropic component. Before presenting our general model, it will be useful to define the /-value as a measure of the positive charge of a cationic group (NH: g r o u p ) . This parameter is analogous to the c-value devised to measure the electron density of an anionic oxyacid group. T h e proton-accepting power of the C=O group and

A N E W MODEL FOR T H E LIVING CELL

43

the proton-donating power of the NH group may also he represented by a cvalue analog and a c’-value analog, respectively. Let us now consider the model of a small segment of a protein chain containing a controlling site (cardinal site) shown in Fig. 24. T h e backbone amide groups have two choices of partners, either amide groups on an adjacent protein segment ( a + a - ) or free hydrogen-bonding molecules such as water (bt, b-). In the absence of an adsorbent at the cardinal site, the c-value analogs of the amide CO groups as well as the 1.’-value analogs of the amide NH groups all have the low value of, say, 1 (see inset). In this state, the electrons are fairly evenly distributed (as in the case of the series of insulators before the electrified rod was brought near, Fig. 23B), and the backbone groups prefer to interact with the fixed groups ;I+ and a- (upper diagram). If a cardinal adsorbent C is then introduced, it reacts with the cardinal site. An inductive effect then raises the c ’ - v h e analog of the nearest neighboring NH group from 1 to 2. At this c-value, this site no longer prefers a- but b- and an a- + h- exchange takes place. Because b- is more polarizing than a-, the replacement of a- by b- has the effect of withdrawing an electron from the neighboring CO group. T h e r-value analog of this group then rises from 1 to 2 leading to an a + + b + exchange. This process continues until all the a + and aare replaced by 13-t and h-. The result of the cardinal adsorption is to create in an all-or-none manner a series of water adsorption sites and (Ling, 1962) to cause an all-or-none dissociation from the adjacent protein and the assumption of a new discrete conformation. If the numerous hydrogen-bonding, ionic, and other sites on a protein molecule were independent, each site would have a large variety of choices in its partner. Such a protein could exist in a variety of conformations not sharply distinguishable from one another. In the present model there is site-to-site interaction between nearest neighbors such that the occupation of neighboring sites by the same adsorbent is favored. This autocooperative interaction leads to the existence of discrete molecular states (conformations) of the protein molecules (see the following section). It is the basic autocooperative nature of adsorption on proteins that makes possible the modulation and control of many sites by a small number of cardinal adsorbents. Such cardinal adsorbents may be hormones, drugs, ATP, and so forth. Their action and inaction constitute the basic step in energy and information transfer over long distances (Ling, 1962). 7.

A /More Sper.ific iModel; The Coopemtiz’e ALJsorptioii of K + aiid N a f lo71

Figure 2 5 shows a protein segment carrying anionic side chains existing in two alternative states in an aqueous environment containing both K + and N a + ion. In one state (B) the carboxyl groups have a relatively low charge density (low

44

GILBERT N. LING

c-value), and prefer K + ion over Na+ ion. The smaller separation of the positive charge of the K + ion from the anionic charge of the carboxyl oxygen makes the whole carboxyl-group-K+ -ion complex a weaker electron-donating source than an ionized carboxyl group by itself. Following the finding about p o l y - ~ alanine, we might anticipate a weakening of the helical structure. In consequence, the protein segment is found in the extended form. Water in the close vicinity of the chain exists in the polarized multilayer state. In the alternative state ( A ) , the anionic side chains are occupied by Na+ ion, which, as our previous calculations have shown (in the low c-value ranges considered here) tends to assume a configuration in which more water molecules separate the cation from the carboxyl group. The result is that the carboxylgroup-H,O-Na+-ion complex functions on the whole as a stronger electrondonating group (than the carboxyl-group-K+ion complex) thereby stabilizing the helical conformation. The surrounding water molecules in this case would be in a free state. The transition between the two states A and B is controlled by the cardinal adsorbent C. 8. T h e Eqnatioiz for Cooperative Adsorption on Proteim

An equation based on the method of the one-dimensional Ising model was published for cooperative adsorption on a protein chain by Yang and Ling in 1964 (see Ling, 1964a). In this model, similar sites on a long chain have similar properties and there is nearest-neighbor interaction among these sides. Each site has two choices of adsorbent. For the adsorption of K + ion in the presence of Na+ ion, the equation has the following form (see Ling, 1964a, for the general equation)

where

As one may recall, I f ] represents the concentration of the adsorbing sites. K & and K i are the intrinsic adsorption constants of the Na+ and K + ion, respecFIG. 21. Diagrammatic illustration of a cooperative transition induced by a cardinal adsorbent. The top figure shows the variation of the free energies of adsorption with changes of the r-value analog of the C=O group and the ?-value analog of the NH group. Figures below demonstrate the stepwise displacement of b+ and b- (NH and C=O groups on the lower polypeptide) by a-t- and a- in consequence of interaction with the cardinal adsorbent at the extreme left. T h e overall result of the cooperative transition is the dissociation of the two peptides (or the uncoiling of a helix).

A NEW MODEL FOR THE LIVING CELL

2

I

. b',

b-

a+,aL

I

c-and c

I

-

I

value analog

45

46

GILBERT N. LING

tively, and are related to the intrinsic standard free energies of adsorption A": and A P i 0 by the relation

- A I c z = RT In I Y ; . ~

(22)

AI;"" I< = RT In KE

(23)

-

The free energy of nearest-neighbor interaction is - y/2. It is equal to the change in free energy each time ;I new neighboring pair of dissimilar adsorbents is created. Thus in a series of three adjacent sites all adsorbing N a + ion ( N a f , N a + , N a + ) , if the middle site is allowed to change its N a + ion for a K t ion ( N a + , K + , N a + ) , the totd energy change is the difference in the intrinsic free -AP,") p h 2 ( - ~ / 2 ) or -y, because of the two new neighenergies >:I boring pairs of "I+, K-t created. O n the other hand, a shift from Na+, Nn+, K + to N a + , K + , K + involves only AFqo - AF"" since no new Na+, K + Si! neighboring pairs are created. An important feature of the equation for cooperative adsorption is that if log ( I K + ] n , l / [ N a + ] n d ) is plotted against log ( [K1,JINaJeS) (as in Fig. 28, for example) the tangent to the resulting curve through the locus at which [ K + ] 1Na+Inll= 1, is described by the following equation (Ling, l964a, 1965b, 1966) :

T h e slope 71 is related to the free energy of nearest-neighbor interaction by the explicit relation 11

= exp

(- h)

when I I = 1, - y/2 = 0 and there is no nearest-neighbor interaction. In this case, the adsorption is effectively noncooperative and indeed follows the wellknown Langmuir adsorption isotherm (Fig. 26). When 12 < 1, - y/2 > 0, and the nearest-neighbor interaction is such that dissimilar adsorbents on adjacent sites are favored (i.e., K, N a , K , N a , etc.), W e have referred to such interactions as heterocooperative (Ling, 1962, p. 1 0 2 ) . If 12 > 1, - y/2 > 0; this means that the nearest-neighbor interaction is such that similar adsorbents on adjacent sites are favored (e.g., K + , K-I-, K + , K + , o r N a + , N a + , N a + , N a + ) . Such

CARDINAL SITE7

COOPERA7 IVE TRANSITION

0

+ u

*

P,

B

FiG. 25. Diagram of a portion of a protein molecule undergoing autocooperative transformation. For simpllcitp. adsorbed water molecules in multilayers are shown as a single 1ayer.W-shaped symbol represents a cardinal adsorbent.

48

GILBERT N. LING

interactions are referred to as autocooperative. It is a major theme of the association-induction hypothesis that this type of cooperative adsorption underlies most, if not all, of the all-or-none phenomena known in cell physiology (Ling, 1962, 1964b). 10

0.8

Xp 0.6

04

0.2

0. I

0

2

4

6

FIG. 26. Cooperative adsorption isotherm for a one-dimensional chain. Linear plots of theoretically calculated isotherms. For - y / 2 = 2.30 kcal./moIe, the isotherm is autocooperative, showing a siginoid shape. For - y / 2 = 0.0 kcal./mole (the Langmuir adsorption isotherm), the curve resembles a hyperbola (Ling, 1 %6b, by permission of Federatiorr Proceedings) ,

9. The General Eqrttatioii for S o h t e Disfribiitioiz

jti

LiiJing Ce1l.c.

Equations (13) and (14) are special equations for K + - and Na+-ion distribution in the case in which there is no site-to-site interaction (i.e., adsorption follows the Langmuir adsorption isotherm). The following more general equation for intracellular K + ion is valid for the case in which the sites have nearestneighbor interaction as well: [K+]iii

= ayR [ K + l r x

+

2

L=l

[fl” -{I+ 2

5” - 1 1(5L-l)2+451,exp(yL/RT)]~

t

(26)

49

A NEW MODEL FOR THE LIVING CELL

where [ f l L , 5'. and y" refer to the Lth type of sites. The equation for the intracellular N a + ion is

Cql 2

5" - 1

-

I(lj'2-

l)2+4E1,exp

(yL/RT)]x (27)

10. Cooperative Adsorption iiz Biologirnl System; Experzmeatal Evideme

Two types of plots are useful in distinguishing cooperative adsorption isotherms from Langmuir adsorption isotherms. W e have already mentioned the use of log-log plot, i.e.,

1og l K + l a d versus log [K+ 1ex "a+ 111d lN"+lI'Y Although the model used in deriving this equation for cooperative adsorption was based on a chain of similar sites, it is often possible to analyze adsorption onto heterogeneous proteins by plotting the data on a log-log plot as if there were only one type of adsorption site. The result appears as a number of straightline segments joining each other at rather sharp angles (Ling, 1966b). For the combination of isotherms shown in Fig. 27, the outermost segments of the curve have slopes of unity. The two inner segments shift sharply from a slope of less than unity to a slope greater than unity. This shift marks the point at which adsorption changes from heterocooperative to autocooperative. The linear plot is perhaps more familiar to most biologists. W e have shown previously that on such a plot a Langmuir adsorption isotherm has the form of a hyperbola. A cooperative (auto-) curve on the other hand is sigmoid. Such curves are seen in the binding of oxygen onto hemoglobin (Wyman, 1964), in the effect of many feedback inhibitors on enzyme reactions (Monod et al., 1965), and in the adsorption of K + ion by living cells (Ling, 1966b). Let us discuss some of these examples in more detail. a. Oxygen Binding on Hemoglobirz. It has long been known that the curve for the binding of oxygen on hemoglobin is not hyperbolic but sigmoid and that this indicates interaction among the four heme groups onto which the oxygen adsorbs (Ling, 1964b, 196613; Wynian, 1964). A. V. Hill introduced an empirical equation to describe this sigmoid ddsorption:

'

log __- = tz log PO2 1 -y

+ n log K ,

(28)

where y and PO, represent the number of adsorbed oxygen molecules per hemo-

50

GILBERT N. LING

globin molecule and the partial pressure of oxygen, respectively, and K, is a constant (Hill, 1910). It has been known that n is some sort of measure of the degree of interaction. This equation is analogous to Eq. (24). Therefore, the Hill coefficient n is related to the free energy of nearest-neighbor interaction according to Eq. (25). The reaction of oxygen and hemoglobin takes place not in a vacuum but in a water solution. Thus K , of Eq. (28) is the intrinsic adsorption constant of oxygen divided by the intrinsic adsorption constant of the alternative adsorbent, namely, water.

lo4

10 Xi -

xi

I

10-1

FIG. 27. Complex cooperative adsorption isotherms. Adsorption isotherms for two linear polymers, one autocooperative ( 8 = 100) and the other heterocooperative ( 0 = & I ) , are shown, respectively, in curves 111 and 11. Circles represent a complex adsorption isotherm for a mixture containing equal amounts of the above linear polymers and treated as though it contained only a single polymer. Curve I has been drawn as a series of straight lines showing the significant parts of the more exact curve that would join all the circles. Note that the two outermost straight lines have slopes of unity; the slopes of the two middle lines have inexact but significant values which reveal the hetero- or autocooperative nature of the component systems (Ling, 1966b, by permission of Federation Proceedings).

A N E W MODEL FOR THE LIVING CELL

51

If we take account of the fact that there are four heme groups per hemoglobin molecule, the complete equation for oxygen adsorption on hemoglobin becomes : PO2

x K, - 1 (29)

In Fig. 28, the data of Lyster (see Rossi-Fanelli et al., 1964) for oxygen

p 0 2 ( rnm Hg )

FIG. 28. A log-log plot of the data of Lyster on oxygen uptake by human hemoglobin a t pH 7.0 at 19°C. Data of Lyster as presented in the review of Rossi-Fanelli et al. (1964). Points are experimental; the line is theoretical according to Eq. (29) with K = 5.88 x 10-6 M and - y / 2 = 0.67 kcal./mole.

binding on hemoglobin have been replotted on a log-log scale. The line is theoretically calculated according to Eq. (29) with K , equal to 5.88 x 10Wg M and - y/2 equal to 0.67 kcal./mole. In 1965 Monod, Wyman and Changeux (Monod et a/., 1965) presented a model for “allosteric transitions” which, superficially at least, bears some resemblance to our model of cooperative adsorption (see also Haber and Koshland, 1967). A short discussion of their model may help to prevent future confusion. Basically, Monod and his co-workers make the following presuppositions: (1) A protein molecule that is capable of undergoing allosteric transitions contains symmetrical subunits with which ligands react in a symmetrical fashion. ( 2 ) This

52

GILBERT N. LING

protein molecule is capable of existing in at least two discrete states which are in equilibrium with each other. ( 3 ) The affinity of the protein molecule for the ligand differs in the two states. ( 4 ) The binding of any one ligand molecule is independent of the binding of any other. On the basis of this model, Monod and his co-workers are able to predict the types of sigmoid curves seen in allosteric interactions in enzymes (Monod et al., 1965) as well as the curve of Lyster for the binding of oxygen to hemoglobin (see also Changeux et al., 1967). The model of Monod, Wyman, and Changeux differs from the model presented above in the following ways: (1) The existence of a small number of discrete states of the protein molecule is an a priori assumption; the reason for the existence of a small number of discrete states rather than a large array of states differing continuously is not made clear. W e have pointed out above that it is the nearest-neighbor interaction that gives the protein its capability of assuming such discrete states. Monod et al. do not include such nearest-neighbor interaction as a part of their model. ( 2 ) Monod et al. refer to “cooperative effects” by which they apparently mean interactions that enhance the binding of substrate molecules; such interactions are thus akin to the autocooperative interactions described above. However, the mechanism they postulate for these effects is not the type of cooperative interaction discussed above, for again nearest-neighbor interactions-an integral part of classic cooperative interactions-are not part of their model. ( 3 ) The Hill coefficient 12 in the equation for the binding of oxygen to hemoglobin is not related to the parameters ( L and c) used by Monod et al. to describe their oxygen-binding curves. (4) In the Monod-Wyman-Changeux model, the change in protein conformation and the binding of single ligands are considered separate events ( L and c are independent parameters). The binding of the ligand is not considered to bring about the conformation change. In the model presented above, the binding of the ligand and the change of the protein conformation are interrelated processes not only because of the role of nearest-neighbor interaction but also because the ligands compete for the same sites that maintain the protein conformation. ( 5 ) The Monod-Changeux-Wyman model is only applicable to symmetrical molecules; our model can be applied to all proteins (for experimental data on cooperative adsorption on denatured proteins, see Ling, 1966b).

b. Detergent Biizdilr’g 012 Bavitze Serzlm Albi~ziiz.There is a fairly large collection of experimental data on i72 vitra adsorption on proteins that are not consistent with the Langmuir adsorption isotherm, but which can be explained on the basis of cooperative adsorption (Ling, 1962). As an example, in Fig. 29 we have replotted the data of Pollansch and Briggs on the adsorption of the

A NEW MODEL FOR T H E LIVING CELL

53

detergent, dodecyl sulfate, on bovine serum albumin. The theory predicts that the point of abrupt change to a higher slope, indicating the beginning of an autocooperative adsorption, should coincide with the beginning of a conformational change. This is indeed the case: At this point the electrophoretic boundary splits from one to two (Pallansch and Briggs, 1954) (for other examples see Ling, 1962).

Double electrophoretic boundary

XDodS ‘-xDod S

.t

lo-’

/

/

Sinale electroDhoretic boundorv

10” 10-5

10-4

Dodecyl sulfate Concentration (moles/litw)

FIG. 29. Adsorption of dodecyl sulfate on bovine serum albumin. Ordinate represents, on a logarithmic scale, the mole fraction of sites binding the anionic detergent, dodecyl sulfate, divided by the mole fraction of sites not binding the detergent. Abscissa represents the dodecyl sulfate concentration also on a logarithmic scale. Total number of binding sites, 102, is the sum of the number of arginin, lysine, and histidine residues per protein molecule. Note that the change from a single to a double electrophoretic boundary (dotted line) occurs at a concentration of dodecyl sulfate corresponding to the abrupt shift of slope from heterocooperative to autocooperative adsorption of detergent. Open and half-filled circles represent two different series of experiments (Pallansch and Briggs, 1954, by permission of The ]oulnal of Amerii-an Chernjcal S o l - i e l y ) ,

c. Cooperative Adsorption of K + l o n in Livjiig Cells. i. Frog muscles. It requires 3-4 days of incubation at room temperature for frog muscles to attain new equilibrium ionic concentrations when the external ion concentrations are varied. In our laboratory we have succeeded, by utilizing tissue culture techniques, in maintaining frog muscle in normal condition in vitro for 7 days or more, more than long enough to attain these new levels. In Fig. 30 we show the results of an experiment in which the external K+-ion concentration was varied while the external Na+ -ion concentration remained constant. The equilibrium internal K+-ion concentrations attained are plotted as a function of the external K+-ion concentration on a linear plot. The sigmoid curve obtained can be compared with the similar curve obtained for the binding of oxygen by hemoglobin

54

GILBERT N . L I N G

(Fig. 28). The experimental data have been fitted with a theoretical cooperative adsorption curve [Eq. (20)] calculated using a value of 665 M-l for K i / K i a and of 0.76 kcal./mole for -y/2. This means that the adsorption of K + is

[K+Iex (rnrnole/liter) FIG. 30. Equilibrium K+-ion concentration in frog sartorius muscle in solutions with low K+-ion concentrations but a high Na+-ion Concentration. Sterily isolated sartorius muscles were shaken for 72 hours at 25°C. in Ringer solutions containing a fixed concentration (100 mmole/liter) of N a + ion and varying K+-ion concentrations. K+ and N a + ion were analyzed by flame photometry on HCI extracts of the muscles. Total intracellular ionic Concentration was obtained from raw analytical data after correcting for extracellular space ( 10%). Adsorbed ionic concentration in millimoles per kilogram of fresh tissue was further computed from the total intracellular concentration by subtracting the interstitial ion concentration (estimated as 10.4% of the equilibrium external ion concentration; this figure represents an average of all values determined to this point). Each point represents a single determination. Inset shows oxygen uptake by human erythrocytes (broken line with filled circles) and by myoglobin (solid line) (from Eastman et at., 1933) (Ling, 196613, by permission of Federation Proceedings).

made more favorable by 2 x 0.76 = 1.52 kcal./mole if the two flanking sites are occupied by K + ion rather than by Na+ ion. ii. Mammalian smooth mzlscles. Jones has demonstrated that the steady level of K+-ion uptake in dog arterial smooth muscles also follows Eq. (20) (confirmed by Gulati, to be published). The K i / K i a and - y/2 values are, respectively, 93 M - 1 and 0.61 kcal./mole (Jones and Karreman, in press). iii. Escberichiu coli. Damadian has shown that a sigmoid curve results when the steady level of K + ion accumulated in an E. coli mutant (RD-2) is plotted against the external K+-ion concentration in the range of 0-0.3 mM external K + ion. At higher concentrations there appear to be a second set of cooperative adsorption sites (Damadian, 1968).

55

A NEW MODEL FOR THE LIVING CELI.

11.

The Coiitsol rind Enes@atinii o f Cooper.crtitie Ad.sosptioFi

a. The Coiitsol of K-Nci Adsosptioii bq' Castliac Glycosides. The theoretical model shown in Fig. 2 predicts that in the same ionic environment the interaction of the protein with a cardinal adsorbent may shift the system from a relative affinity for K + ion to a relative afinity for Na+ ion in all-or-none manner. Figure 31 shows that the cardiac glycoside lanoxin, at pharmacological concentrations, produces a shift in the cooperative adsorption isotherm such that the ratio of the intrinsic equilibrium constants (KO / K O ) shifts toward a lower K Nn K+-ion preference and a higher Na+-ion preference by a factor of 20. The autocooperative nature of the curve (slope > 1 at [K+].,/[Na+Isd = I ) is preserved. This effect has profound physiological significance. The siginoid curve shown in Fig. 30 shows that the muscle can change in a more or less all-or-none manner from adsorbing all K + ion to adsorbing all Na+ ions. However, under physiological conditions, there are no large changes in the plasma concentrations of K+

Q/

1

I& 10-3

10-2

lo-'

FIG.31. Log-log plot of the K + -and Naf-ion distribution in frog sartorius muscles in the absence (left) and presence of lanoxin. The experimental procedures used were similar to those described in Fig. 30. The lanoxin concentration was 2.5 pg./ml. X, and X ,, refer to mole fraction of adsorbed K f and Na+ ion, respectively.

56

GILBERT N. LING

or N a + ions. Under these conditions, changes in adsorbents must be brought about by another agent, preferably one that is active in very small quantities and can influence the adsorbents on a large number of sites by itself. In the present experiment lanoxin is such an agent. Thus in an unvarying ionic environment, interaction with lanoxin changes the site from a state in which K+ ion is the main adsorbent to one in which Na+ ion is the main adsorbent. A similarly controlled all-or-none shift of the K + and Na+-ion preference of the anionic sites on the surface of excitable tissues has been discussed at some length in Section 111, C. b. Etzergization of the Biological Activaiors. In the history of biology, one of the most admirable events in the 1930’s was the way the leaders in the field of metabolism, A. V. Hill and 0. Meyerhof, responded to the findings of the then relatively unknown Lundsgaard. In contradiction to the lactic acid theory of muscle contraction proposed by Hill and Meyerhof, Lundsgaard found that muscle could contract in the complete absence of lactic acid production (Lundsgaard, 1930; Henriques and Lundsgaard, 1931). The immediate confirmation of Lundsgaard’s finding in Meyerhof’s laboratory led Hill to write an article, entitled “A Revolution in Muscle Physiology” (Hill, 1932) which paved the way for the recognition of the important role of ATP in cell function. Subsequent development of the concept of the high-energy phosphate bond led to the postulation that the enzymic hydrolysis of this bond liberates energy for the performance of biological work (Lipmann, 1941). However, the demonstration that the enthalpy (see footnote 3) of the high-energy phosphate bond is not 12 kcal./mole as it was one time thought to be (for a possible source of this error, see Ling, 1962), but only 4.7 kcal./mole (no higher than an ordinary phosphate bond, Podolsky and Kitzinger, 1955; Betzinger and Morales, 1956), has left this theory of the energization of biological work untenable. In terms of the association-induction hypothesis, ATP is conceived to perform its energization function, not through hydrolytic cleavage, but through adsorption on cardinal sites producing or maintaining a particular cooperative state of the cellular protein (Ling, 1962). The unusually high enthalpy of the adsorption of ATP on G-actin [-24 kcal./mole (Asakura, 1961) J indicates that energization could be accomplished in this manner. The effect of ATP on K+-ion adsorption would be essentially similar to the demonstrated effect of lanoxin but in the reverse direction. A prediction of this model is that the K+-ion concentration in living cells is not dependent on the rate of ATP hydrolysis, but instead is determined by the concentration of ATP per se in the cell. Thus each time an ATP molecule adsorbs on a cardinal site, a fixed number of anionic sites (see Fig. 25) will cooperatively adsorb K + ion. Without ATP as the cardinal adsorbent, some other ionic components will occupy the anionic sites; these may be Na+ ion or fixed cations on nearby proteins. This prediction has been confirmed for K+-ion

A NEW MODEL FOR THE LIVING CELL

57

distribution in frog muscle treated with iodoacetate and nitrogen, in human erythocytes, and in E. roli (Ling, 1962, Chapt. 9 ) . Figure 32 shows the K+-ion concentration in frog muscles as a function of the cellular ATP concentration. Once the dependence of K+-ion accumulation on ATP adsorption is realized, 100

[ATPI in (pmoles/gm)

FIG. 3 2 . The correlation of intracellular K+-ion concentration and ATP concentration in frog voluntary muscles. Muscles were treated with /i m M iodoacetic acid for varying lengths of time at room temperature, then chilled in the same bathing solution to 0°C. and allowed to equilibrate at this lower temperature for 1 hour, after which they were analyzed for both their K + ion and ATP contents (Ling, 1962, by permission of Blaisdell Press).

the physiological role of ATP once more becomes understandable. Thus the normal resting cell hydrolyzes ATP slowly, but the activated cell hydrolyzes ATP very rapidly. (This indicates that the ATPase activity itself is under physiological control.) A cyclic event can be visualized as follows: ATP Adsorption on enzymically

/ t

Nonac tive cardinal s i t e s

\

ATP Synthesis

Cooperative K+-ion adsorption

/

Cooperative K+-ion desorption

\ ATP Hydrolysis

t

Stimulus activating ATPase

58

GILBERT N. LING

ACKNOWLEDGMENTS The preparation of this review and the new investigations reported were supported by the National Science Foundation Research Grants GB3921, GB7095, the National Institute of Health Research Grants 2RO1-GM11422-04 and HE-07762-64, and the Office of Naval Research Grant Nonr 4371 (00)-105327. The author is supported by Public Health Service Resrarch Career Development Award K3-GM-19032. The author thanks Dr. Frank Elliott, Dr. Margaret C. Neville, Margaret M. Ochsenfeld. Grace Bohr, and Marie Bowers for their invaluable help; and the John A. Hartford Foundation for providing the basic equipment for the investigations.

REFERENCES Asakura, S. (1961). Arch. Biorhenz. Bioph.7.r. 92, 140. Astbury, W . T. (1951). Sri. Am. 184, 21. Avery, 0.T., MacLeod, C. M.. and McCarty, M. (1944). J. Exptl. Med. 79, 137. Bennett, H. S. (1956). J . Biophys. Biorhem. Cytol. 2, Suppl., 99. Bernstein, J. ( 1902). Arrh. Ges. PhyJiot. PfluegerJ 92, 521. Betzinger, R. J., and Moralrs, M. F. (1956). J. Biol. Chem. 218, 945. Beutner, R. (1920). “Die l?ntstehung electrischer Strome in lebenden Geweben und ihre Kiinstliche Nachahmung durch synthetische organische Substanzen.” Enke, Stuttgart. Boyle, P. J., and Conway, E. J, ( 1941). J. Physiol. (London) 100, 1. Bradley, S. (1936). J. Chem. Sor. p. 1799. Bragg, W. L., and Williams, E. J. (1934). Pror. Roy. Soc. (London) A145, 699. Bratton, C. B., Hodgkin, A. L., and Weinberg, J. W. (1965). Science 147, 738. Bregman, J. I. (1953). A n n . N . Y . Acad. Sci. 57, 125. Butschli, 0.(1894). “Investigations on Microscopic Foams and on Protoplasm” (E. A. Minchin, transl.) . Black, London. Bull, H. (1944). J. Am. Chem. Sac. 66, 1499. Chambers, R., and Hale, H. P. (1932). Pror. Roy. Soc. (London) B110, 336. Chambers, R., and Kao, C. Y. (1952). Exptl. Cell Res. 3, 564. Changeux, J., Thiery, J., Tung, Y . , and Kittel, C. (1967). Proc. Natl. Acud. Sri. US.57, 335. Chapman, G., and McLauchlan, K. A. (1967). Natuve 215, 391. Cole, K.S. (1932). J. Cellular Comp. Pbysiol. 1, 1. Collander, R. (1949). Pbysiol. Plantavum 2, 300. Collander, R.,and Barlund, H. (1933). Acta Botan. Fennicae 11, 1. Conway, E. J. (1946). Nature 157, 715. Cope, F. (1967). J. Gen. Physiol. 50, 1353. Cowan, S. L. (1934). Proc. Roy. Sor. (London) B115, 216. Curtis, H.J., and Cole, K. S. (1942). J. Cellular Comp. Pbysiol. 19, 135. Damadian, R. ( 1968). J. Bacteriol. 95, 113. Davson, H., and Danielli, J. F. (1952). “The Permeability of Natural Membranes,” 2nd Ed. Cambridge Univ. Press, London and New York. Dawson, D. M. (1966). Biol-him. Biophys. Acta 113, 144. Dean, R. B. (1941). Biol. Symp. 3, 331. de Boer, J. H., and Zwikker, C. (1929). 2. Phyrik. Chem. (Leipzig) B3, 407. D e Vries, H. (1885). Juhrb. W‘iJ.r, Botan. 16, 46.5.

A N E W MODEL FOR T H E LIVING CELL

59

Doty, P., and Gratzer, W. B. (1962). I n “Polyamino Acids, Polypeptides and Proteins” (M. A. Stahmann, Ed.), p. 111. Univ. of Wisconsin Press, Madison, Wisconsin. Eastman, N. J., Geiling, E. M. K., DeLander, A . M. (1933). Bull. Johns Hopkins Hosp. 53, 246. Eisenman, G . (1961). Membrane Transport Metab., Proc. Symp., Prague ( A . Kleinzeller and A. Kotyk, eds.), p. 163. Czechoslovak Academy of Science, Prague. Eisenman, G., Rudin, D. O., and Casby, J. U. (1957). Science 126, 831. Epstein, E., and Hagen, C . E. (1952). Plant. Physiol. 27, 457. Falk, G., and Gerard, R. W. (1954). J . Cellular Comp. Physiol. 43, 393. Fenichel, I. R., and Horowitz, S. B. (1963). Acta Physiol. Stand. 222, 1. Fischer, H., and Moore, G. (1908). A m . J . Physiol. 20, 330. Fritz, 0. G., and Swift, T. J. (1967). Biophys. J . 7, 675. Gibbs, J. H., and DiMargio, E. A. (1958). J . Chem. Phys. 28, 1247. Gordy, W., and Stanford, S. C . (1941). J . Cbem. Phyi. 9, 204. Grundfest, H., Kao, C. Y., and Altamirano, M. (1945). J . Gen. Physiol. 38, 245. Haber, J. E., and Koshland, D. E. (1967). Proc. Natl. Acud. Sci. 58, 2087. Hallett, J. ( 1965). Federation Proc. 24, S-34. Hamburger, H . J. (1889). 2. B i d . 26, 414. Hammett, L. P. ( 1940). “Physical Organic Chemistry.” McGraw-Hill, New York. Harkins, W. D. (1945). Scietice 102, 292. Harrington, W. F., and Schellman, J. A. (1956). Compt. Rend. Trau. Lab. Carlsberg, Ser. Chim. 30, No. 6, 21. Harris, E. J. (1950). Trans. Faraday Sor. 334, 872. Harris, E. J., and Burn, G . P. (1949). Tram. Faraday Sor. 45, 508. Heckmann, K. (1953). Naturwissenschaften 40, 478. Henriques, V., and Lundsgaard, E. (1931). Biochem. 2. 236, 219. Heppel, L. A. (1940). A m . J. Physiol. 128, 449. Hermann, L. (1879). “Handbuch der Physiologie,” Vol. 2 (F. C. W . Vogel, ed.), p. 3. L. Hermann, Leipzig. Hill, A. V. (1910). J. Physiol. (London) 40, iv-vii. Hill, A. V. (1930). Proc. Roy. Sac. (London) B106, 477. Hill, A. V. (1932). Physiol. Rev. 12, 56. Hinke, J. A. M. (1959). Nature 184, 1257. Hodgkin, A. L. (1951). B i d . Rev. Cambridge Phil. Sor. 26, 339. Hodgkin, A. L., and Horowicz, P. (1959). J. Physiol. (London) 148, 127. Hodgkin, A. L., and Katz, R. D . (1949). J . Physiol. (London) 108, 37. Hodgkin, A. L., and Keynes, R. D . (1953). J . Physiol. (Londotr) 119, 513. Holter, H., and Holtzer, H. (1959). Exptl. Cell Res. 18, 421. Holwill, M. E. J., and Burge, R. E. (1963). Arch. Biochem. Biophys. 101, 249. Horowicz, P., and Hodgkin, A. (1957). J . Pbysiol. (London) 145, 405. Huxley, H. E. (1957). J. Bi0phy.r. Biorhem. Cylol. 3, 631. Jacques, M. (1936). J. Gen. Physiol. 19, 397. Jones, A., and Karreman, G. ( 1969). BiophyJ. J . (in press). Keynes, R. D., and Maisel, G. W. (1945). Proc. Roy. Soi. ( L o d o n ) B142, 383. Koketsu, K., and Kimura, Y. (1960). J . Cellular Comp. Physiol. 55, 219. Krogh, A. (1946). Proc. Roy. Soc. (London) B133, 140. Langmuir, I. (1917). J. A m . Chem. Soc. 39, 1848. Levi, H., and Ussing, H . H . (1948). Acta Physiol. Srand. 16, 232. Lewis, M. S., and Saroff, H. A. (1957). J. Am. Chem. Sor. 79, 2112.

60

GILBERT N. LING

Ling, G. N . (1951). A m . J. Physiol. 167, 806. Ling, G. N. (1952). In “Phosphorus Metabolism” ( W . D. McElroy and B. Glass, eds.), Vol. 2, p. 748. Johns Hopkins Press, Baltimore, Maryland. Ling, G. N. (1955). J. Phys. Med. 34, 89. Ling, G. N. (1960). J. Gen. Physiol. 43, Suppl., 149. Ling, G. N. (1962). “A Physical Theory of the Living State.” Ginn (Blaisdell), Boston, Massachusetts. Ling, G. N. (1964a). Biopolymers, Symp. 1, 91. Ling, G. N. (l964b). Texas Rept. Biol. Med. 22, 244. Ling, G. N . (196Sa). Ann. N . Y . Arad. Sri. 125, 401. Ling, G. N . (1965b). Federation Pror. 24, Suppl. 15, S-103. Ling, G. N . ( 1 9 6 5 ~ )Perspectives . Biol. Med. 9, 87. Ling, G. N. (1966a). Ann. N.Y. Acad. Sc-i. 137, 837. Ling, G. N . (1966b). Federation Pror. 25, 958. Ling, G. N. (1967a). In “Glass Electrodes for Hydrogen and Other Cations” (G. Eisenman, ed.). Dekker, New York. Ling, G. N. (1967b). Naturzo. Rundschau 20, 415. Ling, G. N. ( 1 9 6 7 ~ )In . “Thermobiology” (A. Rose, ed.), Chapt. 2. Academic Press, New York. Ling, G. N., and Ochsenfeld, M. (1965). Biophys. J. 5, 777. Ling, G . N., and Ochsenfeld, M. M. (1966). J. Gen. Physiol. 49, 819. Ling, G. N., Ochsenfeld, M. M., and Karreman, G. (1967). J. Gen. Physiol. 50, 1807. Ling. G. N., and Ochsenfeld, M. M. (196th). Federation Proc. 27, 702. Ling, G. N., and Ochsenfeld, M. M. (1968b). Prof-. Intern. Physiol. Cungr. Us’aihittgton D.C., 24, 266. Ling, G . N., and Woodbury, J. W . (1949). J. Cellular Comp. Physiol. 34, 407. Lipmann, F. (1941). Advan. Enzymol. 1, 99. Lundsgaard, E. (1930). Biochem. 2. 227, 51. McBain, J. W., and Peaker, C. R. (1930). 1.Phys. Chern. 34,1033. McDonald, J. S. (1900). Pror. Roy. Sor. (London) 67, 310. McLaren, A. D., Jensen, W. A., and Jacobson, L. (1960). Plant. Phy.rio1. 35, 549. Mellon, S. R., and Hoover, E. F. (1950). J. A m . Chem. Sor. 72, 2562. Mond, R., and Netter, H. (1930). Pfiuger’s Arch. 224, 702. Monod, J., Wyman, J., and Changeux, J. (1965). J. Mol. Biol. 12, 88. Mysels, K. J.. and McBain, J. W. (1948). 1. Colloid Sri. 3, 41. Nielsen, J. M., Adamson, A. W., and Cobble, J. W. (1952). J. Am. Chem. Sor. 74, 446. Osterhout, W . J. V. (1936). Barteriol. Rev. 2, 283. Overton, E. (1899). Vierteljuhrsschr. Naturforsrh. GeJ. (Ziirich) 44,88. Overton, E. (1907). In “Handbuch der Physiologie des Menschen” (W. Nagel, ed.), Vol. 2, p. 744. Vieweg, Braunschweig. Pallansch, M. J., and Briggs, D. R. (1954). J. A m . Chem. Soc. 76, 1396. Pfeffer, W . (1921 ) . “Osmotische Untersuchungen.” 2nd Ed. Engelmann, Leipzig. Podolsky, R. J., and Kitzinger, C . (1955). Federation Pror.. 14, 115. Rapatz, G., and Luyet, B. J. (1958). Biodynamira 8, 121. Rossi-Fanelli, A,, Antonini, E., and Caputo, A. (1964). Advarr. Protein Chem. 19, 73. Rotunno, C. A,, Kowalewski, V., and Cereijido, M. (1967). Biorhim. Biophys. Arta 135, 170. Ruhland, W., and Hoffman, C. (1925). Plantu 1, 1. Ryser, H. J. P. (1968). Science 159, 390.

A NEW MODEL FOR THE LIVING C E L L

61

Schellnian, J. A. (1955). Compt. Rend. Truv. Lab. Carlsberg, Ser. Chim. 29, 15. Schwindewolf, U. ( 1953). Nuturwissenschuften 40, 435. Starr, M. P., and Williams, R. C. (1952). J . Barteriol. 63, 701. Steinbach, H. B. (1940). J. Biol. Chem. 133, 695. Taft, R. W . (1960). J. Phys. Chem. 64, 1805. Taft, R. W.. and Lewis, I. C . (1958). J. A m . Chetn. Soc. 80, 2436. Tobias, J. M. (1950). J . CeNular Cotnp. Physiol. 36, 1. Troschin, A. S. (1958). “Das Problem der Zellenpermeabilitat.” Fischer, Jena. Weibull, C. (1960). In “The Bacteria” (1. C. Gunsalus and R. Y . Stanier, eds.), Vol. 1, Chapt. 4. Academic Press, New York. Wyman, J. (1964). Advan. Protein Chem. 19, 223. Zierler, K. (1958). Ann. N . Y . Arud. Sri. 75, 227. Zimm, B. H., and Bragg, J. K. (1958). J. Chem. Phyw. 28, 1246.

This Page Intentionally Left Blank

The Cell Periphery LEONARDWEISS' Department

of

Experimental Pcrthology, RoJiuell Park Memorial Institute, Brrfalo, N e w YorR

I. Introduction.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. Lipid Bilayers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111. Other Models . . . . . . . . . . . . . . . . . . . . . . . . . .... IV. Cell Surface Charge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Measurement of Cellular Electrophoretic Mobility . . . . B. Sialic Acid Moieties . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Ribonuclease-Sus~eptible Groups . . . . . . . . . . . . . . . . . . D. Amino Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E. Other Peripheral Ionopmic Groups . . . . . . . . . . . . . . . . F. Dynamic Aspects of Surface Charge . . . . . . . . . . . . . . . . G . Charge Distribution . . . . . . . . . . . . . . . V. Enzyme Activity and the Cell Periphery . . . A. Sublethal Autolysis . . . . . . . . . .

63 64 70 78 78 81 82 85

86 87

....

VI. The Peripheries of Malignant Cells . . . . . . . . . . . . . . . . . . A. Fine Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Calcium Binding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Surface Charge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

94 95

96 97 99

I. Introduction The structure of the cell periphery is of interest to workers in a number of apparently diverse fields, and a complete discussion of the subject should ideally present an integrated picture. This is not possible at the moment, and any review will necessarily be weighted in the direction of the reviewer's own interests. My own work is concerned with interactions between living cells in cancer and morphogenesis, and involves the biophysical analysis of cellular contact and recognitive phenomena (Weiss, 1967a). In these studies, it is useful to discriminate physically between cell contact, cell adhesion, and cell separation (Weiss 19621, 1967b). This approach requires a distinction to be made between the cell surface, which approximates to a two-dimensional planar structure surrounding the cell and in contact with its environment, and the ceIl peripheral zone, which is a three-dimensional region including the plaimu membrdne or permeability barrier(s). All of these defined regions must be considered in dynamic terms. 1 Some of my own work described here was partially supported by Grant No. P-403A from the American Cancer Society.

63

64

LEONARD WElSS

No attempt will be made to review transport phenomena, or the vast amount of immunochemical data that are clearly fundamental to presentation of any integrated view of the structure of the cell peripheral region. In these and other respects, this review is incomplete. In this field, some of the techniques may be unfamiliar to the general reader. As the techniques themselves influence the interpretation of experimental data, I have on occasion used the work of my colleagues and myself extensively to illustrate interpretative difficulties, as I am most familiar with our own techniques. This should not be taken to imply that I am either unacquainted with, or wish to disregard the work of others. 11. Lipid Bilayers

Gorter and Grendel ( 1725, 1726) observed that when acetone-extracted erythrocyte lipids were spread at the air/water interface of a Langmuir trough, the area of the compressed film was twice that of the calculated surface area of the erythrocytes. On this basis it was suggested that the erythroycte membrane consisted of a lipid bilayer (Fig. 1). It has since been noted, by Winkler and

FIG. 1. The Gorter-Grendel model

Bungenberg de Jong (1741) and Hoffman (1762), that Gorter and Grendel underestimated the surface area of the erythrocytes by about 50%), and that the ratio of total lipid surface area to cell surface area is nearer 1 :1 than 2:1; however, as acetone does not extract all of the lipids from erythrocytes, the original bilayer estimute is probably correct. Davson (1962) has quoted more reliable data derived from lipids extracted from rabbit, guinea pig, and human erythrocytes, in which the ratios of film areas to the computed areas of the cells are all approximately 2:1. A modified lipid bilayer hypothesis was also advanced independently by Danielli and Davson (1735). Danielli and Harvey ( 1 9 3 4 ) observed that the tensions measured at the peripheries of a number of oil droplets and cells corresponded to maximum estimates for their surface tensions of 1-3 dynes per cm. The surface tensions of lipid films was expected to be of the order of 10-30 dynes per cm., and experiments had shown that various proteins could markedly reduce the surface tension of lipids. The now well-known model was

THE CELL PERIPHERY

65

therefore advanced for the plasma membrane, in which a lipid bilayer is coated on both its inner and outer surfaces with protein. The early and later versions of this model are shown in Figs. 2 and 3. It is not my purpose here to review in detail the copious literature relating to the lipid bilayer model for the cell membrane. It is difficult to resist a discussion Hvdrocarbon

FIG. 2 .

Polor groups

The Danielli-Davsorl model [After Danielli and Davson (1935)l.

of the fascinating experiments made on black lipid films, originating with the work of Mueller and his colleagues (Mueller et ul., 1962; MueIler and Rudin, 1963) and recently reviewed by Tien and Diana (1968). However, while these films mimic certain natural membrane functions to a remarkable degree, particularly in respect to the action potentials seen in the nervous system, their role in elucidating the structure of cell membranes in general is not clear at the moment. By the same token, no discussion will be given of the lipid vesicles studied by Bangham and his colleagues (Bangham et ul., 1965; Bangham and Haydon, 1968). If it becomes technically possible to incorporate membrane proteins and carbohydrates in these lipid systems, or to make them from lipoglycoproteins, then possibly other membrane models, apart from the lipid bilayer, can be evaluated. Many of the other relevant papers on model systems have been discussed by Kavanau (1965) and myself (Weiss, 1967a), among others. In 1962, Danielli stated that “The problem of the basic structure of the plasma membrane was essentiaIly solved by about 1940. At that time there was

66

LEONARD WEISS

adequate evidence that the membrane was a bimolecular lipoid leaflet with adsorbed protein layers on both surfaces . . . .” As this model has dominated so much of the thought and work on membrane structure fur nearly 30 years, it is pertinent to examine critically the bases on which it depends in order to determine whether it is hypothesis or fact. Lipoid molecule

Protein molecule

Polar pore

FIG. 3. The Danielli-Davson-Harvey model, modified by the addition of polar pores (Danielli, 1958). [Reprinted with permission from “Surface Phenomena in Chemistry and Biology“ (J. F. Danielli, K. G. A. Pankhurst, and A. C. Riddiford, eds.), Pergamon Press Ltd. (195S).l

While other techniques are of considerable historical interest, it seems that current concepts of the validity of the lipid bilayer model are largely dependent on observations made by electron microscopy and X-ray diffraction. Earlier work, which required considerable technical and interpretative ingenuity, has been shown to be somewhat ambiguous in light of present-day experience, and will not be reviewed here; the reader is referred to Kavanau’s (1965) monograph for details. When the peripheral regions of cells are examined under the electron microscope, the well-known trilaminar structure is seen. Robertson (1960) has reviewed the many observations and variations in technique, which all revealed two electron-dense lines of 20-A. width separated by 35 A., and which led him to the concept of a basic “75-A. unit.” This concept, however, is not tenable in view of the different widths described by Zetterqvist (1956), Freeman (1959),

THE CELL PERIPHERY

67

Karrer (1960), Smith ( 1961), Sjostrand ( 1 963b), Yamoto (1963), Cunningham and Crane (1966), and Parsons (1967), among others. Sjostrand’s (1963b) observations are particularly valid, since his measurements of the width of trilaminar structures were all made on adjacent membranes within the same small field. In the mouse kidney and pancreas, mitochondria1 membranes and a-cytomembranes measure 50 A. in osmium-fixed material and 60 A. in specimens fixed with permanganate. In the same material, smooth cytomembranes measure 60 A. in osmium-fixed and 70-80 A . in permanganate-fixed preparations, whereas plasma membranes and the membranes surrounding zymogen granules measure 90-100 A. Quite apart from the usual artifacts associated with electron microscopy (Weiss, 1962b; Elbers, 1964; and others), the interpretation of the trilaminar structure in terms of lipid orientation presents many difficulties. One of these is that the arrangement of lipid molecules is in part determined by the amount of water present. Bangham (1963) has reviewed much of the literature on model systems, which indicates the formation of different phases as a function of lipid concentration. As water is removed from such systems, the lipids tend to form bilayers as these have the lowest free-energy configuration (Haydon and Taylor, 1963) and hence are the most stable structures. Thus, whatever the arrangement of lipid in model systems originally, a bilayer lamella would tend to result as a preparative artifact on dehydration. Sjostrand (1967) considered that from a knowledge of such model systems it is not possible to draw any conclusions about phase transitions in membrane lipids where they constitute only about 30% of the dry weight. However, work on model systems does indicate that lipids may well exist in a more dynamic state than implied by the bilayer concept (Weiss, 1962b; Lucy and Glauert, 1 9 6 4 ) , and polymorphism in lipids, in the aqueous phase, has been discussed extensively by Luzzati and his colleagues (Luzzati et d.,1957, 1958, 1960, 1962). Chapman et a]. (1967) and Clifford et (11, (1968) have examined the membranes of human erythrocyte ghosts in aqueous suspension by nuclear magnetic resonance spectroscopy and cannot detect freely mobile lipid hydrocarbon chains over a temperature range of 0-60OC. These negative findings argue against the existence of both lipid bilayer and micellar structures, since in these configurations the hydrocarbon chains would be expected to undergo easily detectable motion. Korn (1966a,b) and Korn and Weisman (1966) have attempted to evaluate the use of electron microscopy in determining the molecular relationships of lipids within the cell periphery. In a series of systematic studies on amebas, they have considered the reactions of various stains with lipids, which part of the lipid molecules are marked, whether the lipid or a marker is present in the specimen when it is examined under the electron microscope, and whether or

68

LEONARD WEISS

not the position of the lipid or its marker is oriented in the same way as in the untreated cell. Their experimental results indicate that following normal preparative techniques, Acunthumoebu cells fixed in glutaraldehyde cannot contain lipid at the time of microscopy, and cells fixed in osmium tetroxide or potassium permanganate lose most of their neutral lipids and a sizeable proportion of their phospholipids. Casley-Smith (1967) also considers that there is no general fixative for lipids. In spite of this loss of lipids, regular trilaminar structures are seen at the cell periphery. These results recall the work of Green and Fleischer (1963) in which no electron micrograph changes were observed in the trilaminar structure of mitochondrial membranes after acetone extraction of about 85% of their phospholipids. The work of Fleischer et ul. (1967) on the fine structure of lipid-depleted mitochondria is also difficut to reconcile with the bilayer model. These authors observed the persistence of trilaminar structures after removing 95% of the lipid from mitochondrial membranes, whereas if the bilayer model were correct, they would have expected either collapse or separation of the two electron-dense bands of the unit membrane. If, as suggested by Sjostrand (1963a), crosslinkage between the outer layers of proteins holds the trilaminar structures together, then it would be expected that these cross-links would be visible after lipid extraction. However, Fleischer et al. did not observe them. While these observations argue against the applicability of the bilayer model to mitochondrial membranes, the authors note that erythrocyte ghosts, for example, collapse when lipids are extracted by similar techniques. These results seem to argue more against a common “unit” membrane than to provide unequivocal disproof of the bilayer model. Wigglesworth (1947) and Baker ( 1958) suggested that OsO, acts by linking the double bonds of fatty acids by forming stable diesters. This suggestion is supported by the work of Stoeckenius and Mahr (1965), who could not demonstrate a direct reaction of OsO, with the polar groups of phospholipids, except phosphatidyl serine; and by Korn’s ( 1967) chromatographic and spectroscopic studies on unsaturated lipids which also show diester linkages with the unsaturated fatty acid chains through OsO,. However, Stoeckenius and Mahr also showed by infrared spectroscopy that secondary reactions involving hydrophilic groups do in fact occur when phospholipids react with OsO,; and Stoeckenius ( 1962) had earlier interpreted electron micrographs of lipid-water model systems “fixed” with OsO, to indicate the deposition of osmium at the polar heads of the phospholipids following breakdown of the osmic ester. Korn’s data show the formation of only one molecule of esterified osmium for every two molecules of fatty acid, indicating that it cannot necessarily be assumed that osmium reacts with, or marks, the polar groups of phospholipids during the fixation of biological specimens. Furthermore, in cellular membranes, marked

69

THE CELL PERIPHERY

polar groups may be related to proteins and carbohydrates as distinct from phospholipids. The overall impression gained is that whereas in defined model systems electron microscopy of stained preparations of lipids may be used to deduce their molecular arrangements, the same is not true for cell membranes. It is often considered that the strongest direct support in favor of the lipid bilayer membrane model is provided by X-ray diffraction studies of the myelin sheath, following the pioneering work of Schmitt, Bear, and Clark (1935). Geren (1954), Maturana (1960), and Peters (1960) have shown that the myelin sheath is formed by a remarkable rotation of Schwann cells; however, in view of its extremely specialized functions and electrical properties, it seems doubtful that the myelin sheath can be regarded as a useful model for other membranous structures. Finean ( 1962) has carried out extensive studies correlating the appearance of myelin sheath material under the electron microscope with X-ray diffraction patterns. This work has been generally accepted as providing rigid proof of the existence of a lipid bilayer structure in the myelin sheath. Some authors have used the results of the work on myelin as strong supportive evidence in favor of similar structures in the peripheries of other cells. The work of Finean and his colleagues on the X-ray diffraction of myelin sheath material has been interpreted in terms of a membrane about 80-90 A. in thickness, in which the distance separating the polar, phosphatic heads of the contained phospholipids is 50 A. (Finean, 1962) as shown in Fig. 4. Finean and Burge (1963) studied the X-ray diffraction patterns of myelin sheath under different degrees of swelling. From their measurements of intensity distributions under different conditions of swelling, they attempted to ascribe phases to the different peaks of the normal X-ray pattern. This interpretation of intensity PhosDhotidvlserine I Cholestirol

/

nyelin

171A

Myelin

\

FIG. 4. A proposed structure for myelin sheath (Finean and Robertson, 1958). L, lipid; Pr, protein; P, phosphorus.

70

LEONARD WEISS

distribution is open to criticism since it assumes that the only parameter to change when myelin swells is the spacing. As the swelling of myelin is ill-understood, and is not uniform, it seems unwise to assume that it is not accompanied by molecular rearrangements. That the interpretation of intensity distributions of swollen materials may be ambiguous is apparent from Perutz’ (1954) early discussions of his X-ray diffraction data on hemoglobin in which phase data derived from swelling of hemoglobin crystals are compared with those obtained using the technique of isomorphous replacement. When attempts are made to extrapolate electron microscope and X-ray diffraction data of osmium and permanganate-fixed myelin to fresh, unfixed myelin, serious difficulties are encountered. First, these two reagents alter the properties of the membrane as demonstrated with the polarizing microscope by Schmidt (1936, 1938), although the exact nature of the change cannot be interpreted, since in all polarization measurements form birefringence cannot be separated satisfactorily from intrinsic birefringence. Shah’s ( 1968) experiments with monolayers of lipids also indicate a change on osmium fixation. Stoeckenius et al. (1960) showed that when phospholipid lamellar phase systems react with OsO, the repeat pattern indicates that their lamellar structure is preserved. However, under these conditions there is loss of the 4.5-A. band which identifies the hydrocarbon chains of the fatty acid component of the phospholipids (Luzzati and Reiss-Husson, 1962). It has therefore been suggested that after fixation the lamellae are held together by relatively few cross-links between fatty acids on opposite sides of the bilayers, but that the packing of the remaining chains must be “severely disturbed.” Parsons and Akers (1968) have studied the effects of varying concentrations and reaction times of OsO, with myelin sheath. Their electron microscopic and X-ray diffraction data show that under the conditions used routinely in making preparations for electron microscopy there are obvious rearrangements within the sheath. The evidence quoted indicates that the X-ray diffraction data on the myelin sheath may not be interpreted as unequivocally as once thought. Various chemical fixation procedures used to prepare specimens for electron microscopic observation may seriously disturb the original molecular arrangement, and interpretations of fine structural detail should be made with considerable caution.

111. Other Models While it is probably true that the lipid bilayer model has not been unequivocally disproved, the fact that it cannot by any means be regarded as proven enables other concepts to be examined more impartially than previously. A multiplicity of models has been suggested, ranging from entirely lipid structures (Osterhout, 1940), through lipid/protein mosaics (Nathanson, 1904) , to the

THE CELL PERIPHERY

71

layered structures proposed by Danielli and Davson. Among the more interesting of these models, in a conceptual sense, was that postulated by Parpart and Ballentine (1952) on indirect evidence, and shown in Figs. 5 and 6. This membrane model is of particular interest in that it shifted the emphasis from lipid to protein structures by suggesting that lipids could exist within a primarily proteinaceous framework with their polar heads oriented into aqueous spaces.

r---------3

50 A.

FIG. 5 . Parpart and Ballentine’s (1952) model in tranverse section, showing the polar heads of phospholipids (small circles) interacting with membrane proteins (large circles) and oriented into aqueous spaces to form pores.

F

50 A

FIG. 6 . Parpart and Ballentine’s (1952) model in tangential section. Protein is designated by cross-hatching; water by stippling; nonaqueous phase by clear area; phospholipids by rectangles with circular heads; and cholesterol by rectangles.

Detailed analyses of human erythrocyte membranes have been made by Bakerman and Wasemiller (1967) among others. Their work shows that membrane material is a lipoglycoprotein containing 55% protein, 35% lipid, and 10% carbohydrate. The weight-average molecular weight of the complex was approximately 44,400, and 22,200 for the protein. These authors view their evidence

72

LEONARD WEISS

as indicating repeating units of erythrocyte structural membrane. Although average parameters can present an oversimplified view of cell peripheral structure, treatment of the membrane in terms of an integrated complex, involving all of its major components, appears to be the most promising approach at the moment. An indication of the diversity of possible membrane structures of lipoprotein, as compared to phospholipid, comes from the reviews of W. H. Cook and Martin (1962) and Oncley (1964). Although lipoproteins are known to contain phospholipids, neutral lipids, and proteins in varying proportions, little is known of their structure or binding, other than that they are combined in nonstoichiometric proportions by forces weaker than covalent bonds. This in turn indicates that the determined composition of the complexes will be very sensitive to techniques used to isolate them, and warns against too ready acceptance of analytical data in which drastic procedures have been used to isolate them from membranes. Cook and Martin classify the lipoproteins into those with a protein content below 33% (LPL) and those with a protein content above 33% (HPL). In the LPL class the neutral lipid-phospholipid ratio varies from 1: 1 to 10:1, but this ratio remains at 1:1 in the HPL class. LPL globules tend to have both proteins and phospholipids at their interface with an aqueous environment, and resemble micelles. The HPL class maintains a structural integrity approaching that of protein molecules. Conversion of the LPL to HPL classes is problematical, and would involve more profound changes than simple loss or gain of lipid. The relevance of Cook and Martin’s and Oncley’s reviews to cell membranes is suggested by Lenard and Singer’s (1968) studies on erythrocyte membranes treated with phospholipase C. Following treatment, which releases 68-74% of the total membrane phosphorus into solution, the membranes remain intact as observed by phase-contrast microscopy, and the average protein conformation in them as determined by circular dichroism measurements in the ultraviolet remains unaffected. These results are interpreted by the authors as an indication that the phosphoester bonds of membrane phospholipids are readily accessible to the enzyme and that electrostatic interactions between such phosphatic groups and membrane proteins play at most a minor role in maintaining both membrane integrity and the conformation of membrane proteins. These findings are considered to be more consistent with a scheme advanced by Lenard and Singer (1966) (Fig. 7 ) , on the basis of their studies of membrane protein conformation by the techniques of optical rotatory dispersion and circular dichroism, than the Danielli-Davson model, Lenard and Singer (1966) have proposed that the ionic and polar heads of the lipid molecules, together with all of the ionic side chains of the structural protein, are at the true surface of the cell, in contact with the environment. The nonpolar residues of the protein, together with the hydrophobic “tails” of the phospholipids and relatively nonpolar lipids such as

T H E CELL P E R I P H E R Y

73

cholesterol, are inside the membrane. They also postulate that helical portions of membrane protein are inside the membrane and that interactions of the types listed determine the overall conformation of structural proteins. They speculate that the subunits may be formed having these general arrangements, which could aggregate to form an intact membrane in a manner similar to that pro-

a 86

Hellcal coil portions

/vvv\ Random cal

portions

Lipids

FIG. 7. T h e Lenard and Singer (1966) model (cf. Fig. 2 ) . The proteins on the outer surfaces of the membrane consist of helical and random coil portions. The polar lipids are oriented in a bimolecular leaflet with their polar heads (circles) facing outward. The cross-hatched areas are assumed to be occupied by relatively nonpolar constituents (hydrophobic amino acid residues or lipids).

posed by Green and Perdue (1966). In assessing this model, caution must be used in correlating the accessibility of the phosphoester bonds of phospholipids to their position relative to the cell surface, as Seaman and Cook (1965) have accounted for the electrical charge at the erythrocyte surface in terms of the carboxyl of sialic acids and glutamyl residues and there is no direct evidence for the existence of positively charged groups associated with the polar heads of phospholipids in this region. A model similar to that proposed by Lenard and Singer has been put forward by Benson (1966) (Fig. 8) on the data derived from the lamellae of plant chloroplasts, which consist of arrays of subunits called quantasomes. In the quantasomes are found four amphophilic lipids, each containing a limited and specific group of fatty acids. Benson suggests that the hydrophobic hydrocarbon chain of these fatty acids may associate with specific hydrophobic amino acid residues of the membrane protein. In this model, a proteinaceous framework occupies the full thickness of the membrane; the polar groups of both the protein and lipids are located at its outer aspects while the central region is hydrophobic. The molecular arrangement would account for the trilaminar structure seen in electron micrographs of stained membranes, since the stains would col-

74

LEONARD WEISS

lect at the polar regions, leaving the central area electron-transparent. Benson makes the interesting speculation that metabolically driven alterations in the conformation of a flexible lipoprotein ion-exchange membrane of this type may play a role in transport phenomena.

FIG. 8. Benson’s ( 1966) model, showing a proteinaceous framework containing phospholipids.

Studies made of mitochondrial membranes may well be relevant to the structure of the cell periphery although it is obviously unwise to extrapolate from one membrane system to another. The mitochondrial data are extremely useful since, as noted by Parsons (1967), in contrast to other membranes at least 6070% of the proteins from mitochondrial membranes have been identified. Green et al. (1961) and Richardson et al. (1963) have shown that 40-50% of mitochondrial protein is present in an insoluble form which they have termed “structural” protein. It is of considerable interest that this structural protein has a molecular weight of 22,500 (Criddle et al., 1962, 1966), which corresponds closely to the protein of molecular weight 22,200 isolated from the erythrocyte membrane by Bakerman and Wasemiller (1967). Mitochondria1 protein selfpolymerizes readily, which possibly explains why trilaminar structures persist in mitochondria after removal of 95% of their lipids (Fleischer et al., 1967). The mitochondrial protein also forms strong complexes with phospholipid. Parsons (1967) cites unpublished work of Racker and Stoeckenius showing that lipidfree protein similar to structural protein has an amorphous electron microscopic appearance, however, the addition of a small quantity of phospholipid to the

75

T H E CELL PERIPHERY

system results in the formation of sheets and vesicles resembling fragments of mitochondrial inner membrane. Of great importance is the observation that membranelike structures apparently can only be formed by structural protein. McConnell et al. (1966) have described how membranes can be formed by the so-called repeating units of mitochondrial cytochrome oxidase. “Repeating units” are “the ultimate lipoprotein units of membranes . . (Green et al., 1967) and may be prepared by treatment of membranous structures with bile salts. On removal of the bile salts, membranous structures are formed. Under controlled conditions, the ability of the repeating units to form membranes is lost following the removal of phospholipid from them and may be restored following the addition of lipid to the system. Green et al. (1967) have presented electron microscopic evidence for membrane formation by the repeating units of mitochondrial membranes, the chloroplast membrane of spinach, the outer segments of bovine photoreceptors, a bovine liver microsomal membrane, and the plasma membrane of bovine erythrocytes. In all of these, lipid depletion reversibly inhibits membrane formation. The repeating units themselves, in the case of mitochondria1 membranes, are thought to be the so-called base pieces and may be visualized as approximately cuboidal proteinaceous structures, parallelopipeds (FernLndez-Morin et al., 1964) measuring 114 x 50 x 114 A. (Fig. 9 ) . Green and his colleagues postulate that when lipid is present it is confined to two of the faces of the parallelopipeds and that the repeating units can only interact at the four remaining faces that do not con.I’

Repeating units of membrane (base pieces)

Minus lipid ( b u l k phase)

Plus lipid (membrane)

FIG. 9. The unit model of Green et al. (1967) in which nesting repeating units are shown to combine to form a membrane, with the (dark) polar heads of lipids oriented to present to its inner and outer surfaces.

76

LEONARD WElSS

tain lipid, with consequent membrane formation. When lipid is absent, the repeating units can interact at all faces, leading to an amorphous aggregate. According to this ingenious hypothesis, to have the ability to form membranes a protein must meet four conditions. It must be able to polymerize into threedimensional aggregates; it should combine hydrophobically with negatively charged phospholipid ; it must combine asymmetrically with phospholipid in order to cover 2/6 faces only (vide siipra) ; and finally, the phospholipid-coated protein repeating unit must be capable of hydrophobic bonding with other units to form a curved sheet. Green et al. stress that, apart from the repeating units themselves, they have not found any other protein that can give rise to “authentic” membrane structures by interacting with phospholipids. These authors criticize the report of Kagawa and Racker (1966) that structural protein can form membrmous vesicles under the influence of phospholipids on the grounds that their published electron micrographs do not enable a distinction to be made between the domains of phospholipid micelles that are present and “authentic” vesicular membranes. No doubt this important point of controversy will be settled by experiment. In general, it might be expected that the proteins combining with phospholipids, by hydrophobic bonds, should contain a high proportion of amino acid residues with lipophilic side-chains. It is of interest that these conditions are met by the mitochondrial structural proteins (Criddle et al., 1362), myelin protein (Hulcher, 1963), and the protein isolated from erythroycte membranes (Bakerman and Waserniller, 1967). All of the more recent proposed membrane models tend to favor the membrane protein existing in globular form, in contrast to the Danielli-Davson model in which the protein adjacent to the polar heads of the phospholipids is depicted as in the extended 0-configuration. It is therefore of interest that Maddy and Malcolm’s (1965) examination of erythrocyte ghosts by optical rotatory dispersion and infrared spectroscopy reveals no evidence for proteins in the (3-configuration. Studies on the mitochondrial inner membrane made by a number of workers suggest that it is only 5 5 A. thick. This value is too low to be explained in terms of a lipid bilayer model. Parsons (1967) has suggested that phospholipids could be packed in these membranes in the form of “flat micelles” of minimum thickness 9 A., with their polar heads centrally placed, and their nonpolar tails forming a lattice of variable dimensions into which proteins could fit (Fig. 10). Whether or not this concept will be compatible with others in providing details of the combination of phospholipid and protein remains to be seen. Additional weight to the suggestion that Jome membranes may be built up from repeating units comes from the electron micrograph studies reviewed by Sjostrand (1967). In sections of permanganate-fixed cells of the proximal tubules of mouse kidney, Sjostrand observed mitochondrial membranes and

THE CELL PERIPHERY

77

smooth-surfaced membranes adjacent to plasma membrane. Although the plasma membrane appeared as a trilaminar structure, the other two membranes exhibited a well-marked globular substructure in which the globules were approximately 50 A. in diameter in the mitochondria and 60-70 A. in the smoothsurfaced membranes. Freeze-dried material also exhibited similar substructure.

FIG. 10. Parson’s ( 1967) model for mitochondria1 inner membranes. Speculative diagram for “flat micelle” type of packing of membrane phospholipid for units containing six phospholipid molecules per micelle (A, B) , and eight molecules per micelle (C, D). A and C are closely packed arrangements, B and D are more open arrangements. Such arrangements could occur in membranes of low phospholipid content and would have the advantage of leaving sufficient room for cytochromes and enzymes in a globular form. T h e models are only approximately to scale and do not show kinking of hydrocarbon chains because of unsaturated bonds. Actual models indicate that the polar base part of the phospholipid molecule readily assumes a near central position as indicated in the diagram.

Sjostrand emphasizes that he has not visualized globular substructure in either the plasma membrane or in the myelin sheath and points out that this in itself would suggest that at least two types of membrane exist, quite apart from this detailed substructure at the molecular level. It is of some interest that the myelin sheath does not exhibit the globular substructure, as Folch-Pi (1967) has concluded that in these structures phospholipids are bound to the protein by electrostatic and ionic bonds, which is contrary to the views expressed for other membranes. It may be mentioned here that Rendi and Vatter’s (1967) electron

78

LEONARD WEISS

microscopic studies of mitochondria1 membranes also lead them to postulate a globular substructure but they visualize their findings in terms of separate “granules” of phospholipid and protein about 20 A. in diameter (Fig. 11). While emphasizing that they cannot interpret their observations in terms of molecular arrangements, they make the point that two layers of assorted granules

FIG. 1 1 . Rendi and Vatter’s (1967) model. A diagramatic presentation of the image of the unit membrane interpreted by Robertson (left). Observation of the fine structure of sectioned and negatively stained membranes show that the membrane could be interpreted to be a mosaic of granules. T h e granules (right) are labeled to indicate phospholipids (PL) and “structural” protein (SP). (The arrangement of the two types of units is not to suggest their organization but to account for the proportions of the two components in the membrane.)

separated by 30 A. would give the same image of a trilaminar structure, thus supporting the Danielli-Davson model. The difficulties of interpretation of electron micrograph images of membranous regions, which are reviewed and discussed in detail by Elbers (1964), leave me with the strong impression that the problems of their detailed molecular arrangement cannot be solved with existing electron micrograph techniques, although as Sjostrand (1967) has observed, “The evolution of any concept regarding structure and function of living systems depends on exploring the techniques that are available, irrespective of their crudeness.”

IV. Cell Surface Charge A. MEASUREMENT OF CELLULAR ELECTROPHORETIC MOBILITY All cells from vertebrates so far examined carry a net negative charge at their surfaces. When such cells are suspended in an electrolyte solution, through which a direct current passes, they migrate toward the anode. In an electrophoresis apparatus, the velocity of individual cells located in the “stationary layer” with respect to the walls of the observation chamber in which they migrate may be measured directly. The electrophoretic mobilities of the cells are then expressed in terms of the observed velocity in microns per second per volt per centimeter of potential gradient (p. sec.-l volt.-1 cm.). The various

T HE CELL PERIPHERY

79

techniques used are discussed in the symposium report edited by Ambrose (1965) and will not be described here. Of prime importance to the biologist is the question: What do measurements of electrophoretic mobility mean and what information about the cell periphery can be derived from them? Some aspects will be dealt with here, but for a full discussion of the underlying theory the reader is referred to the recent detailed reviews of Overbeek (1950), Booth (1953), James (1957), Brinton and Lauffer (1959), Lyklema and Overbeek (1963), Haydon (1964), and Wiersema, Loeb, and Overbeek (1966). When a cell migrates through an electrolyte solution in an electrical field some of the environment moves with it. The interface between the environment moving with the cell and the bulk phase of the environment is the so-called “hydrodynamic slip plane,” and measurements of electrophoretic mobility reflect the potential at this plane (the zeta potential) 5, which may be regarded as the electrokinetic surface of the cell. A charged surface preferentially attracts ions of opposite charge, giving rise to a diffuse electrical double layer. The effective thickness of this layer may be defined in terms of the Debye-Huckel parameter 1 / K which is the distance from the true plane of surface charges at which the potential falls from I#~ to l / e x I$~.The value for 1/K is dependent on ionic strength and valency, as indicated by

I/K = 3.05.1-1/2 where I

= 1/2 Br 22 (Lewis and Randall, 1921), c = ionic concentration, and i i i

= valency. In “physiological” salines, 1/K is probably

8-10 A. (Heard and Seaman, 1960). A cell is not a spherical particle and the question has repeatedly arisen as to the value for the radius of curvature to be ascribed in electrokinetic studies of ceIls and the effects of crevices and filaments on the interpretation of mobility measurements. If fluid flows freely through the crevices and round the filaments, then from an electrokinetic standpoint the dimensions of the crevices or pits and the radii of the filaments must be taken into account. However, as discussed by Haydon (1964) among others, in the absence of a heavily filamented surface, and when the cell periphery is regarded as a porous macromolecular assembly, the movements of fluid may be assumed to occur at a plane, but ill-defined, surface, and the radius of curvature considered is that of the whole cell. These considerations present exceptionally difficult hydrodynamic problems which have to be considered for each type of cell surface geometry and appear somewhat intractable. z

80

LEONARD WEISS

Where u is the gross radius of a particle, when Ku > 300 which, from what has been said, is the case with cells in physiological saline, then electrophoretic mobility p, may be related to zeta potential 5 by the Smoluchowski equation

where E and TI are the dielectric constant and viscosity, respectively, in the region of the hydrodynamic slip plane. It is necessary to point out that the use of bulk-phase values for E and ‘(1 may well be a source of numerical error (Henniker, 1949). Haydon (1964) observes that when Ka > 300, 5 can be found for any shaped particle by the use of the above equation, if the distribution of charges in the electrical double layer is not affected by the field applied in electrophoresis, and if the cell surface conductivity is not large. Gittens and James (1963) drew attention to the influence of the electrical conductivity of the surfaces of bacteria on their electrophoretic mobilities. The higher the conductivity, the lower the observed mobility. Carstensen et al. (1968) have shown the changes in the conductivities of osmium-fixed sheep erythrocytes are sufficient to cause significant changes in their mobilities. Inspection of these workers’ data, however, reveals that although in solutions of below 0.02 M NaCl the erythrocyte mobilities are significantly higher than when conductivity changes are ignored, no significant change attributable to conductivity could be detected in solutions of greater ionic strength than 0.05 M NaC1. It therefore seems highly unlikely that surface conductivity is high enough to significantly affect the interpretation of mobility data made in physiological salines (10.145 M NaCI). Electrophoretic mobility may also be related to surface charge density (T by

p=-

(T

KE

The use of these equations in the present context makes the biologically unwarrantable assumption that the peripheries of cells are impenetrable to counter ions. Haydon (1961) has studied this problem and concludes that

where a is the fraction of the total space within the “surface” that cannot be occupied by counter ions; k is Boltzmann’s constant; T is absolute temperature; and YZ is the number of ions per unit volume of the bulk phase. Neglect of Haydon’s “a” could therefore lead to an underestimate of charge density by a factor of 2.

THE CELL PERIPHERY

81

In the above equation, it can be seen that the symbol v,,, for surface potential, appears. Consideration of the diffuse double layer reveals that the technique of cell electrophoresis give measurements of zeta potential, not surface potential, and that 5 < *q~".However, Haydon's (1960) measurements of 5 for oil droplets in aqueous solutions of surface-active ions are the same as measurements of ?loobtained by surface potential measurements of corresponding flat films, provided the surface potential is less thdn 50 mV. As the surface potentials of cells studied so far are thought to be less than -50 mV., Haydon equates +,) and 5 in the field of cellular electrokinetics. This may lead to considerable error. It appears that the technique of measurement of cellular electrophoretic mobility may be used to advantage in comparing the mobilities of similar cells measured under similar environmental conditions. Although e.rtimater of cell surface potential, surface charge density, and zeta potential can be made from mobility measurements, and although these estimates can be used for comparative purposes, too much confidence cannot be placed on the numerical estimates themselves. Some indication of the usefulness of electrophoretic techniques in studying the chemical nature of the cellular electrokinetic surface will be indicated below.

B. SIALICACIDMOIETIES Burnet et d. (1746) showed that filtrates from Vibvio chokvue and Clostridjum zuelchii destroyed the receptor sites for influenza viruses at the surface of the human erythrocyte. Hanig (1748) noted that when erythrocytes adsorb PR8 virus their electrophoretic mobilities are reduced. Thus, some of the negative charges on the human erythrocyte were in some way associated with virus adsorption. Ada and Stone ( 1750) demonstrated that the receptor-destroying enzyme from V . chdeme reduced the net surface negativity of erythrocytes by more than 80% and suggested that the moiety enzymically cleaved from the erythrocyte surface contained acidic groups. In 1958, Klenk suggested that acylated neuraminic acids could contribute to the negative charge on erythrocytes. The final steps in demonstrating this suggestion followed Ada and French's ( 1759) purification of a receptor-destroying enzyme which was shown to be neuraminidase, And Gottschalk's (1957) characterization of its specificity, which is the hydrolytic cleavage of the glycosidic bond joining the keto group of N-acetylneuraminic acid to a sugar or sugar derivative. In 1960, Heard and Seaman demonstrated an 80% reduction in the electrophoretic mobilities of human erythrocytes with purified neuraminidase and showed by analysis that such incubation resulted in the liberation of free sialic acid from the cells into the medium. It was later demonstrated (Weiss, Igblc, 1963b) that after incubation with

82

LEONARD WEISS

pure neuraminidase rat fibroblasts cultured on glass were detached more easily than their controls. This was interpreted as an indication that sialic acid moieties were also present at the surface of tissue cells. More direct proof of the presence of sialic acids at the cell surface has come from experiments in which a significant reduction in cellular electrophoretic mobility, indicating loss of anionic surface groups, occurs following incubation of cells with neuraminidase. In the investigations of normal and tumor cells by Wallach and Eylar (1761), G.M.W. Cook et al. (1962, 1963), and Miller et ul. (1963), among many others, only the N-acetyl and N-glycolyl derivatives of neuraminic acid have been observed. An additional note of caution against regarding any cellular membrane as a good model for any other comes from the observations that the electrophoretic mobilities of a variety of cells are not altered by neuraminidase treatment (Naaman et al., 1965; Chaudhuri and Lieberman, 1965; Wallach and Perez-Esandi, 1964). Of intracellular membranes, those of the nuclei of liver cells (Marcus et a/., 1965) and Ehrlich ascites cells (Mayhew and Nordling, 1966) had their mobilities reduced by neuraminidase, whereas the charge characteristics of the endoplasmic reticulum membranes of Ehrlich ascites cells were unchanged by such enzymic treatment (Wallach and Kamat, 1966). Mayhew and Nordling (1966) also made the interesting observation that while the electrophoretic mobilities of murine Ehrlich ascites, sarcoma 37 ascites, and liver cells and their homologous isolated nuclei were similar (indicating similar surface charge densities at the cell peripheries and at their own nuclear membranes) the reduction in mobility produced by neuraminidase differed between the peripheries and nuclei, indicating that the similarities in mobility were attributable to different anionic species. C. RIBONUCLEASE-SUSCEPTIBLE GROUPS Lansing and Rosenthal (1952) suggested that RNA was present in the peripheries of Arburia eggs and Elodeu cells; de Kloet (1961) suggested that it was present in the peripheries of the protoplasts of Sacchavomyces cnrlrbeygensir, and Chaudhuri and Lieberman (1965) described RNA on the surface of the liver cell nuclear membrane. Systematic attempts have been made to demonstrate the presence of RNA within the peripheries of mammalian cells by Mayhew and myself (Weiss and Mayhew, 1966, 1967; Mayhew and Weiss, 1968). Earlier work had shown that cells detached from glass left pieces of their peripheral zones on the glass surface (Weiss, 1961a,b; Weiss and Coombs, 1763; Weiss and Lachmann, 1964). When cells growing on glass slides in vitro and labeled with tritiated uridine were carefully removed from the glass, they left behind “footprints,” demonstrable by radioautography, over which cell outlines could be superimposed. These “footprints” were removed

THE CELL PERIPHERY

83

from the glass by treating it with RNase. The detachment of similar cells from glass was facilitated by treating them with active RNase but was unaffected by incubation with inactivated enzyme. These results suggested that RNA might be present as a structural component of the peripheries of the cells examined, although other interpretations of the experimental data were possible. The electrophoretic mobilities of Some cells were significantly reduced following incubation with RNase but were not reduced in cells incubated with RNase that had been inactivated by the method of Barnard and Stein (1958). The interpretation of these electrophoretic data is instructive in that it illustrates some of the more general problems associated with this experimental approach. On the one hand, reduction in cellular electrophoretic mobility following incubation with RNase could be the result of il loss of negatively charged surface groups associated with, and susceptible to, the enzyme; bovine pancreatic RNase acts on the phosphodiester bond between the 5’-positions of the ribose moieties in RNA (Brown and Todd, 1955) and is highly specific for pyrimidine nucleoside linkages (Volkin and Cohn, 1953). Thus, if the reduction in the net negativity of the surfaces of cells treated with RNase can be shown to result from its enzymic activity, this finding may be taken to indicate the presence of RNA in the cell periphery. On the other hand, it is well known that RNase is a basic protein, and its nonspecific adsorption to the cell surface would also produce a reduction in electrophoretic mobility because of a fall in the ?let negativity of the surface to which it adsorbs. A very strong indication that the effects observed by Mayhew and myself were not the result of nonspecific adsorption of the enzyme comes from a detailed consideration of the inactivated enzyme, which had no detectable effect at the cell surface. The inactivated enzyme used by us was prepared according to the method of Barnard and Stein (1958), by carboxymethylation of native RNase with bromoacetic acid, which removes the positively charged group at histidine residue 119 (Crestfield et a]., 1963). Thus, of the 19 positively charged groups in RNase, only one is removed by the inactivation procedure. This small change in net charge is evident from the fact that both the inactive and active enzymes have isoelectric points close to p H 9.6, and on ion-exchange columns there is little difference between them (Glick et al., 1967). Thus, from the electrostatic viewpoint, the differences between the active and inactive forms of the enzyme are unlikely to be accounted for by their different adsorptive capacities. Kartha’s (1968) X-ray diffraction studies of active RNase and enzyme inactivated by Barnard and Stein’s technique reveal only very slight conformational changes as shown in the electron density pattern at 4-A. resolution. The changes found in the inactive molecule are located around the active center at which it reacts with RNA-phosphate. If the changes in cellular electrophoretic mobility observed by us were the result of preferential adsorption of the active RNase

84

LEONARD WEISS

over the inactivated enzyme attributable to conformational changes, then it could be argued that only phosphatic groups at the cell surface would show this amount of discrimination between the two adsorbents, and that these are most likely to be those associated with phospholipids or RNA. Additional weight to the suggestion that nonspecific adsorption of RNase does not account for the reduction in the net negativity at cell surfaces comes from the observation that the electrophoretic mobilities of a number of different types of cells, including erythrocytes of three species, are not demonstrably affected by incubation with active RNase even though they all contain phospholipids in their peripheral regions. The peripheral RNA postulated by Mayhew and myself is not an adsorbed contaminant, since attempts to remove it by washing cells up to 12 times have been unsuccessful, and attempts to deliberately contaminate cells, both before and after treatment with RNase, by incubating them in suspensions of lysed cells, have not revealed adsorption of RNA reflected in measurements of electrophoretic mobility. The peripheral RNA is not attributable to the presence of PPLO-like organisms at the cell surface, since repeated careful examinations of cells obtained from both suspension culture and mouse ascites tumors by cultural techniques were consistently negative for mycoplasma over the periods of study. In the case of the Ehrlich ascites tumors, many electron micrographs have failed to reveal the presence of (RNA) viruses at the cell surface, which could have accounted for peripheral RNA. Very recent work by Mayhew and myself has shown that the net surface negativity of a number of different mammalian cells may also be reduced by incubation with TI ribonuclease which, in contrast to ribonuclease A, has a net negative charge. The effects of TI ribonuclease are therefore not attributable to its adsorption, but are consistent with the other data which indicate the presence of RNA and ribonuclease-susceptible anionic groups at the peripheries of some cells. Ion-binding studies (Weiss and Mayhew, 1967) have shown that calcium binding at the cell surface is reduced following incubation of cells with RNase. Our results indicate that calcium binds more avidly to RNase-susceptible groups than to neuraminidase-susceptible groups, and that it binds most avidly to as yet unidentified acidic groups. So far, RNase-susceptible acidic groups have been demonstrated by cell electrophoresis at the surfaces of murine ascites tumors (Ehrlich L1210 and sarcoma 37), permanent cell lines derived from human osteogenic sarcoma and murine mastocytoma, lymphocytes and possible polymorphonuclear leucocytes from human blood, freshly isolated mouse thymocytes, and liver cells. Additional supporting evidence for the presence of RNA within the cell perhiphery comes from other analytical approaches. Warren et nl. (1967) have

THE CELL PERIPHERY

a5

shown that i-2% of the total RNA of L cells is associated with their isolated peripheral membranes ; Lansing (1966) found small constant amounts of RNA associated with isolated, electron microscopically “clean” preparations of liver-cell peripheral membranes; and Burka et ul. (1967) have reported RNA in reticulocyte membranes. The disadvantage of all of these later techniques is that in addition to the problem of contamination of the isolated membrane specimens by intracellular contents, conventional analyses do not indicate where the RNA is located within the cell peripheral zone. By means of electrokinetic techniques, Mayhew and I have been unable to detect surface RNA in human, mouse, and chicken erythrocytes, human monocytes and platelets, mouse peritoneal macrophages, or in cell cultures derived from two Burkitt tumors. W e are currently attempting to examine as many different types of cells as possible. An obvious iind important question that relates to surface RNA concerns its type and function. Although it is possible to speculate (Weiss, 1 9 6 8 ~ ) on some of the possible consequences of surface RNA, it must be emphasized that the validity of such speculations depends very much on its characterization, which has not yet been accomplished. Crawling movements of cells over or through cellular or noncellular substrata can only be accomplished by the actively moving cells continuously making and breaking contacts with their substrata. It was postulated on theoretical grounds (Weiss, 1962a) that an inescapable part of active cell movement is that small parts of peripheral material will be torn off the moving cell and left behind on the surface over which it moves; pieces of the substrate over which a cell crawls may be ruptured off and carried away on the surface of the crawling cell, or both processes may occur, leading to a two-way exchange of peripheral material. If the postulated surface RNA is “informational,” its transfer from one cell to another could conceivably be regarded as transfer of information.

D. AMINO GROUPS Bangham et ul. (1958, 1962; Bangham and Pethica, 1960) failed to detect significant changes in the electrophoretic mobilities of a variety of cells over the pH range 7-9. As the pK of amino groups are in the range pH 7-10, it would be expected that increases in cellular net surface negativity would become apparent as the environmental pH approaches these values. Although small increases in mobility have been observed at values of approximately pH 10, the general validity of this type of experiment is questionable since at such high pH values other surface changes are likely to occur in addition to the ionization of amino groups (Pulvertaft and Weiss, 1963). This view is reinforced in the case of human erythrocytes by the observations that their mobilities were unaffected by treatment with formaldehyde, acetaldehyde (Heard and Seaman,

86

LEONARD WEISS

1961), p-toluenesulfonyl chloride (Seaman and Heard, 1960), or by 2:4dinitrofluorobenzene (Seaman and Cook, 1965), as all of these reagents are expected to react with accessible amino groups causing loss of positivity. More recently, Weiss, Bello, and Cudney (1968) have studied the electrophoretic mobilities of human and mouse erythrocytes and cultured and ascites tumor cells after treatment with freshly generated formaldehyde, 2,4,6-trinitrobenzenesulfonic acid, 2-chloro-3,5-dinitropyridine,or 2-chloro-3,5-dinitrobenzoic acid. None of the reagents lysed the erythrocytes, and the three aromatic reagents were used in concentrations that were nonlethal to the nucleated cells. Of the possible basic groups present in the cell periphery, Gasic et al. (1968) listed the side-chain amino groups of lysine and hydroxylysine, terminal protein aamino groups of arginine, and phospholipid and glycolipid nmines. The aromatic compounds used by us do not react with the guanidine groups of arginine, although formaldehyde does (Fraenkel-Conrat and Olcott, 1948). None of the reagents react with the positively charged quaternary ammonium ion of lecithin, and their reactions with other peripheral phospholipids is problematical owing to the possibility of phosphatcamine interactions. None of the four types of cells studied had their electrophoretic mobilities consistently increased by the reagents in spite of evidence of reaction. This absence of loss of surface positivity was not attributable to screening by sialic acids or trypsin-susceptible groups, as treatment with the various reagents following incubation of the cells with either neuraminidase or trypsin produced no significant changes in mobility differing from those of the cells treated with the enzyme alone. It was therefore concluded that positively charged groups associated with proteins were not present in detectable amounts at the electrokinetic surfaces of cells. It is to be emphasized that these electrokinetic studies do not indicate that positively charged amino groups associated with proteins are not present within the peripheral zones of the studied cells, but rather that they are not detectable at their electrokinetic surfaces. Gasic et al. (1968) have suggested on electron micrograph evidence that positively charged groups, which react with electron-dense negatively charged colloidal particles, lie within the peripheral zone deep to the surface sialic acid moieties, and our own findings are consistent with these conclusions.

E. OTHERPERIPHERAL IONOCENIC GROUPS It is to be expected from a consideration of some of the newer models for surface membranes that the negatively charged phosphatic groups of polar lipids will be present at the electrokinetic surfaces of cells. A noteworthy attempt to identify anionic groups at the cell periphery was made by Bangham and Pethica (1960), who studied the concentrations of

THE CELL P E R I P H E R Y

87

various cations required to induce reversal of charge. Prom this work they concluded that phosphatic groups were present at the electrokinetic surfaces of a number of different cells. It now appears that this technique is not as specific as once thought and that simple charge reversal spectra do not permit unequivocal identification of surface ionogenic groups. In the case of human erythrocytes, where surface charge was once thought to be largely attributable to ionized phosphate groups, Seaman and Cook (1965) have shown that the dominant ionogenic group at the electrokinetic surface is the carboxyl of Nacetylneuraminic acid, with a minor contribution from the a-carboxyl of glutamic acid. These workers also showed that when aldehyde-fixed erythrocytes were treated with diazomethane, which esterified acid groups, they were isoelectric between p H 6 and 8. If it can be assumed that this drastic treatment leaves the erythrocyte surface in a state relevant to that in the normal cell, then this observation would argue against nonspecific, asymmetric distribution of environmental ions near the cell surface making a significant contribution to cellular electrophoretic mobility. It is to be noted that following treatment of a variety of cells with both neuraminidase and RNase, the cells do not become isoelectric (Weiss and Mayhew, 1967), indicating that other groups are present at their electrokinetic surfaces. The nature of these groups remains obscure; they may well be phospholipid phosphates. The positive identification of these ionogenic species by means of electrophoretic techniques as described above will depend on the availability of highly purified enzymes which will cleave them from the cell surface.

F. DYNAMIC ASPECTSO F SURFACE CHARGE

So far, the cell periphery has been discussed in rather static terms. As in any other organelle, this region of the cell must be in a dynamic state with respect to anabolic and catabolic processes, and any given part must be regarded in terms of a half-life period. A detailed examination of the biochemical synthesis involved in maintaining membranes in a steady state is beyond the scope of this review; indeed, studies of this aspect of membrane physiology in metazoan cells has hardly begun and is complicated by the difficulties inherent in obtaining pure membrane fractions for analysis. Changes in the cell periphery associated with cytodifferentiation, induction, modulation, and virus infection have been reviewed in some detail elsewhere (Weiss, 1967a) and will also not be dealt with here. In 1962, Eisenberg et al. made an attempt to relate the electrophoretic mobilities of rat liver cells to their growth rate. The liver cells were isolated from their parent organs following partial hepatectomy and during postnatal growth. The mobilities in the regenerative phase following hepatectomy and in the

88

L E O N A R D WEISS

early neonatal period were significantly higher than in cells from normal adults. These results are in general agreement with those of Ben-Or et al. (1960), Heard, Seaman, and Simm-Reuss ( 1961) and Ruhenstroth-Bauer and Fuhrmann (1961), who observed that some cells derived from embryonic or regenerating tissues have significantly higher mobilities than their “normal” or adult counterparts. However, from their own studies on cultured cells, SimonReuss et al. (t964) concluded that it was inipossible to generalize about the effects of age and regenerative processes on electrophoretic mobility. Later and more sophisticated experiments, which will shortly be described, suggest that it may well be possible to make some general statements about the effects of metabolism and other intracellular events on electrophoretic mobility and that these predictions lend themselves to experimental testing. In any studies of cells isolated from solid tissues, there is always the danger that the dissociation procedures will modify the cell periphery. This is particularly true of cells that are isolated from tissues by trypsinization, since trypsin may remain at cellular electrokinetic surfaces for some hours and reduce the net negativity by virtue of its own net positive charge (Barnard et al., 1969). Mechanical isolation of cells from their parent tissues may result in alterations of their surfaces because the plane of separation does not coincide with the plane of the cell surface (Weiss, 1967b). If cells are separated from their parent tissues without irreversible damage and are cultured for comparatively short periods to enable them to recover, it could be reasonably questioned whether or not an isolated cell could ever reconstruct its peripheral regions in exactly the same way they were in solid tissues, since it has to adapt to the requirements of its existence as a unicellular organism instead of being surrounded by other cells with which it may communicate (Loewenstein, 1967) and a variable amount of connective tissues. It also seems reasonable to argue that some cells adapt to our various arbitrary culture media faster and/or better than others and that the hypothetical recovery periods for isolated cells in such media vary accordingly. Mayhew and O’Grady (1965) and Mayhew (1966) have made a series of studies on cells in suspension culture in which parasynchrony was induced and have shown conclusively that, in the strain of cells studied, electrophoretic mobility is significantly higher during the mitotic peak phase than at any other time in the mitotic cycle. Regardless o f their phase in the mitotic cycle, treatment of these particular cells with neuraminidase reduced their electrophoretic mobilities to a common value, suggesting that the observed increases in net surface negativity observed at mitotic peak phase are attributable to an increased density of ionized carboxyl groups of sialic acid moieties at the cellular electrokinetic surface. Kraemer’s (1967) analyticnl data suggest that there is a constant amount of sialic acid per unit cell volume irrespective of mitotic phase. Mayhew’s

THE CELL P E R I P H E R Y

89

interpretations are perfectly consistent with Kraemer’s data, since Mayhew’s electrokinetic data measure only those charged moieties located not more than 10 A. from the hydrodynamic slip plane, whereas Kraemer’s measurements refer to sialic acid liberated from cells when they are incubated with neuraminidase. T h e sialic acid moieties detected by Kraemer could thus be located anywhere within the depth of the three-dimensional peripheral zone of the cell, which is consistent with Seaman’s concept that this region may be regarded as a polyanionic sponge; and Mayhew’s observation could be interpreted as indicating that the increase in sialic acid-dependent surface charge density at mitotic peak phase is the result of a structural rearrangement of sialic acids within the peripheral zone as distinct from de uoz’o synthesis. On the other hand, Kraemer makes the assumption that when cells are incubated with neuraminidase, this enzyme reacts only with their surface regions and neither enters cells, as assumed by Wallach and Eylar (1961), nor reacts with intracellular membranes after entry. Nordling and Mayhew ( 1966) have convincingly demonstrated that neuraminidase both enters cells and, after entry, reduces the surface charge density of their nuclei; this raises the possibility that some of the sialic acid liberated when cells are incubated with neuraminidase comes from intracellular structures. However, it appears likely that a major cause of discrepancy between “chemical” and “electrokinetic” estimates of cell-surface sialic acid lies in the shortcomings in relating zeta potential to surface charge density. An attempt was made to correlate electrophoretic mobility with cellular metabolic activity, in a line of tumor cells growing in suspension culture, by studying the effects of environment:tl temperature on both of these parameters. It was observed that there was a good correlation between mobility and oxygen consumption over the temperature range 2°-600C., and the speculation was raised that there might be a causal relationship between energy-dependent conformational changes at the cell periphery and the charge density at its electrokinetic surface (Weiss, 1966). Intensive studies made over 2 years (Weiss and Ratcliffe, 1968) on two types of tumor cells, maintained in suspension culture and on a murine ascites tumor, have confirmed that same cells d o show true, rapid increases in electrophoretic mobility as their environmental temperature is raised from 10 to 37”C., but that the changes, although stat~stically significant, are small. Temperature-dependent mobility changes, when they occur, are associated with increases in mean cell volume and in susceptible cells may also be induced by exposure to hypotonic media; it has been suggested on this and other evidence that the mobility changes are the result of unspecified interactions of the cell surface with serum constituents, together with expansive movements of the cell surface. T h e temperature effect is not observed in all cells; thus, on the one hand, Merishi and Seaman (1966) have failed to detect it, but on the other hand, Nordling (1967) has. Weiss and Ratcliffe treated cells with a variety of anti-

90

LEONARD WEISS

metabolites and noted that in the cells examined severe depression of oxygen utilization, anaerobic glycolysis, and uncoupling of oxidative phosphorylation can occur, with no appreciable alteration in electrophoretic mobility. It would therefore appear that there is no direct causal realtionship between short-term metabolic changes of the types mentioned and cellular electrophoretic mobility. Longer-term studies by Mayhew and Weiss (1968) on a line of cells maintained in suspension culture have shown that when their growth rates are increased over periods of several days by increasing the amount of serum in the culture medium, there is a reversible increase in electrophoretic mobility. Studies on the effects of neuraminidase and ribonuclease on the mobilities of these cells strongly suggest that the surface density of RNase-swjceptible groups increases with growth rate, whereas the surface density of sialic: acid moieties remains relatively constant irrespective of growth rate. When cells are dying or moribund, because of nutritional deprivation, RNase-susceptible groups are not detectable at the cell surface; however, within one generation after previously starved cells are supplied with fresh media, RNase-susceptible groups reappear. Of the 13 cell types described by Mayhew and myself, only actively growing cells showed a marked reduction in electrophoretic mobility (20% or more) on incubation with RNase. Growth rate cmnot invariably be correlated with the possible presence of RNA at the cell surface, as treatment with RNase only reduces the electrophoretic mobilities of L1210 cells by 4-9% even when these cells grow very rapidly in vitra.

G. CHARGEDISTRIBUTION Consideration of electrophoretic mobility measurements indicates that at best they provide a crude index of surface charge density. The measurements do not indicate the arrangements of the ionogenic groups at the cell surface. Experiments made on the activities of penicillinase at the surfaces of B U C ~ ~ ~ Z L J . siibtilis suggested that the densities of charged groups in some surface regions were higher than in others (Weiss, 1963a). Later work on the deformability of mammalian cells showed that whereas the cells became more easily deformed after incubation with neuraminidase, no change was detectable after treatment with ribonuclease (Weiss, 1965a, 1968a). One explanation offered for these experimental data was that expansion of the cell peripheral zone into a micropipet by suction was accomplished by an “unplcating” process, and that partial resistance of “unpleating” was attributable to the electrostatic repulsion between the ionized carboxyl groups of sialic acids which were present in a higher-thanaverage density over the pleats. Studies of the effects of temperature and hypotonic media on electrophoretic mobility also led to the suggestion that some of the charged groups at the cellular electrokinetic surface had zonal distribution in

T H E CELL P E R I P H E R Y

91

regions of above-average surface charge density (Weiss, 1966; Weiss and Ratcliffe, 1968). I t has been suggested that contact phenomena between cells may in part be regulated by electrostatic repulsion between their surfaces (for review, see Weiss, 19672). In many time-lapse cinemicrograph studies of contact between a variety of animal cells in culture, cells are seen to explore each other's surfaces in a manner strongly suggesting some sort of spatial specificity. In n sense, spatial specificity implies structural heterogeneity. Electron micrograph evidence of such surface specialization comes from the work of Farquhar and Palade ( 1963), Fawcett (1965), Roth and Porter ( 1964), and Bowers ( 1954), among others. It is known from the work of Chambers and Fell (1931) and Ambrose (1961) that when cells make contact with glass substrata they do so over small regions of the total glass/cell interface. Correlation of experimental observations un the contact interactions of different cells with glass surfaces and the computed interactions forces between the cells and glass strongly suggest that close contact at distances permitting the formation of adhesive bonds cannot be accomplished by cells having uniformly distributed ionogenic units at their surfaces (Weiss, 1968b). The various evidence in favor of heterogeneity in surface charge distribution presented here is indirect; nonetheless, awareness of the possibility of its existence may be of some importance in the biophysicJ analysis of surface-dependent cell contact phenomena.

V. Enzyme Activity and the Cell Periphery A. SUBLETHALAUTOLYSIS Another aspect of change at the cell periphery is that attributable to enzymes. Fell and Mellanby (1952) showed that in the presence of excess vitamin A, cultures of chick embryo cartilaginous limb bone rudiments show loss of metachromasia on staining with toluidine blue because of a loss of intercellular matrix. Thomas et ul. (1960) produced a histological picture similar to that seen in hypervitaminosis A with papain. Further studies by Lucy et al. (1961) suggested that normal chondrocytes contain enzymes capable of degrading cartilaginous matrix, producing an effect similar to that of vitamin A. Dingle (1961) showed that the enzymes involved were, in fact, lysosomal hydrolases, as defined by de Duve (1959) and his colleagues. The observation that pretreatment of rat dermal fibroblasts cultured on glass with excess vitamin A facilitated their detachment from this substratum also suggested that released lysosomal enzymes could attack and weaken the peripheral zones of cells (Weiss, 1962a,b). Following the work of Bitensky and her colleagues (Bitensky, 1963) on the cytoxic activity of antibodies, Weiss and Dingle (1964) studied the

92

LEONARD WEISS

effects of exposing rat liver lysosomes in suspension, rat liver slices, and cultures of rat “fibroblasts” to an antiserum prepared by injecting rabbits with partially purified rat-liver lysosome preparations. The loss of (lysosomal) acid phosphatase after exposure to antiserum was demonstrated in both the liver slices and the cell cultures but, interestingly enough, not in the lysosome suspension. This finding that antisera were without direct effect on isolated lysosomes, as later confirmed by Dumonde et ul. (1765), resulted in the conclusion that lysosomal activation within cells caused by antiserum was riot the result of direct action on the lysosomes themselves. This conclusion supported Bitensky’s ( 1 963) suggestion that antiserum affects lysosomes indirectly by increasing the permeability of the cell membrane. Dingle et ul. (1967) have postulated that adsorption of complement-sufficient antiserum to the plasma membrane casues local alterations facilitating the fusion of the membrane with prin-rary lysosomes, and Dingle (1,968) has discussed such fusion in terms of the stability of emulsions. More direct evidence that antisera could act upon cells causing the release of enzymes capable of degrading intercellular matrix came from in zitro studies of the effects of antisera on fetal mouse bones (Fell and Weiss, 1965) and embryonic chick limb bone rudiments (Fell and Weiss, 1964). These degradative changes were inhibited by hydrocortisone. Weiss (1965b) observed that the detachment of cells growing on glass substrata could be facilitated by exposing them to low concentrations of antisera and that this facilitation was inhibited by microgram quantities of hydrocortisone, which also reduced the loss of intracellular acid phosphatase. It was suggested on tlilis and other evidence that the cell periphery may undergo continuous modifica.tion by sublethal autolysis and that this may be under endocrine and other physiological control. This concept o f “chronic weeping lesions” at the cellular level has been discussed in terms of cell interactions (Weiss, 196713). The reviews of Weissmann (1965) and Straus ( 1967), among others, indicate that lysosomal activation may be induced by many agents under a wide variety of pathological and physiological circumstances. This may indicate that sublethal autolysis of the cell periphery may commonly occur, and may well impose the necessity for a high rate of peripheral turnover on many cells.

B. SURFACE PH Quite apart from the sublethal autolysis discussed above, the enzymic degradation of intercellular materials by enzymes is thought to play an important role in infiltrative and metastatic processes in malignancy (Sylvin and Malmgren, 1957). As enzyme activity can be controlled by environmental pH, the question of hydrogen ion concentration near the cell periphery is of some importance. When aqueous solutions of su1fon;ited acidic dyes are shaken with benzene to

THE CELL PERIPHERY

93

form an emulsion, there is a color change indicative of a lower pH at the benzene/water interface than in the bulk aqueous phase (Deutsch, 1927, 1928). The existence of differences between bulk-phase, and interfacial p H , are also indicated by studies on the relationship between the tensions at fatty acid/water and oil/water interfaces, and bulk-phase pH (Reinders, 19 10; Jahrisch, 1922; Hartridge and Peters, 1922; Peters, 1931; Danielli, 1937). The studies showed that although the curves relating pH and fatty acid dissociation were of a shape similar to those relating pH and interfacial tension, there was a shift of approximately 2 pH units between the two curves, suggesting that the hydrogen ion Concentration near the fatty acid/water interface was 100 times higher than in the bulk phase. A more mathematical approach to the question of surface pH was presented by Hartley and Roe (1940), who postulated that the hydrogen ion concentration near a negatively charged surface is the product of the bulk-phase concentration and the factor exp( --eS/kT) where, e = electronic charge; = zeta potential; R r Boltzmann’s constant, and T = absolute temperature. T h e effective dissociation constant at the surface, K,, is given by

K , = K,, exp (-ec/KT) = K , exp (-F
= pH,,,,,,

+ S/60

94

LEONARD WEISS

well be different than that calculated from aver.ige electrokinetic data because of the presence of higher-than-average and lower-than-average charge density zones (Weiss, 1963a). Other evidence showing that the chiirge on a! substrate may affect its enzymic degradation conies from experiments made or1 the digestion of thin films of lecithin on a Langmuir trough by lecithinase (Bangham and Dawson, 1958; Dnwson and Bangham, 1959). These experiments do not indicate whether the effect of the charge on the lecithin films is mediated directly through ApH, or through electrostatic orientation of the lecithinase with respect to the lecithin. It is now generally appreciated that many cells are surrounded by carbohydrates and that use of the term “glycocalyx” for this region (Bennett, 1963) is apt. This material may have antigenic properties, as discussed by Watkins (1967), and niay be demonstrable as “fuzz” to electron microscopists, as discussed by Revel and Ito (1967), among others. It is often overlooked that in tissues the dividing line between the cell surface and intercellular matrix is a purely arbitrary one and that a cell may usefully be regarded as extending out for an ill-defined distance into its connective tis:iue domain. The bulk volume of these domains may be great owing to the associated water; the specific hydrodynamic volume of hyaluronate for example, is of the order of 200-500 ml./gm. (Rogers, 1961) because of the “tangled-skein” configuration of the hyaluronate complex occupying a verly large solvent domain. Rogers considered the role of hyaluronate in terms o f its effect on water retention, rates of diffusion, and inhibiting enzyme activity by virtue o f its macroanionic properties. These enzyme regulatory functions have been considered in some detail by Weiss (1962b, L967a) in relation to mammalian cells, in which it is postulated that by virtue of ApH effects hyaluronates and similar polyanions could, on the one hand, inhibit enzymes having higher than “physiological” optimum pH and, on the other hand, might be expected to optimize the activity of enzymes working best at lower than physiological pH. By limiting outward diffusion from the cell, hyaluronate would also tend to localize the activity of exoenzynies to the immediate pericellular region. Also, as suggested by Rogers, removal of macroanionic matrix by an hyaluronate-hyaluronidase interaction in this region could also control pericellular enzymic activity. Changes in electrical charge density at cell surfaces have been described above and have been related to cellular growth rate and to the mitotic cycle. All of these activities might therefore also play a role in controlling enzyme activity in the region of the cell periphery and in integrating this activity with other aspects of cell function.

VI. The Peripheries of Malignant Cells Many claims have been made in the past in which structures and properties have been described that are unique to the peripheries of malignant cells. Most

T H E CELL PERIPHERY

95

of these postulated differences between the peripheral regions of normal and malignant cells have proved to be either unfounded, or differences in degree only. The mere bulk of the literature on this topic prevents its exhaustive review, and only some of the conclusions reached in a detailed examination of the literature (e.g., Weiss, 1967a) can be presented here. It should be emphasized that some 273 human neoplasms are cataloged in the International Union Against Cancer (U.I.C.C.) Illustrated Tumour Nomenclature. This large number permits a great deal of individuality in the nature of the periphery of cells representative of any particular neoplasm and cautions against uncritical extrapolation of data from not only one type of human tumor to another, but also of data obtained from tumors in experimental animals to the human situation. It is often overlooked by those unfamiliar with histopathology that tumors are not homogeneous structures containing only viable cells. In any malignant tumor there is often obvious cytological and histological variation among apparently viable cells in addition to various necrotic foci. This necessitates careful definition of the neoplastic material studied, which is often lacking in papers of a biophysical nature. When studies are made on cells mechanically isolated from tumors there is a definite possibility that the plane of separation between the cells is spatially different from their adhesion interface (Weiss, 1967b). While this would not be expected to affect gross analytical data, it may profoundly affect true surface parameters such as electrophoretic mobility. A.

FINE

STRUCTURE

As summarized by Mercer (1963), many of the published electron microscopic data depict changes typical of cell death or degeneration as distinct from changes unique to malignancy per se. Mercer was of the opinion that no unique properties of malignant cell peripheries had been observed in thin sections by electron microscopy. As far as I am aware, this conclusion is still valid. In electron micrographs of rapidly proliferating neoplasms, considerable variation of junctional substructure is seen in which loose junctions and simplifications of peripheral structures are prominent features. However, similar changes have been observed by Lane and Becker (1966) in regenerating rat livers following partial hepatectomy. Some of the changes observed in neoplasms could also be the result of sublethal autolysis, as Overton (1962) has described the disappearance of desmosomes in enzyme-dissociated embryonic chick cells. An early attempt to study the surface contours of malignant cells was made by Coman and Anderson (1955) using a replication technique. On comparing the surfaces of VX2 epidermoid cancer cells from rabbits with normal squamous cells from depilated skin, Coman and Anderson observed that whereas the nor-

96

LEONARD WElSS

ma1 cells revealed surfaces covered uniformly with plaques some 30-60 A. in diameter, the surfaces of the malignant cells were much more hetergeneous, being covered with plaques ranging from 30 to 300 A. These observations were essentially confirmed by Berwick (1959), who also reported that the surface contours of cells of Shope papillomata, which can be regarded as intermediates between normal epithelial cells and those of the VX2 tumor, were homogeneously particulate. However, comparisons of the surface morphology of normal and leukemic lymphocytes of the rat by Nowell and Berwick (1958) failed to reveal consistent differences, since both cell types wcre covered with unevenly distributed plaques some 100-300 A. in diameter. Similarly, no significant differences were detected between normal cells of the hamster renal cortex and tumor cells of the same region by Easty and Mercer (1960), and Catalano et ul. (1960) observed no differences in surface substructure that could be correlated with the very different j u i i i w behavior of two sublines of the MCIM mouse sarcoma. It will be of interest to see if differences in surface structure between malignant cells and their normal counterparts can be revealed by means of the newer freezefracture techniques described by Bullivant and Ames (1966), Branton (1966), and Weinstein and Bullivant ( 1967) among others, although present findings indicate that differences attributable to malignancy per se are unlikely. In connection with possible junctional differences between normal and malignant cells, it is of interest that Loewenstein and Kanno (1966) demonstrated low electrical junctional resistance between microelectrodes placed in adjoining normal liver cells, but obtained no evidence of low-resistance junctions between various hepatomata. These results must be treated with considerable caution, quite apart from the obvious pitfalls in extrapolation to other tumors, as the “negative” finding of high junctional resistance could possibly be the result of difficulties inherent in this technique.

B.

CALCIUM

BINDING

Beebe (1904) and Clowes and Frisbie (1905) observed that the total calcium content of a limited number of malignant tumors was somewhat less than in norindl tissues. Carruthers and Suntzeff ( 1944) noted that following treatment with the carcinogen methylcholanthrene, the calcium concentration in mouse epidermis fell during the phase of benign hyperplasia. As the lesions became frankly malignant, a second fall in the calcium level occurred. In line with these findings, Brunschwig et ul. (1946a) found that human gastric carcinomata contained less calcium than the adjxent uninvolved mucosa, but that in a benign papilloma located between two discrete carcinomata, the calcium level was the same as in the malignant lesions. Brunschwig & al. (1946b) also noted that a number of human colonic carcinomata contained less calcium than both adjacent

T H E CELL PERIPHERY

97

normal mucosa and benign papillomata. However, the more recent studies of Kalant et al. (1964) on the DAB-induced hepatoma have not revealed a positive correlation between malignancy and low tissue calcium. The electrokinetic studies of Bangham and Pethica (1960) have shown that the environmental calcium concentration required to produce reversal of surface charge in mmse cells is approximately the same for those of Ehrlich ascites tumors as for those isolated from normal livers, and that lymphocytes and erythrocytes require even higher concentrations, indicating still lower calcium-binding capacities than the tumor cells. Thus, the commonly made statement that tumor cells in general contain less calcium, or bind this ion less avidly than normal cells, is unacceptable on existing evidence. It would be very reassuring if more tissue calcium levels could be determined on intact tumors of wide variety using modern analytical techniques. In 1900, Herbst showed that Echinus mirrotiiberri/latl/s blastomeres could be dispersed into single cells by treating them with calcium-free seawater. Since that time many workers have shown that removal of calcium facilitates the separation of cells from each other and from a variety of substrata. Coman (1953) suggested that the decreased “adhesion” he observed between malignant cells might be related to their decreased calcium content, and Zeidman (1947) had shown that separation of cells was indeed facilitated by calcium removal. Coman therefore postulated a sequential relationship between the low calcium content of tumors, the facilitation of separation of malignant cells from them, and metastases. The role of calcium in holding cells together has been envisaged as a “bridge” linking anionic sites on adjacent cells (Steinberg, 1958), in terms of its effects on electrical double layers surrounding cells (Weiss, 1960), in terms of desolvation (Schmitt, 194 1) , and in stabilizing coacervates of mucosubstances (Rinaldini, 19j S ) . All of these interactions are more complex than generally appreciated, as discussed by Katchalsky (1964). Recent work on murine sarcoma 37 cells (Weiss, 1 9 6 7 ~ )and cultured cells (Weiss, 1967d) suggests that calcium binds to their peripheries, increasing their mechanical strength and thereby hindering their separation, but does not demonstrably affect their mutual adhesion. Calcium is thus visualized as binding “tangentially” within the peripheries of individual cells as distinct from “radially,” in which position it can bind one cell to another cell or to intercellular substance. Thus, the present status of calcium in the tumor cell periphery, its mode of binding, and its function in periphery-dependent contact phenomena, is not clear.

C. SURFACECHARGE Ambrose (1967) has reviewed some of the various evidence comparing the surface charge of “normal” cells with their malignant counterparts. Cells from

98

LEONARD WEISS

stilbestrol-induced kidney tumors of the hamster, and from butter-yellow hepatomata of the rat, had higher electrophoretic rnobilities than their normal analogs. When cultured hamster fibroblasts were converted by polyma virus, they fell into two colonial types. One type had electrophoretic mobilities indistinguishable from the original fibroblasts, whereas the other cell type exhibited an increase in neuraminidase-susceptible mobility. This latter observation was consistent with that of Defendi and Gasic (196.5), who noted that polyoma-induced conversion of embryonic hamster cells W.IS associated with the acquisition of a dense, pericellular, neuraminidase-susceptible region of acid mucopolysaccharide, demonstrable by cytochemical techniques. Other cell surface changes after virus infection, which may possibly correlate with those mentioned above, relate to the adsorption of erythrocytes (Marcus, 1962) and the development of specific viral antigens (Vogt and Rubin, 1962; Haughton, 1965; Pasternak, 1965; Tevethia et nl., 1965). The question as to whether or not increased cellular electrophoretic mobility is correlated with malignancy, as so often suggested by Ambrose, is a complex one. Fuhrmann (1965) does not regard increased electrophoretic mobility as a nonspecific expression of cell proliferation on the evidence that whereas the mobilities of both proliferating liver cells and cells of an ascitic hepatoma are high, only the mobilities of the malignant cells are reduced by incubation with neuraminidase. H e therefore suggests that malignancy is in some way associated with sialic acid-mediated increases in cell-surface charge density. However, in contrast to Fuhrmann, Chaudhuri and Lieberman (1965) have shown that the increased mobility of cells from regenerating rat livers, does in fact result from increased amounts of surface sialic acids. The work quoted earlier also requires that blefore valid comparisions can be made between the mobilities of normal and malignant cells the cells themselves have similar growth rates and mitotic indices. Unless these conditions are met, it is impossible to attribute increased mobilities to malignancy per se. This latter suggestion is strongly supported by the work of Vassar (1963) and Vassar et al. (1967), in which no consistent difference was demonstrated between the mobilities of mechanically isolated cells from gastrointestinal carcinomata and adjacent normal epithelium, which also proliferates rapidly. Purdom et al. (1958) made a noteworthy attempt to relate the degree of malignancy, by which is usually meant the time taken to kill the host, to the electrophoretic mobility of tumor cells. In the series of murine tumors studied, cells with increased malignancy showed a trend toward having increased electrophoretic mobilities. The growth pattern of a tumor in a host depends partly on host response. If increased mobility reflects the presence of increased amounts of sialic acid moieties at the cell surface, it is to be asked whether these could in some way mask antigenic or other sites in this region and thereby prevent the

T H E CELL PERIPHERY

99

destruction of the tumor cells by the host defense mechanisms. Sanford (1967) observed that neuraminidase-treated TA3 mouse ascites tumor cells caused significantly fewer takes in mice than untreated controls. She interpreted these findings as indicating loss of transplantation specificity in the untreated neoplastic cells because of masking of surface isoantigens by sialomucins, which were removed by neuraminidase treatment. However, when this interpretation was tested by Hauschka and Weiss (1968) in our own laboratory with similar cells and mice, it could not be confirmed, since neither the untreated controls, nor the neuraminidase-treated cells adsorbed specific isoantibodies of the H-2 type. In addition, when the same cells were injected into six different strains of mice, their survival times varied from 7.2 to 12.6 days. In spite of these highly significant differences, no correlation could be found between the electrophoretic mobilities of the recovered cells, before or after incubation with neuraminidase, and survival time. It would thus appear that in this experimental system the total response of the host is not governed by a first-order relationship to the surface charge density of the tumor cell. The faster tumor cells grow in their respective hosts, the faster they can be expected to kill them. From the discussion on cell proliferation, it would be expected that faster growth rate might well cause higher mobilities, rather than the higher mobilities causing higher growth rates. However, our own studies on the TA3 tumor in different hosts, in which there was great variation in growth rates, did not reveal significant, correlative changes in electrophoretic mobility. At the moment, it therefore seems premature to correlate malignancy, or the degree of malignancy, with cellular electrophoretic mobility alone. This is hardly surprising when one considers the complexity of the malignant process, and its dependence on host/malignant cell interactions as well as the properties of the isolated malignant cells themselves.

REFERENCES Ada, G. L., and French, E. L. (1959). Nature 183, 1740. Ada, G. L., and Stone, J. D. (1950). Brir. J . Exptl. Pathol. 31, 263. Ambrose, E. J. (1961). Exptl. Cell Res. Suppl. 8, 54. Ambrose, E. J. (1965). “Cell Electrophoresis.” Churchill, London. Ambrose, E. J. (1967). Proc. Can. Cancer Re.r. Conf. 7, 247. Baker, J. R. (1958). 1. Historhem. Cyrochenz. 6, 303. Bakerman, S., and Wasemiller, G . (1967). Biochemistry 6. 1100. . 65. Bangham, A. D. (1963). Adz,an. Lil~idR ~ J1, Bangham, A . D., and Dawson, R . M. C. (1958). Nature 182, 1292. Bangham, A. D., and Haydon, D. A. (1968). Brit. Med. Bull. 24, 124. Bangham, A . D., and Pethica, B. A . (1960). Proc. Roy. Phys. Soc. (Edinburgh) 28, 43. Bangham, A. D., Pethica, B. A , , and Seaman, G. V. F. (1958). Biochem. J. 69, 12. Bangham, A . D., Glover, J. C., Hollingshead, S., and Pethice, B. A. (1962). Biochem. J . 84, 513.

100

LEONARD WEJ!IS

Bangham, A . D., Standish, M. M., and Watkins, J. C . (1965). 1. Mol. Biol. 13, 238. Barnard, E., and Stein, W. D. (1958). J . Mol. B i d . 1, 339. Barnard, P. J., Weiss, L., and Ratcliffe, T. M. (1969). Exptl. Cell Res, (in press). Beebe, S. P. (1904). Am. J . Physjol. 12, 167. Bennett, H. S. (1963). J . Hisfochem. Cytorhem. 11, 2. Ben-Or, S., Eisenberg, S.. and Doljanski, F. (1960). Nature 188, 1200. Benson, A . A . (1966). J . A m . Oil Chemists’ Soc. 43, 265. Berwick, K. (1959). Cancer Res. 19, 853. Bitensky, L. (1963). Brit. Med. Bull. 19, 241. Booth, F. (1953). Progr. Biophys. Biophys. Chem. 3, 175. Bowers, M. B. (1964). Protoplasma 59, 351. Branton, D. (1966). Pror-. Natl. Ar-ad. S r i . U.S. 55, 1048. Brinton, C. C., and Lauffer, M. A. (1959). I n “Electrophoresis” M. Bier, ed.), p. 42. Academic Press, New York. Brown, D . M., and Todd, A. R. (1955). Iiz “The Nuclcic Acids” (E. Chargatt and J. N. Davidson. eds.), Vol. 1, p. 409. Academic Preis, New York. Brunschwig, A., Dunham, L. J.. and Nichols, S. (1946a). Cavcer Res. 6, 230. 233. Brunschwig, A., Dunham, L. J., and Nichols, S. (1946b). Cancer Re.(. 6, 233. Bullivant, S., and Atnes, A. (1966). j . Cell Biol. 29, 435. Burka, E., Schreml, W.. and Kick, C. (1967). Biorllern. Biophys. Rer. Comrnun. 26, 334. Burnet, F. M., McCrea, J. F., and Stone, J. D . (1946). Brit. J. E x p t l . Pathol. 27, 228. Carruthers, C., and Suntzeff, V. (1944). Sc-ienre 19, 245. Carstensen. E. L., Fuhrmann, G . F.. Smearing. R. W., and Klein, L. A . (1968). Biorhim. Biophys. Aria 156, 39.4. CasleySmith, J. R. (1967). J . Roy. Mjrro,cop. Sor. 87, 463. Catalano, P., Nowell, P., Berwick, L.. and Klein. G . (1960). Bxprl. CeiJ Rrs. 20, 633. Chambers, R., and Fell, H . B. (1931). Pror. Roy. Soc. ( L o z d o n ) B109. 380. Chapman, D., Kamat, V. B., De Gier, J.. and Penkett, S. A. (1967). Nature 213, 74. Chaudhuri, S., and Lieberman, I. (1965). Expil. Cell Re.(. 40. 527. Clifford, J., Pethica, B. A., and Smith, E. G. (1968). I n “Membrane Models and Membrane Formation,” 1967 Frascati Conf. (L. Bolis and B. A. Pethica, eds.). NorthHolland Publ., Amsterdam. Clowes, G. H. A., and Frisbie, W . S. (1905). Am. j. l’hyrjol. 14, 173. Coman, D . R. (1953). Cavrcer Re.!. 13, 397. Coman, D. R., and Anderson, T. F. (1955). Cancer Re.r. 15, 541. Cook, G. M. W., Heard, D . H., and Seaman, G. V. 1.: (1962). E x p d . Cell Re.(. 28, 27. Cook, G . M. W., Seaman, G . V. F., and Weiss, L. (1963). Catjcer Res. 23, 1813. Cook, W . H., and Martin, W. G. (1962). CUE.J . Bd;oi-hem.Phy.rio1. 40, 1273. Crestfield. A. M., Stein, W. D., and Moore, S. (1963). J . B i d . Chem. 238, 2413. Criddle, R. S., Bock, R. M., Green, D. E., and Tisdale, H. (1962). Biochemniiiry 1,827. Criddle, R. S., Edwards, D . L., and Petersen, T. G. (1966). Bioi.hrmisiry 5, 578. Cunningham. W. P., and Crane, F. L. (1966). Exxpil. Cell Res. 44, 31. Danielli, J. F. (1937). Proc. Roy. SOL.. ( L o n d o n ) BL22, 155. Danielli, J. F. (1958). 1)) “Surface Phenomena in Chemistry and Biology” (J. F. Danielli, K. G. A. Pankhurst, and A. C . Riddiford, eds.), p. 216. Pergamon, Oxford. Danielli, J. F. (1962). Circulation 26, 1163, Danielli, J. F., and Davson, H. A. (1935). I . Cellular Comt). Physiol. 5. 695. Danielli, J. F., and Harvey, E. N . (191-1). 1. C e l l u h Co712p. Phpiol. 5, 483, 495. Davson, H . (1962). Circulation 26, 1022.

THE C E L L PI’RIPHERY

10 1

Dawson, R. M. C., and Bangham, A. D. (1959). Blochem. J . 72, 493. de Duve, C. (1959). I n ”Subcellular Particles” (T. Hayashi. ed.), p. 128. Ronald Press. New York. Defendi, V., and Gasic, G . ( 1963). I . Cellz~lurComp. Physiol. 62, 23. de Kloet, S. W., (1961). Studies on protein synthesis by protoplasts of saccharomyces carlsbergensis. Ph.D. Thesis, L’niv. of Utrecht. Deutsch, D. (1927). Sitzber. Akad. W’ii~. V’ien 60, 1036. Deutsch, D. (1928). Z . Physik. Chem. (Leipzig) 136, 3 5 3 . Dingle, J. T. (1961). Biol-hem. J . 79, 509. Dingle, J. T . (1968). Brit. Med. B d . 24, 141. Dingle, J. T., Fell, H. B., and Coornbs, R. R. A. (1967). Intern. Arrh. Allergy Appl. Immunal. 31, 283. Dumonde, D. C., Roodyn, D. B.. and Prose, P. H. (1065). In~munology9, 177. iiasty. G. C., and Mercer. E. H . (1960). Cancer Re.(. 20, 1608. Eisenberg, S., Ben-Or, S.. and Doljanski, F. (1962). E x p / l . Cell Rrs. 26, 4 5 1 . Elbers, P. F. (196.L).lit “Recent Progress in Surface Science” (J. F. Daniclli, K. G. A. Pankhurst, and A . C. Riddiford), Vol. 2 , p. ‘1.13. Academic Press, New York. Farquhar, M. G., and Palade. G . E. (1963). J . Cell B i d . 17, 375. Fawcett, D. W. ( 1965). J . Hntoi hem. C’yroc-hem. 13, 75. Fell, H. B., and hfellanby, E. ( 1 9 5 2 ) . I . Phy-iol. (London) 116, 320. Fell, H. B., and Weiss. L. (196-4). linpublished observations. Fell, H. B., and Weiss. L. ( 1 9 6 5 ) . /. Ex:xp/l. Ned. 121, 551. Fernindez-Morin. H., Oda, T., Blair, P. V., and Green, D . I . (1964). J . Cell B i d . 22, 63. Finean, J. B. ( 1 9 6 2 ) . Cimdntion 2. 1 1 5 1 . Finean, J. B., atid Burge. R. I . (1963). /. M o l . Bjol. 7? 672. Finean, J. B., and Robertson. J. D . ( 1 9 5 8 ) . Brit. Med. Btill. 14, 267. Fleischer, S., Fleisrher, H., and Stoeckenius, W . (1967). J . Cell Biol. 32, 193. Folch-Pi, J. ( 1967). 1’l’/J/fJ/JirU?Ud63, 160. Fraenkel-Conrat, H.. and Olcott. H. S. (1‘918). J . Ant. Chem. Sui.. 70, 2673. Freeman, J. A. ( 1 9 5 9 ) . J . Biophys. BioL-hem. Cy~ol.2, 353. Fuhrinann, G . F. ( 1965). 1n “Cell Electrophoresis” ( 1 . J. Ambrose, e d . ) , p. 92. Churchill, London. Gasic, G. J., Berwick, L., and Sorrentino. hf. (1968). Ldb. I n m i . 18, 63. Geren, B. B. (1954). Exptl. Cell Re.(. 7. 558. Gittens, G . J., and James. A. M. (1963). Biorhitn. Biophyr. Acts 66, 2 5 0 . Glick, P. M., Gorcn, H. J., and Barnard, E. A . ( 1 9 6 7 ) . Biorhem. 1. 102, 7C. Gorter. I:.. and Grmdel, F. ( 1 9 2 5 ) . 1. Exptl. M e d . 41, 439. Gorter, E . , and Grendel, F. (1926). Koninkl. N e d . Akad. Il~’etensi-hap., Proc. 29, 315. Gottschalk. A. ( 1 9 5 7 ) . Biul-hint. Biophys. Ai./a 23, 645. Green, D. I:.. and Fleischer, S. (1963). Bimhzni. B/ophg.r. Ac/a 70, 55-1. Green, D. E., and Perdue. J. F. (1966). P w i - . NN/(. A ~ - n dSi-i. . U.S. 55. 1295. Green, D. I:., l‘isdale, H. D., Criddle, R. S., Chen, P. Y., and Bock, R. M. (1961). Bio. rhern. Biophjs. Re.(. Conininii. 5 , 109. Green, D. E., Allmann, D. W., Bachmann, I!., Haum, H., Kopaczyk, K., Korman, I:. F., Lipton, S., MacLennan, D. H., McConnell, D. G.. Perdue. J. F., Rieske, J. S., and Tzagoloff’. A. (1967). Alch. Biol-hem. Biophys. 119, 3 1 2 . Hanig, M. (1948). YTOC.Sor. Exptl. B i d . Mcd. 68, 385. Hartley, G . S., and Roe, J. W. ( l O . 4 0 ) . 1’van.r. Farnday Sor. 36, 101. Hartridge, H.. and Peters, R. A. ( 1 9 2 2 ) . Proi;. Roy. Sol-. ( L o n d o n ) A101, 3-18.

102

LEONARD WE1S.S

Haughton. G. (1965). Science 147, 506. Hauschka, T.,and Weiss, L. (1968). Unpublished otiservations. Haydon, D. A. (1960). Proc. Roy. SOC. (London) h258, 319. Haydon, D. A. (1961). Biochim. Bjophys. Artu 50, 450. Haydon, D. A. (1964). In “Recent Progress in Surface Science” (J. F. Danielli, K. G. A. Pankhurst, and A. C. Riddiford, eds.), Vol. 1, p. 94. Academic Press, New York. Haydon, D. A., and Taylor, J. (1963). J . Theoret. Biol. 4, 281. Heard, D. H., and Seaman, G.V. F. (1960). J. Gen. Physiol. 43, 635. Heard, D.H., and Seaman, G. V. F. (1961). Biorhim. Biophyr. Arta 53, 366. Heard, D. H., Seaman, G. V. F., and Simon-Reuss, I. (1961). Nature 190, 1009. Henniker, J. (1949). Rev. Mod. Phys. 21, 322. Herbst, C. (1900). Arch. Entwirklun~.cmech.Organ. lVilhelm Roux 9,424. Hoffman, J. F. (1962). Circulation 26, 1073. Hulcher, F. H . (1963). Arch. Biorhem. Bzophys. 100, 237. Jahrisch, S. (1922). Biochem. 2. 134, 163. James, A. M. (1957). Progr. BiophyI. Biophys. Chem. 8, 96. Kagawa, Y., and Racker, E. (1966). J . Biol. Chem. 241, 2475. Kalant, H., Murray, R. K., and Mous, W . (1964). Cancer Res. 24, 570. Karrer, H . E. (1960). J. Biophys. Biochem. Cytol. 7, 357. Kartha, G. (1968). Personal communication. Katchalsky, A. (1964). Ciba Found. Symp. Intercelldar Macromolecules p. 9. Kavanau, J. L. (1965). “Structure and Function in Biological Membranes.” Holden-Day, San Francisco, California. Klenk, E. (1958). Ciba Found. Symp. Chem. B i d . Muropolysaccharides p. 311. Korn, E. D. (1966a). Biochim. Biophys. Arta 116, 3.17. Korn, E. D. (1966b). Biorhim. Biophys. Acta 116,325. Korn, E. D. (1967). J. Cell Biol. 34,627. Korn, B. D., and Weisman, R. A. (1966). Biochim. 13iophys. Aria 116, 309. Kraemer, P. M. (1967). J . Cell Biol. 33, 197. Lane, B. P., and Becker, F. F. (1966). Am. J . Pathol. 48, 183. Lansing, A. I., and Rosenthal, T. B. (1952). J . Cellular Comp. Physiol. 40, 337. Lansing, A. I., (1966). Personal communication. Lenard, J., and Singer, S. J. (1966). Pror. Natl. Acad. Sri. US.56, 1828. Lenard, J., and Singer, S. J. (1968). Science 159, 738. Lewis, G . N., and Randall, M. (1921). J. A m . Chezn. Soc. 43, 1112. Loewenstein, W . R. (1967). Develop. Biol. 15, 503. Loewenstein, W. R., and Kanno, Y. (1966). Nature 209, 1248. Lucy, J. A,, and Glauert, A. M. (1964). J. Mol. Biol. 8, 727. Lucy, J. A., Dingle, J. T., and Fell, H . B. (1961). Biorhem. J . 79, 500. Luzzati, V.,and Reiss-Husson, F. (1962). J . Cell B i d . 12, 207. Luzzati, V., Mustacchi, H., and Skoulios, A. (1957). Nature 180, 600. Luzzati, V., Mustacchi, H . , and Skoulios, A. (1958). Discussions Furaduy Soc. 25, 43. Luzzati, V.,Mustacchi, H., Skoulios, A., and Reiss-Husson, F. (1960). Artu Cryst. 13, 660. Lyklema, J., and Overbeek, J. T. G . (1963). J . Colloid Sci. 16, 501. McConnell, D.G., Tzagoloff, A., MacLennan, D. H., and Green, D . E. (1966). J. Biol. Chem. 241, 2373. McLaren, A. D., and Babcock, K. L. (1959). In “Subcellular Particles” (T. Hayashi, ed.), p. 23. Ronald Press, New York.

T H E CELL PERIPHERY

103

McLaren, A. D., and Estermann, I:. F. (1957). Arc-h. Biorhem. Bi0phy.r. 68, 157. Maddy, A. H., and Malcolm, B. R. (1965). Srieizre 150, 1616. Marcus, P. I. (1962). Cold Spring Harbor Syrnp. Quant. Biol. 27. 351. Marcus, P. I., Salb, J. M., and Schwartz, V. G . (1965). Nature 208, 1122. Maturana, H . R. (1960). J . BiophyJ. Biorhem. Cytol. 7, 107. Mayhew, E. (1966). J . Grn. Phgsiol. 49, 717. Mayhew, E., and Nordling, S. (1966). J . Cell. Phy.riol. 68, 75. Mayhew, E., and O’Grady, E. A. (1965). Nature 207, 86. Mayhew, E., and Weiss, L. (1968). Exptl. Cell Res. 50, d l . Mercer, E. H. (1963). In “Cancer Progress Volume” (R. W . Raven, ed.), p. 66. Butterworth, London and Washington, D.C. Merishi, J. N., and Seaman, G . V. F. (1966). Biochim. Biophys. Arta 112, 15.3. Miller, A.. Sullivan, J. F., and Katz. J, H. (1963). Cclncer Res. 23, 4 8 5 . Mueller. P., and Rudin, D. 0. (1963). J . Theorrt. B i d . 4, 268. Mueller, P., Rudin, D. 0..Tien, H. T., and Wescott, W. C. (1962). Nature 194. 979. Naaman, J., Eisenberg, S., and Doljanski, F. (1965). Lab. Inwst. 14, 1396. Nathanson, A. (1904). Jahrh. Y i s s . Botan. 39, 607. Nordling, S. (1967). Acta Pathol. Microbid. Srand. Suppl. 192, 1. Nordling, S., and Mayhew, E. (1966). Exptf. Cell Res. 44, 552. Nowell, P. C., and Berwick, L. (1958). Canrer Res. 18, 1067. Oncley, J. L. (1964). Pror. Intern. Symp. Lipid Transport, Nashidlr, 1963 p. 70. Osterhout, W. J. V. (1940). Cold Spring Harbor Synzp. Qtcant. Biol. 8, 51. Overbeek, J. T. G. (1950). A d ~ a n .Colloid Sci. 3, 97. Overton, J. (1962). Deimelop. Biol. 4, 532. Parpart. A . K., and Ballentine, R. (1952). In “Modern Trends in Physiology and Biochemistry” (E. S. G. Barron, etc.) p. 135. Academic Press, New York. Parsons, D. F. (1967). Proc. Can. Cancer Res. Conf. 7, 193. Parsons, D. F., and Akers, C . K. (1968). Personal communication. Pasternak, G . (1965). J . Natl. Canrer 1mt. 34, 71. Perutz, M. F. (1954). Pror. Roy. Soc. (London) A225, 264. Peters, R. A. (1960). J . Biophys. Biochem. Cytol. 7, 121. Peters, R. A. (1931). Pror. Roy. Sor. (Londom) A133, 140. Pulvertaft. R. J. V., and Weiss, L. (1963). J . Puthol. Bacteriol. 85, 475. Purdom. L., Ambrose, E. J.. and Klein, G. (1958). Nature 181, 1586. Reinders. (1910). Cited in Peters (1931). Rendi, R.. and Vatter. A . E. (1967). Protoplasma 63. 202. Revel, J.-P., and Ito, S. (1967). In “The Specificity of Cell Surfaces” (B. D. Davis and L. Warren, eds.), p. 211. Prentice-Hall, Englewood Cliffs, New Jersey. Richardson, S. H., Hultin, H. O., and Green, D. E. (1963). Pror. N a d . Arad. Sci. US. 50, 821. Rinaldini, L. M. (1958). Intern. Rri:. Cytol. 7. 587. Robertson. J. D . (1960). Progr. Bi0phy.r. Biophys. Chetri. 10, 343. Rogers, H. J. (1961). Biorhrm. Sor. Symp. (Canzbrid.ce, En,yl.) 20, 51. Roth. T. F., and Porter, K. R. (1964). J. Cell B i d . 20, 313. Ruhenstroth-Bauer, G., and Fuhrniann, G. F. (1961). 2. Naturforsrh. 16b, 252. Sanford, B. H . (1967). Transplantation 5, 1273. Schmidt, W. J. (1936). Z. Zellforjrh. Mikroskop. Anat. 23, 261. Schmidt, W. J. (1938). Kolloid-Z. 84, 137. Schrnitt, F. 0. (1941). Growth 5, Suppl., 1.

I04

LEONARD WEISS

Schmitt. F. O., Bear, R. S., and Clark, G. L. (1935). Radiology 25, 131. Seaman, G. V. F., and Cook, G. M. W . (1965). In “Cell Electrophoresis” (E. J. Ambrose, ed.), p. 48. Churchill, London. Seaman, G. V. F., and Heard, D. H. (1960). J. Gen. Physiol. 44, 251. Shah. D. 0. (1968). Biophys. J. Abstr. A-189. Simon-Reuss, I., Cook, G. M. W., Seaman, G. V. F,, and Heard, D . H. (1964). Cancer Res. 24, 2038. Sjostrand, F. S. (1963a). Nature 199, 1262. Sjostrand, F. S. (1963b). J. Ultrastrurt. Res. 9, 561. Sjostrand, F. S. (1967). Protoplasma 63, 248. Smith, D. S. (1961). J. Biophys. Biochem. Cytol. 10, 123. Steinbeg, M. S. (1958). A m . Naturalist 92, 65. Stoeckenius, W. (1962). J. Cell B i d . 12, 2 2 1 . Stoeckenius, W . , and Mahr, S. C. (1965). Lab. Znzwst. 14, 1196. Stoeckenius, W . , Schulman, J. H., and Prince, L. M. (1960). Kolloid-2. 169, 170. Straus, W. (1967). I n “Enzyme Cytology” ( D . 8. Roodyn, ed.), p. 219. Academic Press, New York. Sylvh, B., and Malmgren, H. (1957). Ac/a IZadiol. Scand. Su/~pl.154. Tevethia, S. S., Katz, M., and Rapp, F. (1965). Prar. Sor. Exptl. Biol. N e d . 119, 896. Thomas, L., McCluskey, R. T., Potter, J. L., and Weissmann, G. (1960). J. Exptl. Me d. 111, 705. Tieo, H . T., and Diana, A . L. (1968). Chem. Phy.1. Lipids 2, 55. Vassar, P. S. (1963). Lab. Inve.rt. 12, 1072. Vassar. P. S., Seaman, G. V. F., and Brooks, D. E. (1967). Proi. Can. Cancer Res. Conj. 7, 268. Vogt, P. K., and Rubin, H. (1962). Cold Spring Harbor Symp. Quant. B i d . 27, 395. Volkin, E., and Cohn, W. 1:. (1953). J. B i d . Chem. 205, 767. Wallach, D . F. H., and Eylar, E. H . (1961). Biorbim. Biophys. Aria 62, 594. Wallach, D . F. H., and Kamat, V. B. (1966). J . Cidl Biol. 30, 660. Wallach, D. F. H., and Perez-Esandi, M. V. ( 1 9 6 4 ) . Biochim. Biophys. Acra 52, 594. Warren, L., Glick, M. C., and Nass, M. M. K. (1967). In “The Specificity of Cell Surfaces” (B. D. Davis and L. Warren, eds.), p. 109. Prentice-Hall, Englewood Cliffs, New Jersey. Watkins, W . M. (1967). In “The Specificity of Cell Surfaces” (B. D. Davis and L. Warren, eds.), p. 257. Prentice-Hall, Englewood Cliffs, New Jersey. Weinstein, R. S., and Bullivant, S. (1967). Blood 29. 780. Weiss. L. (1960). Exptl. Crll Res. 21, 71. Weiss, L. ( I 9 6 l a ) . Exptl. Cell Re.r, Suppl. 8, 141. Weiss, L. ( 1 9 6 t b ) . Exptl. Cell Res. 25, 504. . 191, 1108. Weiss, L. ( 1 9 6 1 ~ )Nature Weiss, L. (1962a). J. T h e o w . B i d . 2, 236. Weiss, L. (1962b). Biochem. Sor. Symp. (Cambridge, Engl.) 22, 32. Weiss, L. (1963a). J. Gen. Microbiol. 32, 331. Weiss, L. (1963b). Exptl. Cell Rrs. 30, 509. Weiss, L. (1965~1).J , Cell Biol. 26, 735. Weiss, L. (1965b). Exptl. Cell ReJ-. 3‘1, 540. Weiss, L. (1966). J. Null. Cancer Inst. 36, 837. Weiss, L. (1967a). “The Cell Periphery, Metastasis, and Other Contact Phenomena.” North-Holland Publ., Amsterdam.

THE CELL PERIPHERY

Weiss, L. (1967b). Yroc. Can. Caticer Res. Conf. 7, 292. Weiss, L. ( 1 9 6 7 ~ ) J. . Cell B i d . 33, 341. Weiss, L. (1967d). J . Cell Biol. 35, 347. Weiss, L. (1968a). J . Theoret. Biol. 18, 9. Weiss, L. (1968b). Expil. Cell Res. 53, 603. Weiss, L. ( 1 9 6 8 ~ ) IP”.rtar . I v s / . S.ynip. Mo>io,qr. 8, 7.1. Weise, L., and Coombs. R. R. A. (1963). Exptl. Cell Res. 30, 331. Weiss, L., and Dingle, J . T. (1964). A m . Kheumatir DLreases 23, 57. Weiss, L., and Lachmann. P. J. (1964). Exprl. Cell Res. 36, 86. Weiss, L., and Mayhew, E. (1966). J . Cell. Phy.rio1. 68, 3/15, Weiss, L., and Mayhew, E. (1967). J . Cell. Physial. 69, 281. Weiss, L., and Ratcliffe, T. M. (1968). J. Natl. Canrer Itist. 41, 957. Weiss, L.,Bello, J., and Cudney, T. L. (1968). Intern. J. Cancer 3, 795. Weissmann, G. (1965). N e w Etigl. J. Med. 273, 1084, 1143. Wiersema, P. H.. Loeb, A . L.. and Overbeek, J. T. G. (1966). J. Colloid Interface Sci. 22, 78. Wigglesworth, V. €3. (1947). YYOC. Roy. Snr. (Lnadoiz) B147, 185. Winkler, K. C., and Bungenberg de Jong, H. G. (19.itl). Arch. Need. Physiol. 25, 431, 467. Yamoto, T. (1963). J . Cell B i d . 17, 413. Zeidman. I. ( 1947). Cizncer Res. 7. 386. Zetterqvist, H. ( 1956). “The Liltrastructural Organization of the Columnar Absorbing Cells of the Mouse Intestine.” Godvil, Stockholm.

This Page Intentionally Left Blank

Mitochondria1 DNA: Physicochemical Properties, Replication, and Genetic Function P. BORSTAND A. M. KROON D.epartment of Medrral Enzymology, Lciboratory o/ Bzorberni~t, Unri e+srty of Amstetdam, Ainiteidunz, The Netherlunds I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. Base Composition of Mitochondrial D N A . . . 111. Nearest-Neighbor Frequencies of Mitochondria IV. Differences in Base Composition and Base Sequence of the Complementary Strands of Mitochondria1 DNA's . . . . . . . . V. Size and Structure of Mitochondrial D N A from Animal Tissues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Introduction .................. .... B. Characterizati Mitochondria[ Closed DUplex D N A and Its Derivatives . . . . . . . . . . . . . . . . . . . . C. T h e Number of Superhelical Turns in Mitochondrial DNA ..........._.._..l......._...,........... D. Size and Circularity of Mitochondrial D N A from Animal Tissues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E. Oligomers of Mitochondrial D N A . . . . . . . . . . . . . . . . . F. The Behavior of Mitochondrial D N A in Alkali . . . . . . G . Composition of Mitochondrial D N A from Animal Tissues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI . Size and Structure of Mitochondrial D N A from Plants and Unicellular Organisms . . . . . . . . . . . . . . , . , . . . . . . . . . . . . . . VII. The Amount of Mitochondrial D N A per Mitochondrion and per Cell . , . . . . . . . . . . .......................... VIII. Replication of Mitochondrial D N A . . . . . . . . . A. Timing of Mitochondrial D N A Synthesis i the Cell Replication Cycle . . . . . . . . . . . . . B. Turnover of Mitochondrial D N A . C. T h e Mechanism of Mitochondrial Intact Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D. Incorporation of Deoxyribonucleotides into the D N A of Isolated Mitochondria . . . . , . . . . . . . . . . . . . . . . . . . . . . . IX. Effects on Yeast Mitochondrial D N A of Anaerobiosis, Glucose Repression, and Mutagenic Agents . . . . . . . . . . . . . . . . A . Anaerobiosis . . . . . , . . . . . . . . . . . . . . . . . . . B. Glucose Repression .......................... C. Mutagenic Agents . . . . . . . . . . . . . . . . . .. . X. Recombination of Mitochondrial D N A . . . . . . . . . . . . . . . . . XI. Renaturation Studies with Mitochondrial D N A XII. Evolution of Mitochondrial D N A and the Relation between Mitochondrial and Nuclear D N A . . . . . . . . . . . . . XIII. Genetic Function of Mitochondrial D N A . . . . . . . . . . . . . . . A. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . B. DNA-RNA Hybridization Experiments . . . . . . . . . . . . . . 107

108 109 117 117

118 118 121 128

130 133 136 137

139 143 145 145 146 149 152

154 154 155

156 163 165 167 168 168

169

108

P. BORST A N D A. M .

XIV.

KROON

C. The Product of Mitochondrial Protein Synthesis . . . . . . D. Identification of Mitochondrial Proteins Coded for by Nuclear DNA or Synthesized outside the Mitochondria E. Mitochondrial Enzymes Found in Cytophsmic Petite Mutants of Yeast . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . F. Correlations of Changes in Mito'chondrial Proteins with Changes in Mitochondrial DNA . . . . . . . . . . . . . . . . . . . G. Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . Addendum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

172 171 174 176 178 179 181

I. Introduction Research during the last two years has provided increasing evidence for the concept that the biosynthesis of functional mitochondria requires the cooperation of two genetic systems: the nuclear system, involving nuclear mRNA's, probably translated on extr'imitochondrial ribosomes, and a mitochondrial system localized in the mitochondrial matrix space. The genetic continuity and expression of the mitochondrial system appear to be ensured by the presence within the mitochondrial inner membrane of the enzymes required for D N A and RNA synthesis and the complete machinery required for protein synthesis. Since several vital parts of this machinery differ from their extramitochondrial counterparts (e.g., the ribosomes), the mitochondrial system for the synthesis of macromolecules is in part unique and not merely a copy of the extramitochondrial system stationed in the mitochondrial matrix space for the benefit of translating the informat'ion present in mitochondrial DNA. The fact that two genetic systems are involved in the biosynthesis of mitochondria raises obvious questions. What is the use of such a complicated setup for a cell that apparently manages to solve other complex cytological problems, e.g., the biosynthesis of peroxysomes or lysosomes, without invoking a second genetic system? What is the evolutionary advantage of having D N A in the cytoplasm? Why is it necessary to provide this D N A with a unique proteinsynthesizing system rather than put the extramrtochondrial system at its disposal ? How is coordination between the contributions of nuclear and mitochondrial D N A in the cell achieved? The answers to these questions are to be found in mitochondrial D N A and, in our opinion, a full understanding of the genetic function and evolution of mitochondrial D N A and the control of its replication and transcription will represent a major step toward an understanding of the biogenesis of mitochondria. It is the purpose of this review to discuss the possible genetic functions of mitochondrial D N A in the light of our present knowledge of the physicochemical properties of mitochondrial DNA's from a variety of sources. Several aspects of mitochondrial biogenesis have been reviewed in detail in

MITOCHONDRIAL DNA

109

recent years. The genetic aspects and general biological implications of cytoplasmic inheritance have been discussed by Jinks (1964), Wilkie (1964), Gibor and Granick (1964), Granick and Gibor (1967), and Roodyn and Wilkie (1968). The general problem of mitochondrial biosynthesis was recently reviewed by Luck (1965, 1966), Kroon (1966a), Tuppy and Wintersberger (1966), Granick and Gibor (1967), and Roodyn and Wilkie (1968), and exhaustively discussed at the Round-Table Discussion on Biochemical Aspects of the Biogenesis of Mitochondria held in Polignano in 1967 (Slater et ul., 1968). Since earlier work on mitochondrial D N A has also been reviewed by Swift (1965), M. M. K. Nass et cll. (1965), M. M. K. Nass (1967), and Borst et ul. (1967a), we have limited this review to a detailed and critical discussion of recent work on physicochemical properties, replication, and genetic function of mitochondrial DNA. 11. Base Composition of Mitochondrial DNA

The base composition of mitochondrial D N A from a variety of organisms has been studied by direct analysis by measuring the density in CsCl and by determining the T , [midpoint of melting curve of D N A (see Marmur and Doty, 1962)] under standard conditions. A summary of published results is presented in Tables 1-111. T o make results obtained in different laboratories comparable and to restrict the size of Tables I and 11, we have recalculated all buoyant densities to a common basis and grouped results of different authors together, as explained in the footnotes to the tables. From these data a number of useful generalizations emerge : (1) Mitochondrial D N A is double-stranded. For most of the cases presented in Tables 1-111, this follows either from melting behavior or from the characteristic increase of density in CsCl after denaturation. ( 2 ) Mitochondrial D N A is homogeneous in base composition. This follows from the narrow unimodal bands observed in CsCl and the sharp melting curves. The only exception to this generalization reported so far is Neztrospora mitochondrial D N A (Table 11). Neiiro.rporci rrussa mitochondria contain equal amounts of D N A components with equilibrium densities of 1.698 and 1.702 gm./cm.3, while Neziro.rpom sitophilu mitochondria contain a third major component banding at 1.692 gm./cm.". The cause o f this heterogeneity is not known. ( 3 ) Although the buoyant density of all mitochondrial DNA's studied varies between 1.683 gm./cm." (Tetrabymenu) and 1.716 gm./cm.3 (peanut), the density of mitochondrial D N A from related organisms is in all cases similar. (4) N o obvious relation exists between the densities of mitochondrial and nuclear D N A present in the same cell. They may be about the same (e.g., in rodents), or the density of mitochondrial D N A may be up to 2 1 mg./cm.3 lower

TABLE I DENSITY IN CsCI. T-. A N D BASECOMPOSITION O F MITOCHONDRIAL AND NUCLEAR DNA's

FROM

ANIMALTISSUES

T,, in SSC Density in CsCl (gm./cm.3)a

Mitochnndrial Nuclear

DNA Manirnals Man (Leukemic leukocytes) M a n (Chang liver cells) Man

OX

Sheep Rat Mouse ( L v e r ) Mouse (L cells) Giiinri pig Rabbit Birrls Chick Duck Pigeon

Amphibia Frog ( R a m p i p i e m ) Toad (Xenoprrs laeuis) Fish Carp Echinoderms Sea urchin

1.705 1.688

1.702 1.703 1.701 1.701 1.698 1.702 1.703

DNA 1.695 1.699 I .700 1.704 1.703 1.703

("C.) b Mito-

Nuclcar chonsatellite drial Nuclear DNA DNA DNA

Hase coniposition

Direct analysis

(5)GC)c

From density4

From

TmC

~

-

M

~

N

-

M

N

Referenccsb'

-

46

36

1

-

29

40

2

-

41

43 44

1'

3 4.5~5

I

87 85.6

42

42

5 44 14 44

I

1.701

-

1.703

81.7

1.TOO 1.701

-

1.708 1.711 1.707

1.701 1.700 1.700

1.702

1.701

1.702 1.702

1.703

1-69?

1.704

1.694

7 8,5,6,9,10 7,6,11.5.6,12

39

44

f

A*J

-I1

14-5

-

44

42

15

90.0

49

42 41

16,7.17,18

41

14

43 45

43 43

19,18,20,21

44

38

22

44

34

23

88

87.3 I

86.8

52 48

13

15

19,18,10,21

All densities were recalculated using the formula of Sueoka (1961) and a reference density of E . coli DNA 1 1.710 gin./cm.3. Tm = midpoint of melting curve (see Marmur and Doty, 1962); SSC = 0.15 M NaCI and 0.015 M sodium citrate (pH 7.0); the T , given for nuclear DKA is the T , of the major component of nuclear DNA. % GC = mole percentage guanine cytosine. 0.098 ( G C ) 1.660 gm./cm.-3, in which p = buoyant density i n gm.,:cm.-:’b. calculated as described Calculated with the formula p in footnote a, and (GC) = mole fraction guanine plus cytosine. This formula was dcrived empirical!). by Schjldkraut et al. (1962) for DNA’s containing only the four standard bases. Calculated with the formula T m = 69.3”C. 0.41 ( G C ) . derived empirically by Marmur and Doty (1962). f The values presented in the table are those given in. or recalculated from. the first reference given. The data presented in the other references quoted are considered not :o differ significantly from the value given here. K q 10 re!er.enr.c.r; ( 1 ) Clayton and Vinograd ( 1 9 6 7 ) ; ( 2 ) Koch and Stokstad (1967); ( 3 ) Corneo eral. (1967); (4) Kroon (1966b); ( 5 ) Corneo er a]. (1966); ( 6 ) Sinciair (1966); (7) Kroon e? al. (1966); ( 8 ) Schneider and Kuff (1965): ( 9 ) Suyama and Bonner (1966): ( 1 0 ) S. Nass ( 1 9 6 7 ) ; (11) Borst and Ruttenberg (196ha); ( 1 2 ) Sinclair and Stevens (1966); ( 1 3 ) hi. hf. K. Nass (1968); ( 1 4 ) Borst and Ruttenberg (1961%): ( 1 5 ) Borst el a!. (1967a); ( 1 6 ) Borst el al. (1967b); ( 1 7 ) Rabinowitz et al. (1965): (18) Dawid and Wolstrnholme (1968a); ( 1 9 ) Dawid (1965): ( 2 0 ) Daaid (1966); ( 2 1 ) Dawid and Wolstenholme (1967); ( 2 2 ) Van Bruggen ef a / . (1968): ( 2 3 ) Piko e t a ! . (1967). 0 Satellite detected i n whole-cell Dh’A; intracellular localization not known. a

(’

+

+

+

DENSITY IN CSCI.

T,,

A N D BASE COMPOSITION 01' -~

TABLE I1 MITOC~IONDKIAL AND NUCLEAR DNA's

T , in SSC ("C.) Dtnsity

111

CsCl ( g m . / c m . ~ ) ) "

Mitochondrial DNA

Nuclear DNA

Nuclear satellite DNA

Mitochnn-

FKOM ~ ~ K l C I ! l LiI.AR .I ORGAhtSMS

Base composition (% GC)c Direct

From densi tyd

analysis

drial Nuclear -DNA DNA M h

'

M

N

From

T," h

l

N

1.6gs+

.7 I 3

1.702

-

1.692+ 1.698+

.713?

-

-

1702

-

-

1.686

.700

12

1.706"

-633"

13

1.686 1.685

.692

1.686

.mz

,685

11

I1

26 25 26

33

14

25

14

31

14

-

0 - p See footnotes LO Table I. 1 The values presented in the tahle are those given in, or rrcdcutated from, the hrsr references given. The data presented in the other rcfcrences quoted are considered not to differ significantly from the value gixen here. Key t o rejeretrcer: ( 1) fdelman el al. ( 1966) ; ( 2 ) Edelman et al. (1965); ( 3 ) Ray and Hanawalt (1965); ( 4 ) Tewari ei d.(1966); ( 5 ) Corneo et a!. (1966); ( 6 ) Moustacchi and Williamson (1966): ( 7 ) Mounolou ?I al. (1966); ( 8 ) Carnevali ef al. (1966); (9) Borst et al. (1067a); (10) Hollenberg, Ruttenberg, and Borst, unpublished observations; (11) Reich and Luck (1966) ; ( 1 2 ) b.vnns (1966): ( 1 3 ) Suyama and Preer (1965); (14) Suyaina (1966): ( 1 5 ) Sinclair (1966). The nuclear localization of the satellite has not been demonstrated. The density difference observed between this satellite and nuclear DNA varics in different lnborntories between d mg./crn.3 (Cnrneo et mi., 1966) and I I rng.icm.3 (Carnevali er al.. 1966). h The valucs given by rhr authors are 1.702 and 1.689 grn./cm,3. However, in a later paper Suyama (1966) menttons that, owing to a calibration error. the densities given for Trrrahywend DNA by Suyama and Preer (196j) were 4 mg./cm 3 too low. We have assumed that the same Ilolrls for the denuty vf Purume~~jrrmDNA reported in the same paper by Suvama and Preer.

5

c

n

114

P . BORST A N D A . M. KROON

TABLE 111

DENSITY I N CSCl OF MITOCHONDRIAL AND NUCLEAR DNA's

FROM

HIGHER PLANTS

Density in CsCl (gm./cm.3) Nuclear DNAa Marigold (Tagetes patula) Tobacco (Nicotiuna tubacrum) Sweet potato (Ipomoea b a t a t a ) Mung bean (Phaseolus aureus) Turnip (Bra.r.rica rapu) Onion (Allium cepa) Spinach (Spinarea olerureu) Spinach (Spinucea olerarea) Beet (Bela vulguri.r) Swiss chard (Beta vulgaris var. cicla) Lettuce ( L t u l - a sativu) Broad bean (Viciu fuhu) Sweet pea (Lathyrus 0doi.alu.i ) Peanut (Ararhis hypoxaeu)

Chloroplast DNA

1.692 1.696 I ,692 1.691 1.692 1.688 1.695 1.692 1.605

1.702 1.706 -_

1.705? 1.6515 1.705

1.690 1.692 1.692 1.692 1.705

1.700 1.695 1 . 6 ~ 1.605

_. _. _.

__

Mitochondrial DNA 1.707? 1.71n? 1.706 1.706 1.706

Analytical method Referusedh ences"

I.719? 1.705 1.719?

A A € 3 B B B A C A

3 3 3 3 4,5 6 4

1.705? 1.705 1.70) 1.705 1.716

D C C C E

7,s 6 i 6 9

1.706(1.718)

c

1 1.2

a The nuclear D N A of higher plants may contain up to 6% methyl-C. The replacement of C by methyl-C lowers the density of the D N A (Kirk, 1967). 1, ( A ) Reference E . c-oli D N A = 1.710 gm./cm.3; density calculated by Sueoka's formula (Sueoka, 1961). ( B ) As ( A ) . In a later paper Suyama (1966) mentions that, owing to a calibration error, the previously reported density for Tetrahytnena D N A was incorrect by 4 mg./cm.3. Since this Tetrahymena D N A was also used as a second reference D N A in the studies on plant DNA, it is probable that the densities of Suyama and Bonner presented in this table must also be increased by 4 mg./cnl.3. ( C ) Methods not specified. ( D ) As ( A ) hut density calculated according to Vinograd and Hearst (1962). [Contrast Sinclair (1966) in the same laboratory.] ( E ) . Reference Pseudomonas aerugjnosa D N A N15; density calculated by Sueoka's formula (Sueoku, 1961). C Key / o references: ( 1 ) Green and Gordon (1967); ( 2 ) Green and Gordon (1966); ( 3 ) Suyama and Bonner (1966); (4) Chun et ul. (1963); ( 5 ) E. Englert, quoted in Green and Gordon (1967); ( 6 ) Wells and Birnstiel (1967); ( 7 ) Swift (1965); ( 8 ) Kislev et a/. (1965); ( 9 ) Breidenbarh ~t ul. (1967).

(N.dophiku) or up to 16 mg./cm.3 higher (Englend gmciilis) than that of nuclear DNA. ( 5 ) Direct analysis of the base composition of mitochondrial D N A from yeast, rat, mouse, and chick tissues has shown an approximate molar equivalence of A with T and of G with C. No unusual bases were detected at the 1 % sensitivity level but the presence of lower amounts of methylated bases has not been studied. The absence of major amounts of unusual bases in mitochondrial D N A from sheep, frog, toad, sea urchin, and Tetruhymenu mitochondrial D N A

MITOCHONDRIAL DNA

115

is suggested by the agreement of the base composition calculated from the density of these DNA’s in CsCl and their T , in 0.1 5 M sodium chloride, 0.015 M sodium citrate (Tables I and 11). Why the T,,, of yeast mitochondrial DNA is so much lower than that expected on the basis of the base composition of this DNA is not clear (Table 11). Early work on the yeast mitochondrial L-lactate dehydrogenase (cytochrome b,) suggested that this enzyme contains a specific DNA of low molecular weight. More recently, Burgoyne and Symons (1966) have conclusively shown, however, that the association of DNA with this enzyme is the result of a nonspecific binding to the enzyme of small pieces of DNA formed in autolyzing yeast. Whether or not the high proportion of 5-methylcytosine in the nuclear DNA of some plants is also present in the mitochondrial DNA of these plants has not been determined. Although in most cases buoyant densities determined for mitochondrial DNA in different laboratories agree quite well, discrepancies were noted in four cases. The density of mitochondrial DNA from sheep heart was first reported by Kalf and GrPce (1966) to be identical with that of the nuclear satellite DNA of sheep tissues ( 1.714 gm./cm.3). Kroon et al. ( 1966) subsequently showed that the purified closed circular duplex fraction of sheep heart mitochondrial DNA banded at 1.703 gm./cm.3 without any material at 1.714 gm./cm.3, while the 1.714-DNAwas only present as a trace contaminant in crude DNA preparations from sheep heart mitochondria. From these results, it is likely that the value of 1.714 for sheep mitochondrial DNA is the result of an experimental error and we have, therefore, omitted it from Table I. W e have also omitted the value reported in a brief note by Parsons and Dickson (1965) for mitochondrial DNA from Tetrdymetza pyifofarmis, syngen 6 (1.671 gm./cm.:i against 1.685 gm./cm.3 found by Suyama) . It seems likely that in this case a carbohydrate peak (cf. Counts and Flamm, 1966) was mistaken for mitochondrial DNA in the density gradient. The situation is less clear with mitochondrial DNA from human tissues and plants. Clayton and Vinograd (1967) have reported a density of 1.705 gm./cm.3 for mitochondria1 DNA from the leukocytes of three different patients with leukemia. The DNA was isolated on the basis of its restricted uptake of ethidium bromide in a preparative CsCl gradient; it consisted, therefore, exclusively of closed circular duplex DNA and, since only circles of 5 p or a multiple of 5 were present in electron micrographs of this DNA fraction, the conclusion that it was derived from the leukocyte mitochondria, seems well-founded. Also, the density of 1.705 gm./cm.?. of this DNA agrees well with the density found for mitochondria1 DNA from other mammals. In contrast with these results, Koch and Stokstad (1967) have isolated a DNA with a density of 1.688 gm./cm.3 from the mitochondrial fraction

116

P . BORST A N D A. M . KROON

of cultured human Chang liver cells. A minor satellite component of this density, comprising less than 1% of the total cellular D N A and of unknown subcellular localization, had previously been detected in human bone marrow cells by Corneo et al. (1967). This makes it rather unlikely that the D N A could be derived from contaminating bacteria or viruses. In addition, more recent work of Koch and co-workers (private communication) showed that the isolated D N A fraction contained circular D N A with a contour length of about 5.3 p. A major fraction of this D N A banded at the position of closed circular duplex D N A in CsCl gradients containing ethidium bromide. It is difficult to reconcile these results with those of Clayton and Vinograd. If the D N A banding at 1.688 g m . / ~ m . ~ is indeed the mitochondrial D N A from Chang liver cells, it must have undergone a radical change in base composition in respect to the liver cells from which these cells were originally derived, assuming that the equilibrium density of normal human mitochondrial D N A is 1.705 g m . / ~ m . ~A. shift in density of 17 mg./cm.3 could arise from the presence of a fixed amount of covalently bound protein or carbohydrate, the substitution of all C by methyLC, or a drastic decrease in GC content. Changes in GC content have been observed in cytoplasmic petite mutants of yeast (see Section I X ) but in these cases no functional mitochondria are formed, while mitochondrial energy-yielding reactions have been shown to be normal in all tumors studied (see reviews by Borst, 1961; Wenner, 1967). It will be of interest to study the base composition of mitochondrial D N A from normal human tissues. Considerable controversy exists regarding the density in CsCl of rnitochondrial D N A from higher plants, as shown in Table 111 which includes data for chluroplast DNA. N o attempt has been made to recalculate the densities to a common basis in this case for lack of data but, when possible, the method used for buoyant density determination has been indicated. The main difficulty is the distinction between chloroplast and nuclear DNA. According to Wells and Birnstiel, the chloroplast D N A of spinach, lettuce, broad bean, and sweet pea has a buoyant density only 3 n i g . / ~ m .higher ~ than the density of nuclear DNA, while mitochondria1 D N A bands at 1.705 gm./cm.3 They suggest that the DNA identified as chloroplastal by Chun et al. and Shipp et al. (1965) was in fact mitochondrial. They support this suggestion by renaturation studies of their chloroplast D N A which show it to renature much faster than nuclear D N A but slower than the mitochondrial D N A from animal tissues. The only other study of plant extranuclear D N A in which the D N A was characterized by renaturation analysis was that of Tewari and Wildman (1766). In this case, no quantitative renaturation studies were done and it is therefore not possible to decide from their results whether they studied mitochondrial or chloroplast DNA. Moreover, it is noteworthy that they found a difference in density between nuclear and chloroplast DNA in tobacco of only 5 m g . / ~ m . ~which , is closer to the 3 mg./cm.3

MITOCHONDRIAL D N A

117

found by Wells and Birnstiel for other higher plants than the differences of 10 and 13 mg./cm.“ found by Green and Gordon and Shipp el ul., respectively. Wells has suggested that the “nuclear DNA contamination” observed by others in their purified chloroplast fraction was in fact mainly chloroplast DNA, while the enriched satellite bands were the result of enrichment of mitochondria in the chloroplast fraction. Obviously more work should be done on the mitochondrial DNA of plants, and in this work close attention should be paid to the clean separation of subcellular components using marker enzymes to assess the degree of cross-contamination and quantitative renaturation studies to characterize the DNA isolated. Very recently, Whitfeld and Spencer (1968) have also shown that the densities of chloroplast and nuclear DNA are identical in tobacco (1.697 gm./cm.3) and very similar (1.696 and 1.694 gm./cm.s) in spinach. This confirms the conclusions of Wells and Birnstiel. 111. Nearest-Neighbor Frequencies of Mitochondrial DNA

The nearest-neighbor frequencies ending on G were determined by Cummins et al. (1967) for mitochondrial and nuclear DNA of the slime mold Physurmz polycephalrLm, using RNA copies of these DNA’s made in v h o with RNA polymerase. While nuclear DNA contained the low CpG content characteristic of the nuclear DNA of all eucaryotic organisms (Swartz et ul., 1962), the mitochondrial CpG content did not differ significantly from random. Since bacterial DNA’s are also characterized by a CpG content equal to, or greater than, that expected for a random base sequence (Swartz ef nl., 1962), the authors conclude that their observations add another item to the list of similarities between mitochondria and bacteria. It remains to be demonstrated, however, that RNA polymerase makes complete copies of both strands of mitochondrial DNA. It wiIl be of interest to determine the nearest-neighbor frequencies of other mitochondrial DNA’s to see whether a difference between nuclear and mitochondrial DNA is also found in animals in which the difference in base composition between the two is much smaller than the 15% GC found in P . polycephalum.

IV. Differences i n Base Composition and Base Sequence of the Complementary Strands of Mitochondrial DNA’s DNA of several bacteriophages forms two bands in neutral CsCl after denaturation. Marmur and Cordes (1963) have shown that this is because of an unequal distribution of purines and pyrimidines over the complementary strands, the heavy strand being relatively pyrimidine-rich. No band splitting in neutral CsCl has been observed for any mitochondrial DNA after denaturation. How-

118

P. BORST AND A. M. KROON

ever, Dawid and Wolstenholme (1967) have reported that in alkaline CsCl equilibrium gradients toad mitochondrial DNA separates into two bands that differ by 13 mg./cm.3 in density. An even much greater difference (31-33 mg./ cm.3) was found by Smit and Borst (unpublished observations) for rat liver mitochondrial D N A in alkaline CsC1. It seems very likely that the two bands represent the complementary strands of mitochondrial DNA. This point is being further pursued in our lhoratory since the preparative separation of the complementary strands of mitochondrial D N A will be of interest both for experiments on mitochondrial transcription and for an analysis of the evolution of mitochondria1 DNA’s. Recently, Ruttenberg and Borst (1968) have attempted to separate the complementary strands of chick liver mitochondrial DNA in CsCl by complexing them with polyribonucleotides, according to the technique developed by Szybalski (Kubinski et al., 1966; Hradecna and Szybalski, 1967). In the presence of poly-U, the buoyant density of denatured mitochondrial D N A in CsCl increased from 1.723 to 1.751 gm./cm.:’; in the presence of poly-IG of different molecular weights, the increase in density varied between 8 and 13 mg./cm.3. Strand separation was obtained with neither of the ribopolynucleotides.

V. Size and Structure of Mitochondria1 DNA from Animal Tissues A. INTRODUCTION Early in 1966, Borst and Ruttenberg (1966a) and Van Bruggen et a / . (1966) reported that mitochondrial D N A from chick liver, mouse liver, and ox heart consists of a homogeneous population of circular molecules with an average contour length of 5.45 p, equivalent to a molecular weight of 10-11 x 106 daltons (sodium salt). Circular D N A was independently observed in mouse liver mitochondrial D N A by Sinclair and Stevens (1966) and in L-cell mitochondrial D N A by M.M.U. Nass (1966). Two major types of circular molecules were observed by Van Bruggen et nl. (1 966) in electron micrographs of purified chick liver mitochondrial D N A : open (or half open) circles as shown in Fig. 1A and highly twisted circles as shown in Fig. IB, the former sedimenting with an J ~ , ,= , ~ 27 S (component I I ) , the latter with an .r2,,,,,, = 39 S (component I ) , as shown in Fig. 2. The authors suggested that the twisted circles represent the closed circular duplex form of mitochondrial DNA in which FIG. 1. A. Electron micrograph of an open chick liver mitochondrial DNA circle, spread according to the Kleinschmidt protein monolayer technique (cf. Van Bruggen et ul., 1968). B. Electron micrograph of a twisted chick liver mitochondrial DNA circle (cf. Van Bruggen et ul., 1968). Markers indicate 0.2 p.

MITOCHONDRIAL D N A

119

120

P . BORST AND A. M. KROON

both strands are covalently continuous, while the open circles represent molecules with one or more single-strand breaks, in analogy with the situation observed earlier with polyoma D N A (Vinograd et al., 1965). I n the last two years the conclusions of Van Bruggen et al. (1966) have been confirmed and extended in a number of laboratories, mainly by Sinclair and coTop

Meniscus

Bottom

FIG. 2. Band sedimentation of mitochondria1 DNA through neutral CsCl in the analytical ultracentrifuge. Top: densitometer tracing of ultraviolet absorption photograph of chick liver DNA 3 2 minutes after reaching full speed (from Borst el al., 1 9 6 7 ~ ) Bottom: . densitometer tracing of ultraviolet absorption photograph of rat liver DNA 28 minutes after reaching full speed (unpublished experiment of E. M. Smit).

workers (Sinclair and Stevens, 1966; Sinclair, 1966; Sinclair et al., 1967a.b ; Swift et al., I968b) ; Borst and co-workers (Kroon et al., 1966; Borst et al., 1()67a,b,c; Borst et al., 1968; Ruttenberg et al., 1968; Van Bruggen et a/., 1968), Dawid and co-worker (Wolstenholme and Dawid, 1967; Dawid and Wolstenholme, 1967, 1968a,b) and Vinograd and co-workers (Radloff r t al., 1967; Hudson and Vinograd, 1967; Clayton and Vinograd, 1967; Piko rf a/., 1967). The present status of the field can be summarized by three points: (1) Mitochondria1 DNA of all animals analyzed consists of circular molecules homogeneous in size. ( 2 ) The major part of this D N A is present in situ as closed circular duplex DNA, On extraction, this D N A is obtained in a compact, twisted form containing right-handed superhelical turns. In the case of chick and rat liver mitochondrial D N A the number of twists per unit length is about the same as in polyoma D N A and replicative form DNA of phage @XI74.

MITOCHONDRIAL D N A

I21

In addition, two other classes of molecules are found in D N A isolated from mitochondrial preparations, open circles and multimers of mitochondrial DNA. All available evidence indicates that most of the open circles are derived from closed circles during the isolation of the mitochondria and the purification of the DNA. The proportion of multimers is low in all normal cells analyzed, but in tumor cells up to 50% of all mitochondrial D N A may be present as multimers. ( 3 ) The size of D N A from a wide range of animals is remarkably constant and varies around 5-5.5 p. The evidence for these points is discussed below.

B. CHARACTERIZATION O F MITOCHONDRIAL CLOSEDCIRCULAR DUPLEX D N A AND ITS DERIVATIVES The characteristic properties common to all circular duplex DNA’s are best illustrated by a numerical example. Consider a closed circular duplex containing 110 base pairs. In the A configuration with about 11 base pairs per turn of the Watson-Crick helix the molecule contains 10 complete revolutions of one strand around the other. A change in configuration to the B-configuration, with 10 base pairs per turn of the helix, requires formation of one additional complete revolution of one strand around the other. This can only be accomplished in a closed circular duplex by introducing a right-handed superhelical turn in the molecule as a whole. By definition, one superhelical turn leads to the winding of one turn of the Watson-Crick helix and when such a molecule is adsorbed onto a protein monolayer, it will contain, in principle (see below), one crossover similar to a figure eight. Early denaturation of D N A leads to an increase in the number of base pairs per turn of the Watson-Crick helix and therefore to an unwinding of right-handed superhelical turns until the circle is completely open. With increasing denaturation, left-handed superhelical turns are introduced. The molecule with superhelical turns has a higher free energy than an identical molecule without superhelical turns. As a consequence, any process that leads to unwinding of superhelical turns occurs more easily in the twisted circular duplex than in a linear D N A molecule of the same base sequence; the reverse holds for any process leading to the introduction of superhelical turns. One single-strand break in one of the strands of a closed duplex circle is sufficient to release any superhelical turns present, because a “swivel” is created at which free rotation of the two ends, in relation to each other and the other strand, is possible. This brief introduction of the properties of closed circular duplex DNA’s may serve as background for the experimental results discussed in this section. A detailed discussion of this problem can be found in the papers of Vinograd and Lebowitz (1966), Vinograd et al. (196S), Bauer and Vinograd (1968), Wang et al. (1967), and Wang (1969). All closed circular duplex DNA’s synthesized in the intact cell contain righthanded superhelical turns when analyzed in vitro. This results in a series of

122

P. BORST AND A. M . KROON

highly characteristic properties which are listed in Table IV. A number of these properties have been studied for mitochondrial D N A from chick liver, rat liver, frog, t a d , and sea urchin eggs, and malignant cells of human origin. The results obtained are summarized in Table IV and will be briefly discussed. ( 1 ) “Twisted” circular molecules have been observed in electron micrographs of mitochondrial D N A from all animal tissues listed in Table VI. Unfortunately, with the comparatively large circles of mitochondrial D N A a variable degree of entangling also occurs during spreading of open circles. Therefore, the presence of closed duplex molecules in mitochondrial D N A can only be determined by electron microscopy if rigid criteria are developed to distinguish between twisted molecules JBZJZI stricto and entangled open molecules, and if these criteria are verified on samples of purified open circular and closed circular duplex D N A isolated by preparative band sedimentation. As yet, this has only been done by Borst et al. ( 1 9 6 7 ~ )and ~ they showed that with their classification of molecules in electron micrographs only 4% of the molecules of an open circular duplex D N A sample were scored as twisted, while more than 80% of the molecules of a purified closed circular duplex D N A sample were found to be twisted. More recently, Ruttenberg et al. (1968) have shown that the distinction of closed and open circular D N A by electron microscopy can be greatly simplified by spreading the D N A samples on low salt containing a high concentration of ethidium bromide. Under these conditions the configuration of the closed molecules is so characteristic that they cannot possibly be confused with entangled open circles (see below) . ( 2 ) As shown in Fig. 2, mitochondrial D N A from chick and rat liver consists of variable proportions of two homogeneous components sedimenting at 39 S (component I ) and 27 S (component 11). Similar components have been identified in mitochondrial D N A from duck liver (Kroon et al., 1966), sheep heart (Kroon et al., 1966), toad eggs (Dawid and Wolstenholme, 1967), sea urchin eggs (Piko et al., 1967), and human leukemic leukocytes (Clayton and Vinograd, 1967). Treatment with pancreatic deoxyribonuclease of mitochondrial D N A from chick liver (Borst et al,, 1967b), rat liver (Smit and Borst, unpublished observations), or amphibian eggs (Dawid and Wolstenholme, 1967) leads to the conversion of component I into component 11. In the last-mentioned case, conversion was shown to follow single-hit kinetics, in agreement with the concept that one single-strand break is sufficient to convert component I into component I I. (3-6) Limited denaturation of a closed circular duplex D N A leads to unwinding of the right-handed superhelical turns and to a concomitant drop in sedimentation coefficient to that of the open circle. When the degree of denaturation increases, the sedimentation coefficient rises again, probably because further unwinding of the Watson-Crick helix introduces left-handed superhelical

TABLE IV GENERAL PROPERTIES OF CLOSED CIRCULAR DUPLEX D N A OBSERVED FOR INTACT MITOCHONDRIAL DNA

OF

VARIOUS SOURCES

Mitochondria1 DNA source

( 1 ) Twisted circular moleculrs present in electron

micrographs ( 2 ) 3 9 4 DNA converted into 27-S DNA by one or more single-stranded scission ( 3 ) Dip in rhc sedimentation velocity mclting curve ( 4 ) Elevated T,a ( 5 ) Elevated pH, for the alkaline transitionn (6) Elevated sedimentation coeficient in strandseparating solventsa ( 7 ) Elevated buoyant density in alkaline CsClo ( 8 ) Titration with ethidium bromide leads to a characteristic sedimentation velocity curve ( 9 ) Drcreased capacity for binding intercalating dyes such as ethidium bromide'L 0,

Chick liver

Rat liver

+

3-

+ + +

+

Frog or toad eggs

Sea urchin eggs

HeLa cells

Human leukemic leukocytes

+ +

+

+

-t

=i

0 n

+

4-

+

-k

-k

+

+

+ +

4-

'x

+ +

+

+

In relation to a linear DNA of the same base composition and and molecular weight.

c h)

w

124

P. BORST A N D A. M. K R O O N

turns in the molecuIe as a whole. This sequence of events has been demonstrated to occur for component I of mitochondrial D N A by studying its sedimentation coefficient as a function of p H or heating temperature in the presence of formaldehyde. As shown in Fig. 3, the sedimentation coefficient of frog egg mito-

8

12

II

13

PH

FIG. 3. Sedimentation coeficients of mitochondrial DNA from toad eggs ( X . laevjs) as a function of pH. The graph shows the uncorrected sedimentation coegcients determined by analytical band sedimentation in CsCl with a density of 1.33. Open circles, component I; solid circles, component I1 (from Dawid and Wolstenholme, 1967).

chondrial D N A component I decreases between pH 11.5 and 12; above p H 12, it rises to a value of 87 S for the alkaline supercoil at pH 13. Similar alkaline supercoil forms of mitochondrial D N A from chick and rat liver have been observed by Smit in our laboratory. The effect of heating chick liver mitochondrial D N A in the presence of formaldehyde on the Sedimentation coeficient of I and I1 is presented in Fig. 4. Around 50OC. the sedimentation coefficient of I decreases, indicating the unwinding of the right-handed superhelical turns. With higher heating temperatures, the sedimentation coefficient rises to a maximum of 8 3 s. As shown in Fig. 4, heating of component I1 to 60OC. or higher in the presence of formaldehyde Ieads to the appearance of two components with sedimentation coefficients of 32 and 28 S, tentatively identified as the single-stranded ring and the single-stranded broken ring (Borst et al., 1 9 6 7 ~ ) .In principle, these may also be expected to be present when component I1 is sedimented in alkali, but with toad egg mitochondrial D N A only one band was found (Dawid and Wolstenholme, 1967). Figures 3 and 4 illustrate two other characteristic properties that mitochondrial closed circular D N A shares with other closed circular duplex DNA's: The pH

MITOCHONDRIAL DNA

125

necessary for complete denaturation is more than 0.5 p H unit higher than for the “nicked,” open circle (Fig. 3 ) . The experiment presented in Fig. 4 demonstrates that the T , of component I is much higher than the T,, of component 11.

FIG. 4. Sedimentation coefficients of mitochondrial DNA from chick liver heated for minutes at different temperatures in the presence of formaldehyde. Continuous line: component I; broken line: component I1 (modified from Borst et al., 1 9 6 7 ~ ) . 10

( 7 ) An elevated equilibrium density in alkaline CsCl has been observed for polyoma D N A and the replicative form of phage 0 x 1 7 4 (see Vinograd and Lebowitz, 1966). Attempts to duplicate this result with mitochondrial D N A have failed because all samples of mitochondrial D N A isolated so far have been too unstable in alkali to survive a 20-hour CsCl run at pH 1 3 without strand breakage. (8-9) Intercalation of ethidium bromide between the base pairs of duplex DNA leads to unwinding of the helix, and the degree of unwinding is a function of the amount of ethidium bromide intercalated (Radloff et a]., 1967; Crawford and Waring, 1967; Bauer and Vinograd, 1968). Therefore, the sedimentation coefficient of component I decreases in the presence of low concentrations of ethidium bromide. At higher concentrations it increases again because the increasing degree of unwinding probably leads to insertion of left-handed superhelical turns. That this is the case for chick liver mitochondrial component I is

126

P. BORST A N D A. M. KROON

shown by the experiments presented in Fig. 5. Similar results have been obtained with rat liver mitochondrial DNA. At very high ethidium bromide concentrations less ethidium bromide can be intercalated into component I than into component I1 or linear D N A (Radloff et al., 1967; Bauer and Vinograd, 1968). Since the buoyant density of the ethidium bromide-DNA complex in CsCl is

301

.E

f

I \

5

.-

ul

k

Mitochondria1 DNA

I

L-

25

10

'

0

I

I

I

I

005

0 10

0 15

0 20

J 025

Ethidium Br present per nucleotide (molehnole)

FIG. 5 . Sedimentation coefficients of chick liver mitochondrial D N A and polyoma DNA as a function of ethidium bromide concentration. Open circles, component I; triangles, component 11; solid circles, only one component detectable (from Ruttenberg et al., 1968; the data for polyoma D N A were determined by Crawford and Waring, 1967).

lower than that of D N A alone, the buoyant density of component I of SV 40 D N A in CsCl in the presence of saturating concentrations of ethidium bromide is nearly 50 mg./cm.a higher than that of component I1 (Bauer and Vinograd, 1968). A similar difference in density has been observed under these conditions for mitochondrial D N A from HeLa cells (Radloff et al., 1967; Hudson and Vinograd, 1967), human leukocytes (Clayton and Vinograd, 1967), and chick and rat liver (Borst and Smit, unpublished results). In view of these results there is no doubt that components I and I1 represent the closed and open circular duplex forms of mitochondrial DNA. The fact that slight denaturation of component I decreases its sedimentation coefficient to that of component 11 can only be rationalized by assuming that component I contains

MITOCHONDRIAL DNA

127

right-handed superhelical turns in solution. This conclusion is confirmed by the observation that in chick liver component I all analyzable twists in stereoelectron micrographs are right-handed (Van Bruggen, unpublished observations, 1967). The superhelical turns result in a compact structure which sediments about 40% faster than the open circle with the same molecular weight. Conversion of a circular duplex molecule into a linear duplex of the same molecular weight leads to a 10-15% decrease in sedimentation coefficient (see Vinograd and Lebowitz, 1966). Treatment of frog egg mitochondrial D N A with deoxyribonuclease 11, the enzyme that cuts both strands of duplex D N A at the same site, led to the formation of the expected component with a sz0,,, of 24 S (Dawid and Wolstenholme, 1967). A similar component was detected in “aged” chick liver mitochondrial D N A by Borst et al. ( 1 9 6 7 ~ ) .It seems likely that this component represents the linear form of mitochondrial DNA. The relations between the different forms of chick liver mitochondrial D N A discussed in this section are schematically indicated in Fig. 6, adapted from a Denatured DNA

N a t i v e DNA

1.39

s

27 S

I

x

J.

24 S

FIG. 6 . Diagrammatic representation of the various forms of mitochondrial D N A , modified from a similar diagram for polyoma DNA of Vinograd ef al. (1965). T h e denatured forms are those observed after complete denaturation by heating in the presence of formaldehyde.

similar scheme for polyoma DNA by Vinograd et a/. ( 1965). The sedimentation coefficients reported for mitochondrial D N A from chick and rat liver, amphibian eggs, and human leukemic leukocytes are presented in Table V. The agreement is excellent, supporting the idea that there are neither differences in structure nor in molecular weight between these three mitochondrial DNA’s. The relative

128

P. BORST A N D A. M . KROON

sedimentation coefficients of the various forms of mitochondrial D N A and the circular viral DNA's are very similar (Borst et al., 1 9 6 7 ~Dawid ; and Wolstenholme, 1967; Clayton and Vinograd, 1967), with the possible exception of the ratio Sl neu+ral/SII ,lr,,+ml, which is 1.4 for mitochondrial D N A and about 1.3 for most viral circular DNA's. In our opinion, it is not possible to decide from the data available whether o r not this difference is significant. TABLE V SEDIMENTATION COEFFICIENTS O F T H E DIFFERENT FORMSOF MITOCHONDRIAL DNA Source of D N A Chick liver*

Rat liverb

Amphibian eggsc

Human leukocytes"

I Neutral ( N a salt)

49

11 Neutral ( N a salt) 111 Neutral ( N a salt)

27

39 27

37 27

2t

-

39 27 24

Componenl

I Alkali (Cs salt) I Formaldehyde ( N a salt) I1 Alkali ( N a salt) 11 Formaldehyde ( N a salt) h G

d

~~

-

89

87

80

83

-

-

-

-

-

24

32.28

-

-

From Borst et al. ( 1 9 6 7 ~ ) . Srnit and Borst (unpublished results). From Dawid and Wolstenholme (1967). From Clayton and Vinograd (1967).

Using Studier's (1965) formula relating .rzo,dto the molecular weight of linear duplex DNA, a molecular weight of 10.9 x loGdaltons can be calculated for the sodium salt of the 24-S form of mitochondrial DNA. The average contour lengths of chick liver and Xenoptds 1uevi.r mitochondrial D N A are 5.35 and 5.40 p, respectively. Using a value of 1.96 x lo6 daltons per micron of sodium D N A (Thomas, 1966), these contour lengths correspond to molecular weights of 10.5 1 0 6 and 10.6 x 1 0 6 daltons. The molecular weights calculated from contour lengths and sedimentation analysis are therefore in excellent agreement.

x

C. THENUMBER OF SUPERHELICAL TURNS I N MITOCHONDRIAI. DNA

The number of superhelical turns in the closed circular duplex form of mitochondrial D N A has been studied both by electron microscopy and by a dye intercalation technique. Van Bruggen et al., 1968 (see also Borst et al., 1968) counted the number of crossovers in electron micrographs of the closed circular duplex form of chick liver mitochondrial DNA. For purified component I DNA, spread at 20°C. on 0.1 M ammonium acetate, they obtained a value of 33 -+ 7 (S.D.) crossovers per molecule; for component I released from mito-

MITOCHONDRIAL DNA

129

chondria Iysed by osmotic shock at 2o0C., the number of crossovers was 35 _t 6. This is an average of 6.4 crossovers per micron of DNA. For purified component I of the replicative form of @X174 analyzed on 0.1 iM ammonium acetate, they counted 13 2 2 crossovers, or 7.6 crossovers per micron of DNA. This suggests that the number of superhelical turns per unit length of mitochondrial D N A and 0 X D N A is approximately the same. These results were extended by analyzing the sedimentation behavior of chick liver mitochondrial D N A as a function of ethidium bromide concentration. As shown in Fig. 5 , the unwinding of the D N A helix induced by the intercalation of ethidium bromide first leads to a loss of supercoiling and a concomitant decrease in the sedimentation coefficient of component 1 to that of component 11. At higher ethidium bromide concentrations, the sedimentation coefficient of component I rises again, probably because increasing unwinding of the helix leads to insertion of left-handed superhelical turns. It is clear from Fig. 5 that the titration curves of chick liver mitochondrial D N A and polyoma D N A , which were determined under identical conditions, are not significantly different. Since the binding constant of ethidium bromide to D N A is affected neither by the base composition nor by the length of the D N A (Waring, 1965), this result shows that complete unwinding of the right-handed superhelical turns of mitochondrial D N A and polyoma D N A requires the intercalation of the same amount of ethidium bromide per nucleotide. Therefore, the number of superhelical turns per unit length of D N A must be the same. It is possible to calculate the number of superhelical turns per D N A molecule from the results presented in Fig. 5 by using certain reasonable assumptions (Crawford and Waring, 1967). For chick liver mitochondrial D N A at 20°C., a value of 40 turns per molecule was found, or 7.5 turns per micron of D N A (Ruttenberg et ul., 1968). This is in excellent agreement with the estimate of 6.4 turns per micron made by counting crossovers in electron micrographs. Using the dye intercalation technique, Smit and Ruttenberg (unpublished observations) have recently found that rat liver mitochondria1 D N A contains about the same number of superhelical turns per unit length as chick liver DNA. Two explanations have been advanced to account for the right-handed superhelical turns found in all closed circular duplex DNA’s synthesized in the intact cell. (1) When the D N A is synthesized, closure of the chain ends of the newly synthesized strand takes place before all of the winding of the two D N A strands into the Watson-Crick structure is completed. This explanation predicts that superhelical turns are present if? viuo, that the number of superhelical turns per molecule is constant, and that the number of turns per unit length of D N A is, therefore, higher in small circles than in large circles (Vinograd et ul., 1965).

130

P. BORST A N D A. M. KROON

Our finding that two DNA’s that differ by a factor of 3 in molecular weight have the same number of superhelical turns per unit length of D N A strongly suggests that this explanation for the origin of superhelical turns in closed circular duplex D N A is incorrect. ( 2 ) The average rotation per base pair of the D N A helix is lower in the intact cell than in the solvents used for the physical characterization of the extracted DNA. The intracellular condition giving rise to the “underwound” state of the D N A in the intact cell is not known. However, the recent demonstration that the pitch of the D N A helix is dependent both on temperature and on the ionic strength of the solvent (Wang et al., 1967; Wang, 1969, Bode and McHattie, 1968) suggests that ionic conditions may be responsible. This explanation implies that the superhelical turns found in vitro do not exist iyz vivo and that the actual number of twists found in uitro depends on the conditions of analysis chosen. If this is true, it is remarkable that the intracellular conditions under which D N A is synthesized are so similar in vertebrate mitochondria, mammalian nuclei (polyoma), and Escherirhia coli (replicative form of 0 x 1 7 4 ) that the pitch of the helix is the same in all three cases.

D. SIZE

AND

CIRCULARITY O F MITOCHONDRIAL DNA ANIMALTIssuEs

FROM

The size and circularity of mitochondrial D N A from a variety of animals have been studied by electron microscopy, and the results obtained are presented in Table VI. Circular D N A was found in all animal mitochondria analyzed so far, and no difference in contour length was detected for mitochondrial D N A from mouse liver, brain, kidney, and pancreas (Sinclair et al., 1967b). More remarkable, however, is the uniformity in size of mitochondrial D N A from such diverse branches of the evolutionary tree as mammals, sea urchins, and insects. Even the small differences in size given in Table VI may not be significant: Different versions of the Kleinschmidt protein monolayer technique are used in different laboratories; the D N A samples analyzed often contained large amounts of contaminating material; and results obtained for the same D N A in different laboratories (and even at different times in the same laboratory, cf. Borst el al., 1968) may yield different contour lengths, as shown in Table VI. T o establish whether or not real differences in contour length exist, it will be necessary to use highly purified D N A samples and demonstrate the presence of two size classes in mixed preparations of the two animals. Even if such experiments show that the contour lengths of mitochondrial D N A in human tumor cells, chick liver, carp liver, sea urchin eggs, and the flight muscle of house flies are not exactly the same, the fact that they are so similar requires an explanation. Three possibilities can be envisaged:

131

MITOCHONDRIAL DNA

CONTOUR

TABLE V I LENGTHOF CIRCULARD N A FROM ANIMALMITOCHONDRIA~

Species Chordata Mammalia Man

ox Sheep Rat

Techniqueb

K K K-F K K-F K

Guinea pig Aves Chick

Echinodermata Echinoidea Sea urchin ( L . pirtccs) Arthtopoda Insecta Fly (M.domestira)

4.81 5.3" 5.1

0s

K-F

5.6

K

5.35 5.55 5.1 5.26

K K-F K-F

References

Radloff et a/. (1967) Kroon et a!. (1966) Sinclair ef al. (1967b) Kroon et a/. (1966) Sinslair et al. (1967b) Van Bruggen et a/. ( 1968) Van Bruggen et a/. (1968) Kroon et al. (1966) Sinclair and Stevens (1966) M. M. K. Nass (1966) M. M. K. Nass (1966) Sinclair et a/. (1967b)

K-F K-F K

5.1

Borst et al. ( 1 9 6 7 ~ ) Van Bruggen el a/. (1968) Sinclair ei a/. (1967b) Sinslair et al. (1967b) Kroon el nf. (1966)

K-F K-F

5.56 5.40

Wolstenholme and Dawid (1967) Wolstenholme and Dawid (1967)

0s

K

5.4 5.4

Van Bruggen et al. (1968) Van Bruggen et al. (1968)

K

4.45

Piko ef al. (1967)

0s

5.2

Van Bruggen et

0s Duck Amphibia Frog ( R . p i p i e n s ) Toad ( X . laevir) Osteichthyes Carp

(p)

5.4 5.1 4.9 5.4 5.1c 4.96 4.74 5.24

0s Mouse

Average circumference

a/.

(1968)

From Van Bruggen et al., 1968. K, standard protein monolayer technique, purified D N A ; K-F, as K but with 0.5% formaldehyde present (Freifelder and Kleinschmidt, 1965) ; OS, mitochondria lysed by osmotic shock as described by Van Bruggen et al. (1968). c Values are based on less than 10 measurements. 0

b

132

P. RORST A N D A . M. K R O O N

( I ) Mitochondria1 D N A is very resistant to any form of mutagenesis leading to changes in D N A length. The constant length is therefore not the consequence of a rigid selection system but the result of a lack of change. This explanation is not very attractive because mitochondrial D N A is not at all immune to mutagenic events. Table I shows that the base composition of mitochondrial D N A is far from constant, while the results discussed in Section V,E show that mitochondria contain the equipment to form and eliminate multimers of mitochondrial DNA. ( 2 ) The constant size of the mitochondrial genome reflects the constancy of the basic mitochondrial structure and functions. According to this hypothesis, mitochondrial D N A specifies a set of proteins involved in oxidative phosphorylation. Since these proteins could be similar in all animal mitochondria, a similar stretch of D N A will be required to specify them. The difficulty with this hypothesis is that at least one component of the respiratory chain, cytochrome c, is specified by nuclear D N A (see Section Xl1,D). It is likely that the structural genes for some of the Krebs cycle enzymes (e.g., aconitate hydratase) required to channel reducing equivalents into the respiratory chain are also localized on nuclear DNA. It is clear, therefore, that there is no obvious reason in this hypothesis why animal mitochondrial D N A could not be larger than 5 p and take over some of the nuclear genes specifying mi tochondrial enzymes. ( 3 ) If we assume that mitochondria originate from bacterial symbionts, a possibility favored by most of the workers in the field (cf. Lehninger, 1964; M.M.K. Nass, I967 ; Granick and Gibor, 1967 ; Roodyn and Wilkie, 1968). it is necessary to postulate that in the course of evolution most bacterial genes have disappeared and that their task has been taken over by nuclear genes. This process of genomic reduction could have halted at the 5-y stage for several reasons. To be of evolutionary advantage mitochondrial D N A must meet two requirements: First, it should specify certain gene products that are essential for the biosynthesis of the mitochondrion, and the fact that the relevant genes are present in many separate copies in the cytoplasm must provide an evolutionary advantage. Second, the expression of the cytoplasmic and the nuclear genes must be precisely coordinated and a certain number of mitochondrial genes must therefore have a regulatory function. It is possible that the minimal amount of D N A that will still meet both requirements is 5 p. An alternative possibility is that mitochondrial D N A specifies a series of proteins which function as a unit and which cannot be coded for by less than 5 p DNA. Obvious candidates for such a unit are the mitochondria1 ribosomes and the proteins tightly bound to the mitochondrial inner membrane. If the synthesis of such a unit is regulated by one large operon, it is conceivable (although by no means necessary) that the chances of obtaining adequate and coordinated substitution for the separate genes specifying this unit by mutagenic events in the nucleus is virtually nil. A

MITOCHONDRIAL D N A

133

third possibility is that mitochondrial D N A codes for a number of gene products that are required in or on the inner side of the mitochondrial membrane and that cannot be transferred through the mitochondrial membranes. Again mitochondrial ribosomes or inner membrane proteins might be these gene products. It is clear that according to this last hypothesis the 5-y D N A of animal mitochondria represents a minimum, an evolutionary bottleneck on the way to a complete loss of mitochondrial D N A and transfer of its information content to the nucleus. Identification of the genetic function uf mitochondrial D N A will show which of these explanations for the constant size o f animal mitochondrid D N A is correct. From the considerations presented in this section, it is clear that there is no reason to expect that all mitochondrial D N A in nature is circular and 5 p long. The evolutionary advantage of circular D N A does not appear to be decisive, since many viruses survive with linear D N A and mitochondrial D N A of organisms not studied as yet may do likewise. Similar considerations hold for the size of mitochondrial DNA. Even if the general evolutionary trend is toward smaller mitochondrial DNA, occasional higher organisms at the dead ends of evolution might cultivate a larger mitochondrial DNA, while the mitochondria1 DNA from primitive organisms might have retained more of the genes of its supposed bacterial ancestor than the mitochondrial D N A from animals has. In summary, it is not possible to extrapolate from the data presented in Table VI until more is known about the function of mitochondrial DNA.

E. OLIGOMERS O F MITOCHONDRIAL DNA

Two types of oligomers of mitochondrial D N A were discovered by Vinograd and co-workers (Radloff et al., 1967; Hudson and Vinograd, 1967; Clayton and Vinograd, 1967) in DNA from malignant cells: circular oligomers, i.e., rings with a contour length of a multiple of 5 p, and catenated oligomers, i.e., circular D N A molecules consisting of independent, double-stranded circles that are topologically interlocked or catenated as links in a chain. Most of the oligomers found were dimers, but smaller amounts o f catenated higher oligomers up to a septanier were also observed. The fraction of the total mitochondrial D N A and the ratio of catenated to circular oligomers varied depending on the source of the DNA. Mitochondrial D N A from HeLa cells contained about 10% catenated dimers (on a weight basis) and no circular dimers. Mitochondrial D N A from leukocytes of three different cases of human leukemia contained the following fractions of oligomers on a weight basis: case I , 39% circular dimers, 5 % catenated dimers, and 4.5% higher oligomers; case 2, 7% circular dimers, 10% catenated dimers, and 6% higher oligomers; case 3, 3.5% circular dimers, 5 7 0

134

P . BORST A N D A.

M. KROON

catenated dimers, and 0.7% higher oligomers. These percentages are based on an analysis of the D N A in the closed circular duplex band collected from preparative CsCl gradients containing ethidium bromide. As pointed out by Vinograd and co-workers, these values are probably minimum values. The large fraction of oligomers obtained from the malignant cells made it possible to analyze the nature of the oligomers in detail. The circular dimers were found to have the same buoyant density as the monomers, while both their length and their sedimentation properties in neutral salt and alkali were consistent with a closed circular duplex structure with exactly twice the contour length of the 5-p monomer. It is assumed by Vinograd and co-workers that the circular dimer contains the base sequence of two identical monomers sequentially linked. Mild shear degradation of the circular dimers, followed by denaturation and renaturation at low concentrations, should therefore lead to formation of 5-11 circles as the major circular product. This experiment has not yet been reported. The nature of the catenated molecules was analyzed in two ways (Hudson and Vinograd, 1067; Clayton and Vinograd, 1967). Catenated dimers were enriched in a CsCl gradient containing a high concentration of ethidium bromide in a band exactly intermediate in density between the bands of closed circular duplex and open duplex DNA. This proves that these dimers consist of one closed circle (taking up less ethidium bromide) and one open circle (taking up more ethidium bromide) of about equal size and linked by bonds stable in 6 M CsCI. The equal contour length of the molecules linked in catenanes was confirmed by length measurements in electron micrographs, and the appearance of the attachment site was in agreement with the assumption that the molecules are linked by a topological bond (similar to the two strands of the monomers in strand-separating solvents) and not by chemical bonds. To exclude that chemical bonds are also present it will be necessary to demonstrate that the catenated molecules are completely dissociated by introducing a double-strand break in one of the circles. This experiment remains to be performed. Oligomers of mitochondrial D N A are also present in nonmalignant cells, although at much lower concentrations. Oligomers (3%) were found in the closed circular duplex D N A of normal human leukocyte mitochondria, while an unspecified fraction of sea urchin mitochondrial D N A consisted of catenated dimers and a small proportion of higher catenated oligomers (Clayton and Vinograd, 1967). W e have also observed a very faint intermediate band in several preparations of chick and rat liver mitochondrial D N A centrifuged to equilibrium in a CsCl gradient containing ethidium bromide (Borst and Van Bruggen, unpublished observations, 1967), Assuming that nicking of closed circular duplex D N A hits monomers and catenated dimers with the same frequency, we calculated that the fraction of mitochondrial D N A present as catenated dimers is about 1% in chick liver and even less in rat liver. Analysis

MITOCHON1)RIAL D N A

135

of the fraction corresponding to the intermediate band in preparations of chick liver D N A showed 'in appreciable fraction of dimers consisting of one open 5-p circle and one twisted molecule. In addition, the fractions corresponding to the closed band contained about 2% catenated dimers and less than 0.5% open dimers. At these very low proportions of multimers it is often difficult to distinguish circular and catenated dimers from accidentally overlapping molecules and, therefore, these percentages are only approximate. Oligomers of circular D N A have recently been observed for the replicative form of phage 0 x 1 7 4 (Van Bruggen, Jansz, and Pouwels, unpublished observations, 1967; Rush et nl., 1967; Rush and Warner, 1967) and for a bacterial plasmid (Roth and Helinski, 1967). They could be formed from monomers either during D N A replication or during recombination; the latter mechanism is favored by Vinograd and co-workers. As shown in Fig. 7, one symmetric

FIG. 7. The recombination model for formation of mitochondria1 multimers. T h e circular molecules first pair ( a ) , and are then "broken" either once ( b ) or twice ( c ) as shown. If broken once, reunion results in an open dimer ( d ) which can pair again ( e ) and recombine. Half of the products of the second recombination will be catenanes ( f ) , while the other half will be separate circles. If broken twice ( c ) , half of the recombinations will result in separate circles ( i ) , while half will be catenanes ( h ) (from Hudson and Vinograd, 1967).

crossing-over will lead to a double-length circle, while a second crossing-over may lead either to an interlocked dimer or to two unconnected monomers. Multiple crossing-over events may lead to multimers of various lengths and various degrees of catenation. This symmetric recombination model places the various types of mitochondrial D N A species in a sequence running from monomers to circular dimers, to catenated dimers, to higher oligomers. The population of mitochondrial D N A is viewed by Vinograd and co-workers as an equilibrium population and the various possible distributions of mitochondrial species as different positions in a multiple equilibrium. The factors influencing the position of this equilibrium are not known and no explanation is available for the large differences in the equilibrium positions in different tissue samples. If the first recombinational event is relatively improbable in relation to subsequent recombinational events between the original molecules linked as a circular dimer or a catenated dimer, one would expect that multimers are selfeliminating and that the number of multimers can be kept small. However, the finding of 39% circular ditners against 5% catenated dimers in one case of

136

P . BORST A N D A. M . KROON

leukemia does not agree with this hypothesis. Moreover, one would expect that further recombination events involving the two circles of a catenated dimer would lead to catenated dimers in which the monomers are wound around each other at the position of the interlock. This has not been observed. The possibility should therefore be left open that normal mitochondria have additional mechanisms at their disposal to eliminate multimers. It is not yet clear whether the high fraction of oligomers found in mitochondrial D N A from malignant cells is a reflection of rapid multiplication of mitochondria or of an imbalance in the recombination system peculiar to dedifferentiated cells. Since any simple symmetrical recombination mechanism leads to formation of circular dimers from circular monomers, the discovery by Wilkie that recombination of mitochondrial D N A does occur in yeast (Section X ) supports Vinograd's suggestion that the oligomers of mitochondrial D N A arise by recombination. However, if replication of circular D N A involves a double-strand break, as envisaged in several models for D N A replication (see Lark, 1966; Yoshikawa, L967), oligomers could also arise during D N A replication, and there is no direct experimental evidence at the moment arguing against this possibility. F. THEBEHAVIOR O F MITOCHONDRIAL DNA

IN

ALKALI

Borst et al. ( 1 9 6 7 ~ )have reported that mitochondrial D N A from chick and rat liver was rapidly degraded in alkali to a collection of heterogeneous fragments. Recent experiments by Smit in our laboratory have shown that this is a consequence of the isolation procedure. If the D N A is extracted and purified at 0-4"C., instead of room temperature, most of it is stable in alkali. Although the cause of the alkali lability is not known, it seems likely that radicals, present in low concentrations during the preparation of the DNA, may lead at room temperature to liberation of occasional bases without cleavage of phosphodiester bonds (cf. Bode, 1967; Rhaese and Freese, 1968). On exposure to alkali, the chain is cleaved at the sites where bases are missing (Tamm el ul., 1953). Even with mitochondrial D N A preparations, which do not show rapid fragmentation in alkali, a slow alkaline degradation takes place leading to a slow loss of the alkaline supercoil with a half-life measured in hours. This phenomenon, which has been observed with mitochondrial D N A from toad eggs (Dawid and Wolstenholme, 1967), sea urchin eggs (Vinograd, private communication, 1'967), and chick and rat liver (Smit, unpublished observations), has thwarted attempts to determine the equilibrium density of component I in alkaline CsCl. It seems likely that the slow degradation in alkali results from the same factors that underly the rapid alkaline degradation. W e expect that more rigorous purification of the D N A will bring the alkali stability to the same level as that of polyoma D N A (see Weil and Vinograd, 1963). Recently, Pouwels and Jansz and their co-workers (Pouwels et ul., 1966, 1968;

137

MITOCHONDRIAL DNA

Jansz et al., 1968) have shown that component I of replicative form D N A of 0 x 1 7 4 denatured by alkali does not normally renature after neutralization, even under conditions in which denatured component I1 will completely renature. However, after the introduction of one single-strand break in denatured component I, it immediately renatures to form native component 11. The authors suggest that the inability of native denatured component I to renature under these conditions is the result of two effects. (1) When all hydrogen bonds of component I are broken, the complementary strands shift out of register; at neutralization interstrand nonspecific hydrogen bonds immediately form, blocking the movement of the two strands relative to each other in the direction of the helix axis. ( 2 ) In the denatured DNA, the movement of the two strands away from each other is severely curtailed because the original Watson-Crick turns are still conserved in the denatured molecule, probably in part as right-handed turns and in part as left-handed superhelical turns. These two effects in combination prevent the nucleation necessary for successful renaturation. One singlestrand break is sufficient to remove the topological restraint, and renaturation follows immediately. This interpretation of the results obtained with OX-DNA implies that alkaline denaturation is pseudo-irreversible for all closed circular duplex DNA’s. In this light, the results of Dawid and Wolstenholme (1967) are unexpected. They denatured mitochondrial D N A of toad eggs with 0.1 N NaOH at 0°C. and found that after neutralization a portion of the D N A corresponding to the amount of component I present still had the equilibrium density of native D N A in CsCl. Smit (unpublished observations) has recently reinvestigated this point in our laboratory. After denaturation of rat liver mitochondria1 D N A with alkali, followed by neutralization, all alkali-stable component I was converted into a new component, sedimenting with an J . ) ~ , , of 70-75 S, the sedimentation coefficient expected for the neutral denatured form of component I on the basis of the results obtained by Pouwels et al. (1968) with g X - D N A . N o D N A sedimented at the position of native component I. It seems likely, therefore, that the “reversible denaturation” observed for toad egg mitochondrial D N A must have been the result of incomplete denaturation. In conclusion, it now appears that the behavior of mitochondrial D N A in alkali may not be significantly different from that of other circular DNA’s. G. COMPOSITION OF MITOCHONDRIAL DNA

FROM

ANIMALTISSUES

The size distribution in analytical band sedimentation experiments of DNA extracted from chick and rat liver mitochondria is shown in Fig. 2 . Three components are usually found. (1) The closed circular duplex form of mitochondrial DNA, sedimenting with a sedimentation coefficient of 39 S.

138

P. BORST A N D A . M. KROON

( 2 ) The open circular duplex form of mitochondrial D N A at 27 S.

( 3 ) Heterogeneous low-molecular-weight material sedimenting with sedimentation coefficients lower than 27 S. In addition, two minor components are present that do not show up in band sedimentation studies. ( 4 ) Oligomers of mitochondrial D N A represent only about 1-30/0 of the totd high-molecular weight D N A of the normal tissues studied so far, but in tumor cells up to 5 0 7 ~of all D N A may be present as oligomers (see Section VJ). ( 5 ) Since continuous synthesis of mitochondrial DNA is probably taking place in animal cells (see Section VIII), replicating molecules can be expected to be present in the D N A extracted. Nu convincing examples of branched molecules have been detected among the thousands of mitochondrial D N A molecules from various animal sources scrutinized in several laboratories. Possible reasons for this failure are discussed by Borst et ul. ( 1 9 6 7 ~ ) .In addition, intermediates in the formation of oligomers should also be present, but these have not been detected either.

It is likely that most of the heterogeneous material at the top of the gradient in the upper tracing of Fig. 2 represents remnants of nuclear DNA, or RNA degradation products not removed by the isolation procedure. The amount of this material varies from preparation to preparation, while in mitochondrial D N A from amphibian eggs, in which nuclear contamination is no problem, virtually no low-molecular-weight D N A was observed in density gradients (Dawid and Wolstenholme, 1967), and 99% of all molecules in electron micrographs were circular (Wolstenholme and Dawid, 1967). The relative proportions of components I ;and I1 may vary considerably in different preparations from chick liver (Borst et ul,, 1967c), rat liver (Smit and Borst, unpublished observations), frog and toad eggs (Dawid and Wolstenholme, 1967), sheep heart, and duck liver (Kroon et al., 1966). Several lines of evidence discussed by Van Bruggen et al. (1968) indicate that most of the component I1 found is derived from component I, either during purification of the DNA, or during the isolation of the mitochondria through an activation of the mitochondria1 endonuclease described by Curtis and Smellie ( 1966) and Curtis et al. (1966). W e conclude that in the mitochondria studied most extensively, i.e., mitochondria from chick liver, rat liver, and amphibian eggs, closed circular duplex D N A with a contour length of 5-5.5 p represents the major component present in the mitochondria in sitti. Minor components include oligomers of mitochondrial DNA, intermediates in replication, and intermediates in the formation of oligomers. Only the former have been detected as yet. Although these conclu-

MITOCHONDRIAL DNA

139

sions may hold for animal mitochondria in general, there are no a priori grounds to exclude that exceptions may appear. It should also be stressed that the presence of minor components other than the ones mentioned above cannot be rigorously excluded, even in the mitochondrial DNA’s extensively analyzed. The recovery of mitochondrial D N A during extraction snd purification is usually not better than 50%. Loss of a minor component, which is specifically eliminated by the isolation procedures employed and which does not show up in electron micrographs of mitochondrial preparations lysed by osmotic shock (see Van Bruggen et al., 1968) would, therefore, pass unnoticed. There is no positive evidence, however, that this is taking place.

VI. Size and Circularity of Mitochondria1 DNA from Plants and Unicellular Organisms N o conclusive evidence is available as to the circularity and size of mitochondrial D N A from any of the organisms represented in these categories. Suyama (1966) has reported that mitochondrial D N A from Tetrdhymelza sediments with a sedimentation coefficient of 40 S. Since the same sedimentation constant was obtained if precautions were taken to prevent shear degradation, Suyama concluded that this 40-S component is the undegraded l e t m h y m e m mitochondrial DNA. Although it is tempting to infer from Suyama’s results that mitochondrial D N A of Tetv~ihymenais also a 5-p closed circular duplex, attempts by Sinclair (1966) to obtain support for this conclusion were unsuccessful. In electron micrographs of mitochondrial D N A from Tetr~ihyvzenu,he observed only occasional circles with a contour length varying between 1 and 13 p, with a clustering of molecules at sizes 3.0, 4.3, 5.8, 8.4, and 13 p. These experiments are not conclusive, however, since only about one-third of the D N A in Sinclair’s mitochondrial D N A samples renatured under conditions in which mitochondrial D N A from animal tissues or yeast completely renatured. The possibility remains, therefore, that the circles observed by Sinclair were derived from contaminating nuclei, while true mitochondrial 5-p circles were absent because they had been degraded. It is also possible, of course, that mitochondrial DNA of Tetrclhynzenn consists of linear molecules which happen to sediment at 40 S. The mitochondrial D N A from E . gvucih was analyzed in detail by Ray and Hanawalt (1964, 1965). The D N A was prepared by preparative CsCl gradient centrifugation of whole-cell DNA, and from its sedimentation rate in sucrose gradients a molecular weight of 3 - 4 x 1 0 ° was calculated, assuming that the D N A was linear. The possibility that the D N A had been degraded during purification was excluded by conclusive control experiments. However, for the extraction of the D N A the Euglen‘r cells were first extracted twice with 937;

140

P. RORST A N D A . M. KROON

ethanol, prior to dodecyl sodium sulfate lysis. It cannot be ruled out that the ethanol treatment activates mitochondrial nucleases that selectively degrade mitochondrial D N A without degrading the nuclear and chloroplast DNA's. Occasional circles with a contour length of 3.5 p were observed by Sinclair (1966) in the whole-cell D N A of E. gracilis. This is twice as long as the maximal length calculated from the sedimentation coefficient found by Ray and Hanawalt for Euglerzu mitochondrial DNA. Moreover, there is no indication that the circles observed by Sinclair were derived from the mitochondria. Much effort has been expended in a number of laboratories to determine the size and structure of intact mitochondrial D N A from yeast. The major results obtained with isolated yeast mitochondrial D N A are summarized in Table VII. The results obtained by the top four groups are essentially in agreement. Only linear D N A was found in mitochondrial D N A preparations not contaminated by nuclear DNA. Sinclair et al. (1967a) concluded that these linear molecules could not have originated from breakage of circles during isolation and extraction and they suggested that linear molecules of 5-6 ~1 may represent the intact yeast mitochondrial DNA. The marked heterogeneity of the rods obtained, however, makes this interpretation very unattractive, and both Slonimski (in discussion of Swift et al., 1968b) and Borst and co-workers (Borst et al., 1968; Van Bruggen et ul., 1968) concluded that the linear molecules represented breakdown products of a much larger (and possibly circular) DNA. This is not unreasonable, because yeast mitochondria contain about 1 0 times as much D N A per milligram of mitochondrial protein as mammalian liver mitochondria (Schatz et al., 1964b). If we assume that the number of mitochondria per milligram of protein and the number of DNA molecules per mitochondrion are equal for purified liver and yeast mitochondria, yeast mitochondrial D N A could be up to 50 p long. The correctness of this assumption is supported by the following calculation. A diploid yeast cell contains about 5 x 10-14 gm. D N A (Ogur et al., 1952) ; up to 20% of this may be mitochondrial (Moustacchi and Williamson, 1966), and the number of mitochondria per yeast cell is about 5 0 (Avers et al., 1965). Therefore, one yeast mitochondrion contains about 2 x IO-le gm. or 120 x lo8 daltons DNA. To further study the length of undegraded yeast DNA, we subjected yeast mitochondria to the osmotic shock procedure of M. M. K. Nass (1966), in which shear degradation of the D N A is minimized. Most of the D N A was liberated in flowerlike structures containing at least 50 p DNA. However, the D N A in these flowers was not continuous and the longest D N A fragment observed was 18 IL (Borst et al., 1968; Van Bruggen et al., 1968). Since breaks might have been introduced by the mitochondrial endonuclease during purification of the mitochondria, Hollenberg in our laboratory has recently studied the possible presence of mitochondrial closed circular duplex D N A in protoplast lysates with-

SIZE A N D STRUCTURE OF

TABLE VII MITOCHONDRIAL DNA

ISOLATED FROM YEAST ~

Refcrentrs

',

Average

.( -0.W

Trwari ei al. (1966)

3-3

Sinclair

-

et

al. (1967a)

Borst el a/. (1968) Slonimski (see Swift t t at.. 196Sb) Avers ( 1967) a

About 25

--

-

< 5 About 5 About 8 -

Maximal

5.5 9a

> 20 11

Calculated mol. wt. (daltons)

x 1 x >2 x 2

107? in7

107

> 4 x 107

Variable

Longer niolcrules ( u p to 18 p ) observed in mitochondria1 DNA rrlrascd by osmotic shock (see text).

~

_

Circular

DNA Hcttrogcnrity I

+

++ ++ +++

present -

-

+

_

_

K

3

0

i i

8 5

;

142

P. BORST AND A. M . KROON

out prior purification of the mitochondria. Protoplasts were lysed with dodecyI sarkosinate and after addition of solid CsCl and an excess of ethidium bromide, the lysate D N A was centrifuged to equilibrium. Under these conditions, all mitochondria1 D N A banded at the position expected for linear D N A with a buoyant density of 1.684 gm./cni.s in the absence of ethidium bromide. N o closed circular duplex D N A band was detectable. Although these experiments seem to indicate that yeast mitochondrial D N A is not circular, this conclusion is not yet justified for two reasons. ( 1 ) It remains to be demonstrated that the procedure used for isolating D N A from protoplast lysates excludes strand breakage. ( 2 ) The generation time of rapidly growing yeast is 1-2 hours. If yeast mitochondria1 D N A consists of 50-p circles which replicate at a rate of 1 p per minute, similar to nuclear D N A of HeLa cells (Cairns, 1966), Viciu fuba (Taylor, 1968) or Chinese hamster cells (Huberman and Riggs, 1968), most of the mitochondrial D N A from a log phase culture could be replicating DNA. Since replication of a circular D N A requires at least one single-strand break to introduce a swivel (Cairns, 1963), all replicating D N A will be found in CsCIethidium bromide gradients in the band containing open circles and linear DNA. W e are presently studying the mitochondrial D N A of stationary yeast cultures. Since the results obtained in four different laboratories with yeast are so consistent, the report by Avers (1967) that yeast mitochondrial D N A consists of circular D N A molecules varying in contour length between 0.5 and 10 p came as a surprise. In purified yeast mitochondrial D N A the majority of the molecules measured were circular and nearly all circles were smaller than 3 p. As it is impossible by the procedure employed to convert circles smaler than 3 p into linear molecules longer than 5 p, the results of Avers are incompatible with the other results presented in Table VII. W e may therefore point out that none of the circular molecules selected by Avers (1967) to illustrate her paper would have been classified as circular in this laboratory. Even if the circularity of these molecules could be established by more convincing electron micrographs, however, the possibility remains that they are identical with the heterogeneous circular D N A previously observed as a minor component in purified yeast mitochondrial D N A by Sinclair et al. (1967a). Sinclair et ul. (1967a) have shown that these circles have the density in CsCl of yeast nuclear DNA, and that they are presumably derived from nuclei contaminating the mitochondrial preparation. In view of these considerations, it is clear that the question of size and structure of yeast mitochondrial D N A remains open. Heterogeneous linear D N A up to 7 p long was observed by Luck and Reich ( 1964) in electron micrographs of mitochondrial DNA. The interpretation of these results meets with the same difficulties as the other c s e s discussed in this section. Only a limited number of experiments have been done to determine the size

MITOCHONDRIAL D N A

143

and structure of mitochondrial D N A from higher plants. Heterogeneous linear D N A up to 7 p in length was observed by Swift et al. (1768b) in electron micrographs of D N A from bean hypocotyl (Phuseolus udgaris) mitochondria. More recently, Van Bruggen, Borst, and Talen (unpublished results) analyzed the D N A from mitochondria of the spadix of the Voodoo lily (Saaromutum vemJnm), lysed by osmotic shock (cf. Van Bruggen et ul., 1768). Only linear DNA filaments were released, varying in length between < 1 LL and 35 p. The bulk of the molecules were larger than 10 p and the frequency distribution did not show preferred lengths. Taken together, the results discussed in this section suggest that the DNA from plmts and lower organisms may be longer than 5 p and linear.

VII. The Amount of Mitochondria1 DNA per Mitochondrion and per Cell Considerable effort has been spent in several laboratories to determine the amount of mitochondrial D N A per mitochondrion in mitochondria of various sources. Two methods have been used. (1 ) Determination of acid-insoluble deoxyribose in highly purified mitochondrial fractions, usually with the diphenylamine reaction (S. Nass et al., 1965). The assumptions made in this method are that all material reacting in the diphenylamine test is mitochondrial D N A and that all mitochondrial D N A present is measured. The correctness of both assumptions remains to be shown. N o satisfactory test has been devised as yet to demonstrate the absence of nuclear fragments in mitochondrial preparations. In this laboratory the method used most, light microscopy of Feulgen-stained smears, did not prove to be sensitive enough (Kroon, unpublished results). Furthermore, mitochondrial lipids interfere with the diphenylamine reaction and even after repeated extraction of the acidinsoluble material with lipid solvents, the spectrum of the mitochondria1 material reacting with diphenylamine is usually not identical with the spectrum of the product of the diphenylamine reaction with pure DNA. Correction for these greenish impurities is therefore necessary, adding another factor of uncertainty to the determination of mitochondrial D N A by this method. ( 2 ) To distinguish the mitochondrial D N A present in mitochondrial fractions from contaminating nuclear, bacterial, or chlvroplast DNA, the D N A was extracted and characterized by its equilibrium density in CsCl (Suyama and Bonner, 1966; Borst et al., 1967b; Clayton and Vinograd, 1967), renaturation behavior (Kroon et a/., 1966), or sedimentation velocity characteristics (Borst et al., 1767b). This can be done only after the D N A has been purified, and to monitor loss of D N A during purification it is necessary to add a known amount of marker DNA to the mitochondrial lysate, assuming that loss of mitochondrial

144

P. BORST AND A. M. KROON

D N A and marker D N A are the same during purification. Since mitochondrial DNA has unique physicochemical properties and could be attached to the mitochondrial membranes, it is not certain whether the losses of marker D N A in these experiments do parallel losses of mitochondrial DNA. The quantity of mitochondrial D N A per gram of mitochondrial protein, determined with these methods for mitochondria from various sources in different laboratories, is presented in Table VIII. In general, most authors find values TABLE VlII AMOUNT OF DNA FOUND IN MITOCHONDRIAL PREPARATIONS FROM VARIOUS SOURCES

Source of mitochondria Rat liver Chick liver Beef heart Mouse L cells Mung bean Tetrahymena pyriformis Sarrharomyrrr rrrevisiae

DNA per gin. mitochondrial protein (mg.) Method ( 1) a

Method ( 2 ) a

0.25-0.65

-

0.5 0.2‘1 1.1

0.7, 0.3

-

-

0.8

Mitochondria1 DNA per mitochondrion (gm.)

References”

x x 1x 9 x

10-17

3

10-17

10-16 10-16

6 7 8 9 10

0.5-2 7

-

1.7

x

1-4

+ -

0.7-1.0

0.7

6

x

10-16 10-17

10-17

1-4 5

See text for explanation. K e y fo refrrenies: (1) Schneider and Kuff (1965): ( 2 ) S. Nass rt al. (196513); ( 3 ) Schatz el al. (1964a); ( 4 ) P. Parsons and Simpson (1967); ( 5 ) Borst rf al. (1967b); ( 6 ) M. M K. Nass (1966); ( 7 ) Suyama and Bonner (1966); (8) Suyama and Preer (1965); ( 9 ) Schatz rf a/. (1964b): (10) Tewari ef al. (1966). 4 f~

around 0.5 mg. DNA/gm. mitochondrial protein for animal mitochondria. The value for yeast mitochondria is controversial, but in our opinion the higher values given in Table VIII are probably the correct ones (see Borst in discussion after Swift et d.,1968a). In the last column, the values are expressed as amount of mitochondrial D N A per mitochondrion. The lowest value for animal mitochondria is equivalent to two 5-p circles per mitochondrion, the highest values to more than 10. Evidence that several D N A molecules may be present in one mitochondrion also comes from cytochemical studies. Up to six unconnected DNA-containing areas were detected on serial sections of mouse L-cell mitochondria by M. M. K . N m (1968) and up to 10 unconnected areas were observed by Merker ef al. (1967) in the giant mitochondria of rat uterus. Some evidence exists suggesting that changes in the growth rate or the physiological state of cells may influence the amount of D N A per mitochondrion. It has been consistently observed in cytochemical studies that mitochondrial D N A is most easily identified in cells of rapidly growing tissues (M. M. K. Nass et

MITOCHONDRIAL DNA

145

al., 1965; Swift et al., 1968a) and according to Neubert (in discussion after Swift et al., 1968a) more DNA is present per milligram of protein in mitochondria from embryonic rat tissues than in mitochondria from weanling or adult rats. Furthermore, Work (in discussion after Swift et al., 1968a) has reported that injection of triiodothyronine into thyroidectomized rats leads to a >-fold increase in the amount of mitochondrial DNA. These results suggest that the D N A content of mitochondria may increase when the rate of mitochondrial duplication increases. Because the amount of DNA per mitochondrion is rather similar for D N A from widely different organisms (see Table VIII), it is obvious that the relative proportion of total cellular DNA found in the mitochondria only depends on the number of mitochondria per cell and the amount of D N A per nucleus, unless other DNA-containing organelles are present. This proportion varies from over 50% in the gigantic amphibian eggs (Dawid, 1966), through about 3040% in haploid Sacrharomyces cerevlsiae (Hollenberg, unpublished observations) with its small amount of nuclear DNA, through about 15% in diploid Sarcharomyce.r (see Section I X ) , through about 1% in chick liver (Borst et al., 1967b), to 0.2% in mouse L cells (M. M. K. Nass, 1966). VIII. Replication of Mitochondria1 DNA O F MITOCHONDRIAL D N A SYNTHESIS IN RELATION A. TIMING TO T H E CELL REPLICATION CYCLE

The relative timing of mitochondrial and nuclear DNA synthesis in the same cell has been studied by radioautography of pulse-labeled cells or by following the rate of incorporation of precursors into D N A of nuclear and mitochondrial fractions isolated from synchronized cells. The results obtained with Tetvuhymenu (J. A. Parsons, 1965) and P. polycephalzrm (Guttes and Guttes, 1964; Evans, 1966) were essentially identical. While the synthesis of nuclear D N A was limited to part of the cellular replication cycle (S phase), synthesis of mitochondria D N A proceeded at an approximately constant rate throughout the replication cycle. In Chang cells synchronized in tissue culture, maximal incorporation of thymidine into mitochondrial D N A occurred between the S phase and cytokinesis (Koch and Stokstad, 1967). Obviously the temporal control of mitochondrial and nuclear D N A synthesis is different. Recently, Mounolou et a/. (1968) and Swift el a/. (1968a) reported that exposure of anaerobic yeast to oxygen leads to an immediate and intense burst of mitochondrial D N A synthesis, which levels off after 10 minutes. During the period of increased D N A synthesis mitochondrial D N A is very prominent in yeast cell sections (Swift et al., 1968a). This may represent a promising test

146

P. BORST AND A. M. KROON

system to study control of mitochondrial D N A synthesis. The factors involved in this control are unknown.

B. TURNOVER O F MITOCHONDRIAL DNA The turnover of mitochondrial D N A in animal tissues in vivo has been studied in detail by Neubert and co-workers (Bass and Neubert, 1966; Neubert et al., 1968a). The animals were given a pulse of radioactive thymidine, and the specific radioactivity of the D N A isolated from highly purified mitochondrial and nuclear fractions was followed as a function of time. Thymidine incorporation into mitochondrial and nuclear D N A of rat liver was linear for about 6090 minutes following injection and then leveled off. In the week following the pulse, the specific activity of nuclear D N A remained nearly constant, indicating the virtual absence of D N A synthesis. The specific activity of mitochondrial DNA, on the other hand, steadily decreased following an ideal exponential dilution curve with a half-life of 9 days, as shown in Fig. 8. The half-lives of '000:

.....--._. *-. *......# l

400200-

w

\w\x Mitochondria1 DNA \x

a 100: z D

tl

d

2

9 days

40E

,"

20-

,o ......................I,-* 47

,

,

,

f l > 100 days

Nuclear DNA I

1

1 2 I2

.

1

16

'

$0

I

24

'

w

:

2'8

I

32

'

46

FIG. 8. Decline in specific activity of nuclear and mitochondrial DNA in rat liver after a single injection of thymidine-H3. A t day zero, adult rats (body weight 300 gm.) were injected with 1 mCi. thymidine-HS (specific activity, 6 mCi./pmole) per kilogram, followed by repeated injections of 40 mg. cold thymidine per kilogram. Reproduced in modified form froin Neubert et al. ( 1968a).

mitochondrial DNA's in adult rat heart, kidney, and brain were found by Gross et ul. (1968) to be 5.5, 7.9, and 39 days, respectively. Obviously mitochondrial D N A is turning over in resting cells. More recent work by Neubert and co-workers (1968a) has shown that the

MITOCHONDRIAL DNA

147

half-life of mitochondrial D N A depends on the rate of cell division: the higher the rate of cell division, the lower the half-life of mitochondrial DNA. In a rapidy growing Morris hepatoma (half-life nudear DNA = 5 days), the decrease in specific activity of mitochondrial and nuclear D N A was even identical, indicating that this decrease was completely attributable to net synthesis of D N A without any turnover. Neubert et al. (1968a) have suggested that “the increased need for the formation of certain mitochondrial Components during rapid growth is first of all satisfied by an increased stability of the components rather than an increased rate of synthesis.” At higher growth rates the rate of synthesis, of course, also increases. An obvious consequence of these results is that the relative rates of mitochondrial and nuclear DNA synthesis vary, depending on the mitotic index of the tissue studied. Neubert et ul. (1968a) have shown that, following a pulse of thymidine-H3 the specific activity of mitochondrial D N A is 50 times higher than the specific activity of nuclear D N A in adult rat liver. In 200-gm. rats this ration was 22, in 130-gm. rats 3.5, in 60-gm. rats 1.4,and in 13-day-old rat embryos it was 0.7. The preferential incorporation of D N A precursors into mitochondrial D N A in adult rat liver has also been observed by others (Schneider and Kuff, 1965; S. Nass, 1967; Gross et al., 1968). N o attempt has been made so far to determine which part of the turnover of mitochondrial D N A is attributable to replication and which part is attributable to D N A repair. In the subsequent part of this section we will assume that the contribution of D N A repair is negligible. On the basis of the results discussed above, it can be predicted that in regenerating liver the ratio of the specific activities of mitochondrial and nuclear D N A following a thymidine-H3 or P32 pulse will drop precipitously. This was actually observed in two laboratories. While the rate of thymidine-H3 (Chang and Looney, 1966) or P:,? (S. Nass, 1967) incorporation into mitochondrial D N A increased only 2- to 3-fold in regenerating liver, incorporation into nuclear D N A went up 20- to 80-fold. According to s. Nass (1967), incorporation into mitochondrial D N A doubled within 12 hours after hepatectomy, whereas incorporation into nuclear D N A did not increase “until sometime between 1 2 :ind 24 h.” The possibility that this represented an effect on precursor pools was made unlikely by the observation of Nass that net increase in mitochondrial D N A in regenerating liver preceded a net increase in nuclear DNA. It is interesting that the half-life of mitochondrial D N A in adult rat liver is in the same order of magnitude as the half-life of other major mitochondria1 constituents. From a limited number of experimental data, Fletcher and Sanadi concluded in 1961 that soluble proteins, insoluble proteins, cytochrome c, and bulk lipid of rat liver mitochondria turn over with the same half-life of 10-11 days. Although a half-life of 9 days for mitochondrial insoluble protein was also ob-

148

P. BORST A N D A. M. KROON

tained in a more detailed investigation by Bailey et a/. (1967), the decline in the specific activities of mitochondrial soluble protein and phospholipid did not follow a simple exponential curve. The logarithmic plots for soluble protein gave a curve with a continuously decreasing slope, indicating the presence of at least two components with different half-lives (Bailey et al., 1967). Turnover studies of mitochondrial phospholipids, labeled with Pa?,revealed the presence of two components with half-lives of 1.6 and 10 days, respectively (Bailey et al., 1967; Gross et nl., 1968). Moreover, when the phospholipids were labeled with C14 from acetate-C14, a half-life of 2 days was found (Bailey et al., 1967), suggesting that the turnover of the acyl moiety of the phospholipids is higher than that of the rest of the molecule. On the other hand, the slow turnover of mitochondria1 D N A in adult rat brain mitochondria with a half-life of 39 days (Gross et ul., 1968) is mirrored by the slow turnover of bulk phospholipids of brain mitochondria with a halflife of 32 days (Cuzner et al., 1966). Although these results do not agree with the proposal that complete mitochondria turn over as a unit, it would be surprising if the similarity in the half-lives of mitochondrial DNA, mitochondrial membrane-bound protein, and mitochondrial bulk phospholipid were fortuitous. It is likely, therefore, that mitochondria aye degraded as a unit but that some mitochondrial components turn over more rapidly, either because they exchange (an example might be the acyl moiety of mitochondrial phospholipids) or because they are degraded within intact mitochondria (an example might be induced mitochondrial 6aminolevulinate synthetase, which decays with a half-life of 70 minutes (Marver et al., 1966)). Degradation of mitochondria as a unit could occur as a consequence of the engulfment of mitochondria by lysosomes, since lysosomes have been observed to contain mitochondria in various stages of degradation (cf. De Duve and Baudhuin, 1966). If this is correct, the turnover rate of mitochondrial D N A might be an index of the metabolic activity of lysosomes. Extension of Neubert’s results for mitochondrial D N A to other mitochondrial components will show whether or not this concept is correct. The conclusion that mitochondrial D N A of adult rat liver is metabolically unstable has been criticized by S. Nass (1967) for three reasons: ( 1 ) The possibility exists that thymidine is incorporated into RNA or adsorbed to glycogen. Both RNA and the thymidine-glycogen complex (see Counts and Flamm, 1966) may be present as acid-insoluble contaminants of DNA, and turnover of either of these components could have been responsible for the decrease of the specific activity in Neubert’s mitochondrial “DNA” fractions. ( 2 ) Although Nass also finds that P32 is incorporated at a much higher rate

MITOCHONDRIAL D N A

149

into the DNA of mitochondrial than into that of nuclear fractions of rat liver, he concludes from the decay of D N A specific activity with time that mitochondrial DNA is stable and that its turnover is negligible. ( 3 ) Nass claims that similar experiments by Schneider and Kuff (1965) using thymidine-H" as D N A precursor also show (although the authors do not say so) that mitochondrial DNA of adult rat liver is stable. None of these arguments carries much weight in our opinion. First, several control experiments reported by Neubert et al. ( 1968a) and Gross et al. (1968) indicate that the acid-insoluble thymidine was in fact present in DNA. Second, the experimental results reported by S. Nass (1967) do not support his conclusion that mitochondrial D N A is stable. Figure 5 of his paper clearly shows that in the period between 4 and 14 days after the injection of P32, the specific activity of mitochondrial D N A of normal rat liver decreases, while the specific activity of nuclear D N A increases. These results are in good agreement with those of Neubert et al. (1968a) if the differences in turnover of the p32 and thymidine pools are taken into account. Thymidine turnover is very high, and within 3 hours after injection incorporation into mitochondrial D N A has completely stopped, while incorporation of P?' goes on for days. In addition, it is noteworthy that S. Nass (1967) does not provide any explanation for the fact that he too observes a much higher rate of P32 incorporation into the mitochondrial D N A than into the nuclear D N A of rat liver. If mitochondrial D N A were indeed as stable as nuclear DNA, the liver of old rats would be crammed with mitochondrial DNA. This is not the case. Last, the incorporation studies of Schneider and Kuff (1965) cannot be interpreted as support for the metabolic stability of rat liver mitochondrial D N A because in this work the specific activity of mitochondrial D N A was followed for only 18 hours after thymidine injection. One cannot expect to detect a half-life of 240 hours in an 18-hour experiment. W e conclude that the concept that mitochondrial D N A turns over in nondividing cells is well established, at least for the few cases studied. W e have pointed out earlier (Borst et al., 1967a) that this metabolic instability does not interfere with genetic continuity unless all mitochondria in a cell are degraded at the same moment. The only case known in which this might occur is in the anaerobic yeast cell. The way in which genetic continuity is conserved under these conditions will be discussed in Section IX.

C. THEMECHANISM OF MITOCHONDRIAL D N A SYNTHESIS I N INTACTCELLS The replication mechanism of mitochondrial D N A in the intact cell was studied by Reich and Luck (1966) with N. cradssa, using the density-labeling technique first used by Meselson and Stahl (1958) with E . coli. Nezrrospora crassa was grown with N15 as the only nitrogen source, and the buoyant density

150

P. BORST AND A. M . KROON

of D N A extracted from purified nuclear and mitochondrial fractions was analyzed in analytical CsCl gradients at different times after shifting the culture to an N14 medium. The behavior of nuclear D N A was that expected for uncomplicated semiconservative replication. With mitochondrial DNA, most of the DNA synthesized in the first doubling cycle consisted of DNA-N15 undiluted with DNA-NI4. Even after three doubling cycles the N15 content of the D N A was 40%, but in this case most of the DNA-N15 was diluted with NI4, since most of the D N A banded after denaturation at a position intermediate between pure DNA-N1* and DNA-NI5. Reich and Luck (1966) conclude from these results : “The nitrogenous precursors for mitochondrial D N A synthesis are drawn from a pool which is effectively large in relation to the amount of mitochondrial DNA, turns over slowly relative to the rate of mitochondrial D N A synthesis, and resists dilution by exogenous nitrogen sources. The contrary is true for nuclear DNA. Therefore, the replication of the two D N A species is at least metabolically independent, and perhaps topographically isolated, and a precursor-product relationship between the two is excluded.” Reich and Luck further conclude that their results are consistent with a semiconservative replication mechanism for mitochondrial D N A and that the pre-existing polynucleotide chains of mitochondrial D N A are conserved during replication.1 In their paper on renaturation of mitochondrial DNA, Corneo et ul. (1966) briefly mention unpublished results of Grossman and Marmur with yeast that are qualitatively similar to those obtained with Nezrrosporu. The nature of the N15 store on which the mitochondria draw for D N A replication is not known. Reich and Luck conclude from the flow of “4 into mitochondrial D N A during three replication cycles that the precursor pool must be many-fold greater than the amount of mitochondrial D N A present. They continue: “Since the existence of such a large pool of soluble deoxyribonucleotides would be surprising, it may be that in mitochondria, as in some other systems, the turn-over of RNA provides the immediate precursors for D N A synthesis.” This suggestion of Luck and Reich requires four assumptions: (1) The mitochondrial and extramitochondrial deoxyribonucleotide and ribonucleotide pools do not equilibrate. ( 2 ) The mitochondrial deoxyribonucleotide pool is small in relation to the amount of nucleotides present in mitochondrial DNA. ( 3 ) The enzyme system for converting ribonucleotides into deoxyribonucleotides, presumably at the nucleoside di- or triphosphate level (see Larsson and Reichard, 1 In our opinion, the high rate of incorporation of N16 deoxyribonucleotides into mitochondrial D N A after the shift to a “ 4 medium, makes it impossible to conclude from the results of Reich and Luck that the physical continuity of mitochondrial D N A is conserved during replication since similar results would have been obtained if the mitochondrial D N A were continuously depolymerized and resynthesized.

MITOCHONDRIAL D N A

151

1967), is present in the mitochondrial matrix space. (4) The turnover of mitochondrial RNA is high in relation to net RNA synthesis. Some objections can be raised to assumptions ( I ) , ( 3 ) , and ( 4 ) . The inner membrane of intact, isolated mitochondria has a low permeability for nucleotides with the sole exception of ADP and ATP, which are rapidly transported in and out by an adenine nucleotide translocase or permease (Klingenberg and Pfaff, 1966; Kemp and Groot, 1967; Ohnishi et al., 1967; Greenawalt et al., 1967). Although this permeability barrier has been observed in experiments on RNA synthesis by isolated, intact, mammalian mitochondria with added UTP, GTP, and CTP as substrates (Neubert et al., 1968b; Saccone et al., 1968), no evidence for such a barrier was found in iiz vitro studies of D N A synthesis (see Section VII1,D). More important is that no permeability barrier is apparent either in the experiments of Neubert et al. (1968a) on the incorporation of thymidine into mitochondrial D N A in vivo, since thymidine-HS appeared simultaneously in mitochondrial and nuclear D N A within 30 minutes after its intra'venous injection, while incorporation stopped completely 180 minutes after injection. This suggests that in liver the mitochondrial deoxyribonucleotide pool is small, and that the very low permeability of the mitochondrial inner membrane is sufficient to allow the influx of deoxyribonucleotides required for DNA synthesis. Rat liver mitochondria contain about 0.5 pg. mitochondrial D N A per milligram mitochondrial protein or 1.5 nmole D N A per milligram protein (see Section VII). If a rate of D N A synthesis of 1 p per minute is assumed, the influx required is maximally about 0.25 x 0.20 x 1.5 = 75 pmole per minute per milligram protein of each individual deoxyribonucleotide. This is four orders of magnitude smaller than the rate at which ribo-ADP and ribo-ATP are exchanged, and an influx of deoxyribonucleotides at this rate is not excluded by any in vjtro experiments. On the contrary, N A D + for which the permeability of the mitochondrial membrane is also very low (see review by Borst, 1963), was calculated to enter rat liver mitochondria in vivo at a rate of 43 pmole per minute per milligram mitochondrial protein (Purvis and Lowenstein, 1961). Since this rate was a minimum estimate it seems reasonable to assume that the low permeability of the mitochondrial inner membrane for nucleotides is adequate for supplying the mitochondrial complement of NAD, NADP, ribonucleotides, and deoxyribonucleotides in rat liver. Whether the permeability barrier present is detected in in vjtro incorporation studies may depend on the intactness of the mitochondria and the size of the intramitochondrial pool of the nucleotide studied. Nothing is known about the size of the deoxyribonucleotide pools in mitochondria. It seems reasonable, however, to assume that the concentrations of all deoxyribonucleotides are lower than those of the ribouridine and ribocytidine

152

P. BORST AND A. M. KROON

nucleotides, which are 0.1 and 0.2 nmole per milligram mitochondrial protein in rat liver (Heldt and Klingenberg, 1965) and 0.1 and 0.6 nmoles per milligram protein in yeast (Ohnishi et al., 1967), assuming that no nucleotides are lost during purification of the mitochondria. Since rat liver mitochondria contain about 1.5 nmole D N A nucleotide per milligram protein and yeast mitochondria about 4 to 10 times more, the deoxyribonucleotide pool may indeed be far too small to explain a 3-fold increase in the amount of N15 in mitochondrial D N A during growth in NI4-medium. The suggestion that ribonucleotides are converted into deoxyribonucleotides in Neziruspora mitochondria is not satisfactory, however, for several reasons. First, adenine contributes 38% of the nitrogen of Ne~~rosporu mitochondrial DNA, which has a mole % GC of 43% calculated from its buoyant density (Table 11). Since the adenine nucleotide translocase, present in NezlroJpora mitochondria (Greenawalt et al., 1967), completely equilibrates intra- and extramitochondrial adenine nucleotides, the maximal contribution of NI5 from ribonucleotides is 60%. Second, it is likely that most of the mitochondrial R N A in Nez4roJpor.a is ribosomal RNA and tRNA. It would be surprising if these RNA's turn over rapidly. Last, Wintersberger (1966) has found that unlabeled GTP, CTP, and UTP are completely unable to support the incorporation of dATP by isolated yeast mitochondria, which show an absolute requirement for the presence of all four deoxynucleoside triphosphates for D N A synthesis. This suggests that under these conditions no significant conversions of ribonucleotides into deoxyribonucleotides occurred. Although none of these considerations is conclusive, we think that the continuing incorporation of N15 nucleotides into Neuro.rporu mitochondrial D N A following a shift to N14 is not yet satisfactorily explained. Two alternative explanations, that are equally unsatisfactory, may be mentioned. If newly synthesized mitochondria were preferentially lost during purification of the mitochondria, a lag in the appearance of DNA-"+ would occur. If the mitochondrial DNA were contaminated by nuclear DNA, this would lead to an overestimation of mitochondrial DNA-NI5 since this bands only 2 mg./cm.3 above nuclear DNA-NIJ. Neither of these alternatives explains the presence of a large fraction of D N A strands containing both N14 and N15.

D. INCORPORATION O F DEOXYRIBONUCLEOTIDES INTO T H E D N A O F ISOLATEDMITOCHONDRIA Incorporation of deoxyribonucleotides into D N A has been observed with mitochondrial preparations isolated from rat liver (Schmieder and Neubert, 1966; P. Parsons and Simpson, 1967, L968), chick liver (Ter Schegget and Borst, 1968), yeast (Wintersberger, 1966, 1968), and P. polycephalum (Brewer et ul., 1967). The maximal incorporation rates observed for one labeled nucleotide are I. pmole nucleotide per milligram protein per hour (37°C.) for rat liver

MITOCHONDRIAL D N A

I53

mitochondria, 2 pmole per milligram protein per hour for chick liver mitochondria (37OC.), nearly 400 pmole per milligram protein per hour for yeast mitochondria (37OC.), and 170 pmole per milligram protein per hour for mitochondria from P. polyrephalnm (25OC.). Although the incorporation with liver mitochondria is low in comparison with bacterial systems, it is about 100 times as high as incorporation into isolated rat liver nuclei if incorporation is expressed on a D N A basis, as Schmieder and Neubert (1966) have pointed out. In the experiments of P. Parsons and Simpson (1967), maximal net synthesis of D N A by rat liver mitochondria, which incorporate dTTP linearly for a 2-hour incubation period, corresponded to 1'/F of the mitochondrial D N A present (based on a mitochondrial D N A content of 0.25 pg. per milligram protein). Although the initial rate of dATP incorporation in yeast mitochondria is 400 times higher than incorporation by rat liver mitochondria, net synthesis is only three times higher than in rat liver mitochondria, because yeast mitochondria contain 4 pg. D N A per milligram protein and dATP incorporation stops after 1 5 minutes incubation (Wintersberger, 1966, 1968). The incorporation studied with liver mitochondria has the characteristics expected for D N A synthesis by a D N A polymerase taking place within the mitochondrial inner membrane (Schmieder and Neubert, 1966; P. Parsons and Simpson, 1967, 1968; Ter Schegget and Borst, 1968): It is insensitive to deoxyribonuclease; it is not affected by added D N A ; it is inhibited by uncoupling agents and inhibitors of mitochondria1 electron transport; it is only in part dependent on the presence of all four deoxynucleoside triphosphates (probably because a small endogenous pool is present); it is inhibited by inhibitors of D N A synthesis like ,nogalamycin, cinerubin A, phleomycin, or high concentrations of actinomycin' D; and the same incorporation rate is found with semisterile and nonsterile mitochondria. The TMP-CI4 incorporated into rat liver mitochondria was recovered as 3'-TMP after enzymic digestion, showing that most of the incorporation was into internal positions of the D N A chain (P. Parsons and Simpson, 1967). While incorporation of dATP into yeast mitochondrial D N A was also insensitive to high concentrations of pancreatic deoxyribonuclease, incorporation was completely dependent on the presence of dGTP, dCTP, and d T T P in this case, and these nucleotides could not be replaced by GTP, CTP, and UTP (Wintersberger, 1966). Equilibrium density gradient analysis has provided further evidence that the D N A synthesized in zitro by isolated mitochondrial preparations is actually mitochondrial DNA. In the case of chick liver, yeast, and P . polycephalzm, the buoyant density of mitochondrial and nuclear D N A differs in CsCl (cf. Tables I and II), and it was shown that the DNA synthesized in vitro had the equilibrium density of mitochondrial D N A (Brewer et al., 1967; Wintersberger, 1968; Ter Schegget and Borst, 1968). With rat liver, the density of nuclear and mito-

154

P. RORST A N D A. M. KROON

chondrial D N A is approximately the same, but in this case it was reported that the D N A synthesized by isolated mitochondria renatured under conditions in which nuclear D N A did not renature at all (P. Parsons and Simpson, 1967). The physical properties of the D N A product synthesized by chick liver mitochondria in vitro were analyzed by Ter Schegget and Borst (1968). Rather unexpectedly, they found that up to 80% of the incorporated radioactivity, recovered by the standard D N A isolation procedure employed by Borst et ul. (1967b), was present in a DNA identical with marker component I in sucrose gradients and in CsCl gradients containing ethidium bromide. This result indirates that the deoxyribonucleotide incorporation by isolated mitochondria is not the result of aberrant copying of mitochondria1 D N A similar to that observed with bacterial D N A polymerase and double-stranded D N A in subcellular systems (cf. Schildkraut et al., 1964). In addition, the fact that radioactivity is found in closed circular duplex D N A strongly suggests that the enzyme polynucleotide ligase (Gellert, 1967; Becker et al., 1967; Little et ul., 1967) is present in mitochondria. Although the results discussed in this section show that isolated mitochondria from a number of sources are able to incorporate deoxyribonucleotides into their DNA, two major questions have not been answered: (1) Is the incorporation of deoxyribonucIeotides the result of D N A replication or D N A repair? ( 2 ) Do all mitochondria in the suspension incorporate nucleotides at the same rate leading maximally to a 1-3Yo increase in their D N A content, or is all activity the result of a minority of mitochondria replicating their D N A more extensively? Since hromodeoxyuridine is readily incorporated into D N A by isolated chick liver mitochondria (Ter Schegget and Borst, 1968), an answer to both questions should be forthcoming soon.

IX. Effects on Yeast Mitochondria1 D N A of Anaerobiosis, Glucose Repression, and Mutagenic Agents Several yeast species have the fortunate ability to grow without functional mitochondria, allowing experiments on mitochondrial biogenesis that are not readily available in obligate aerobes. Three conditions are known in which the biosynthesis of mitochondria in yeast is altered : anaerobiosis, glucose repression, and in petite mutants. The fate of mitochondrial DNA under these conditions has been explored in different laboratories and the results obtained will be discussed in this section. A. ANAEROBIOSIS

The synthesis of mitochondrial cytochromes in Sacchavomyces only occurs in the presence of oxygen (Somlo and Fukuhara, 1965). Whether other mito-

MITOCHONDRIAL D N A

155

chondrial constituents are present in anaerobically grown cells is dependent on the growth conditions chosen. If a source of fatty acids and sterols is present in the medium, numerous well-defined mitochondrialike structures are present (Lukins et al., 1966; Swift et al., 1968a). These “promitochondria” consist of a double membrane without cristae and in electron micrographs they contain typical mitochondrial D N A fibers and ribosomelike particles (Swift et ul., 1968a). When the particle fraction of a homogenate of cells grown under these conditions was centrifuged to equilibrium in a Urografin gradient, a band was found at the density of normal yeast mitochondria (Schatz, 1965). In this band, two exclusively mitochondrial enzymes, succinate dehydrogenase and oligomycinsensitive ATPase, were concentrated. It seems probable that the particles purified by gradient centrifugation are identical with the promitochondria observed in electron micrographs. N o accurate values for the mitochondrial D N A content of these cells have been reported. Swift et al. (1968a) mention that “DNA isolated from anaerobic cells still showed a satellite band of mitochondrial density, but in reduced amount,” while Fukuhara (1968) concludes that the proportion of mitochondrial D N A in anaerobic cells is “not very different” from that in aerobic cells. A major technical difficulty in these experiments is that respiratory adaptation of anaerobic yeast already occurs at very low oxygen concentrations (Somlo and Fukuhara, 1965). If yeast is grown under anaerobic conditions withozrt a source of fatty acids and sterols, the picture is entirely different (Wallace and Linnane, 1964) : Mitochondrial profiles are completely absent and in electron micrographs “only an occasional single membrane vesicle is observed within the cytoplasm” (Lukins et al., 1966). Moreover, only traces of succinate dehydrogenase could be detected (Lukins et al., 1966). On the basis of indirect arguments, Wilkie (1963) has concluded that under these conditions only one genetically active copy of mitochondrial D N A is preserved. It would be most interesting to know what the actual mitochondrial D N A content of these cells is and whether this mitochondrial D N A is present in a cytoplasmic organelle, in the nucleus, or free in the cytoplasm. N o experimental data are available on these points. Furthermore, one wonders how the anaerobic cell ensures that every daughter cell receives one copy of a master template. Some physical association with the nucleus would seem to be the simplest way to achieve this.

B. GLUCOSE REPRESSION When S. cerevisiae is grown on glucose concentrations above 6 x 10-3 M the biosynthesis of mitochondria is repressed (Slonimski, 1956). The degree of repression appears to be strictly dependent on the rate of fermentation (De Deken, 1966a). At very high glucose concentrations, fermentation is maximal and mitochondrial biosynthesis is nearly completely repressed. Con-

(o.l%),

156

P. BORST A N D A. M. KROON

sequently, mitochondrial profiles in electron micrographs of these cells are few or absent (Yotsuyanagi, 1962a,b; Polakis et al., 1964, 1965; Jayaraman et al., 1966), all cytochromes are present in very low concentrations (Reilly and Sherman, 1965), and the concentrations of several typical mitochondrial enzymes may be decreased to less than 5 % of the aerobic derepressed level (Polakis and Bartley, 1965; Jayaraman et al., 1966). When fermentation is low either because the glucose concentration is low or because other sugars are used as carbon source, for example, melibiose (Reilly and Sherman, 1965), which cannot be rapidly fermented by Sacchalomyre.r, respiration may be as high as that of cells grown on lactate. The effect of glucose repression on the proportion of mitochondrial D N A in the D N A extracted from S . cereviJiae protoplasts was studied by Moustacchi and Williamson ( 1966) using preparative CsCl density gradients. I n extracts from stationary cells, the proportion of mitochondrial D N A of the total cell D N A was 20%. After 8 hours of growth in 0.3 M (5.4%) glucose, the number o f cells had increased 30-fold and the proportion of mitochondrial D N A had decreased to about 3% of the total DNA. This decrease was not attributable to a temporary stop in mitochondrial D N A synthesis but to a decreased rate of synthesis in relation to nuclear D N A synthesis, resulting in a gradual decrease in the proportion of mitochondrial D N A . Unfortunately, the authors did not study the effect of prolonged and maximal glucose repression, and it is not known to what level the proportion of mitochondrial D N A is pushed back under those conditions. The decrease in mitochondrial D N A during glucose repression is completely reversible on removal of the glucose. An interesting exception to the general rule that glucose repression in yeast is completely reversible was reported by Negrotti and Wilkie (1968). They isolated a mutant in which all daughter cells budded off in the presence of more than 0.2% glucose, or under anaerobic conditions were found to be cytoplasmic petiteJ. The fate of mitochondrial D N A in these mutants has not yet been studied. C. MUTAGENIC AGENTS

Slonimski and co-workers have analyzed the mitochondrial D N A from two mutants, and their results are summarized in Table IX. Two main conclusions are obvious: ( 1 ) A nuclear mutation affected neither the buoyant density nor the amount of mitochondrial D N A , even though no functional mitochondri'i are made in this mutant. ( 2 ) In the cytoplasmic mutants, the density of mitochondrial D N A was changed while the amount of mitochondrial D N A was normal. In previous experiments of Corneo et al. (1966), Moustacchi dnd Williamson (1966), and Tewdri et al. (1966), little or no detectable mitochondrial D N A was found in cytoplasmic petite strains. Although this has not

petite

Mi tuchondrial

Kudcnr

Nature of genetic determinants Strain

Chromosomal

Cytoplasmic

QO, (%of norma I

density

Density

in CsCl

in CsCl (gm./cm.::)

(gm./c1m3)

* I).002

s Re1 amount (% of total)

Normal gvmd2d~. p7 rho-

Korma 1

Normal

100

1 7 0 1 fi 9.002

1.687

Chroiriosornal pc/I/?

M ti t a k d

Normal

0

1.701 i O.O(lZ

I hW7 2 0.002

1 I1

Mutad

0

1.701 & 0.001

1.683 =? 0.003

li

ii

1.792 f c).001

1.695 k 0 001

10

p7 rho+

Neutral pe/i/r p 7 rho,-

Suppressive pelite p7 rhu,-

14

recessive

Mutated . rct-cssivc

M u ~ttd a recessive

Modified from hfaunolou t t d.,1966

recessive Mutatrd dominant

7 0

r

r 0 z B 2 > r G 2

*

158

P. RORST AND A. M . KROON

been shown directly, it seems possible that the glucose repression prevailing under the culture conditions of these authors may have suppressed mitochondrial DNA synthesis to a level at which the D N A could not be detected in density gradients. Additional cytoplasmic petite mutants were studied by Carnevali et al. (1966) and Mounolou et ul. (1968). In these cases also, the buoyant density of the mitochondrial D N A was different from that of wild-type yeast, but in one mutant less than 170of the total D N A was mitochondrial against 10-14% in the wild-type cells. The major shifts in the buoyant density of the mitochondrial D N A of some of these cytoplasmic mutants could be attributable either to large scale modification of bases, e.g., by methylation, or to a major change in the GC content. Preliminary experiments of Mounolou et nl. (1968) indicate that in the mutants studied by this group, the change in density of the D N A is accompanied by a change in T,,, suggesting that changes in base composition are responsible f o r the density shifts in these mutants. This was confirmed by direct analysis for the mutant of Carnevali et al. (1966). The mitochondrial D N A of this mutant has an exceptionally low buoyant density in CsCl of 1.670 gm./cm.3 (nuclear D N A 1.699 g m . / ~ m . ~and ) it was found to contain less than 3% GC (Tecce, personal communication). Since alternating dAT has a density of 1.679 gm./cm.3 (Schildkraut et nl., 1962) while dA:dT has a density of 1.6445 gm./cm.3 (Wells and Blair, 1967), the mutant D N A must consist of random sequences of A and T to explain the density of 1.670 gm./cm.3, found by Carnevali et al. (1966). The interesting question how the density shifts in these mutants arise was discussed at length at the Round-Table Conference at Polignano in 1967. Three mechanisms were considered. ( 1 ) Large-scale deletions, which remove sections of the D N A with a base composition that 2iffers substantially from the rest of the molecule. This explanation is highly improbable, however, in view of the narrow unimodal bands observed for yeast mitochondrial D N A in CsCl gradients (Borst et ul., 1968). Since these DNA preparations consist of heterogeneous linear DNA, probably derived from larger molecules by random breakdown (see Section VI), any heterogeneity in base composition of the unbroken molecules would have shown up as density heterogeneity of the fragments. ( 2 ) Slonimski (in discussion of Mounolou et ul., 1.368) has raised the possibility that mitochondrial D N A in wild-type yeast is genetically heterogeneous. According to this hypothesis, most of the molecules have the same base composition but a minority that are too small to show up in density gradients have a base composition very different from the bulk of the mito-

MITOCHONDRIAL DNA

159

chondrial DNA. In the presence of agents inducing petite mutations such as acriflavin, mitochondria containing the minority D N A would be at an advantage and outgrow the normal mitochondria. As pointed out by Slonimski, the mitochondrial heterogeneity in normal yeast cells required in this hypothesis to explain the many different mutants already obtained would seem impossibly high. ( 3 ) The hypothesis preferred by Slonimski (in discussion of Mounolou el al., 1968) is the “mismatch hypothesis.” Agents inducing petite mutations lead to errors in the replication of mitochondrial DNA, rendering the newly synthesized D N A nonfunctional. If all replicas of mitochondrial D N A in a cell contain errors, the cell will be a cytoplasmic petite. When the dye is withdrawn the cell will at first contain many different molecules of nonsense DNA. In successive generations the nonsense mitochondrial D N A replicating most effectively will be selected for, and a petite clone with one type of mitochondrial D N A will finally be obtained. Apparently Slonimski assumes that the large-scale changes in base composition already occur during the initial incubation with the intercalating dye because he has stated that cytoplasmic mutants “result from a change in the buoyant density of their mitochondrial D N A ” (Slonimski et al,, 1968). This is not very likely, however, since only a few molecules of acriflavin per mitochondrion are sufficient to convert a yeast cell into a petite mutant (cf. Wilkie in discussion of Mounolou et al., 1968). We, therefore, prefer the idea that the primary mutagenic event may only introduce small errors in the mitochondrial DNA. The minimal error required for this is not known, but it is possible that all gene products of mitochondrial D N A are essential for the synthesis of a functional mitochondrion so that every mutation leading to one nonfunctional gene product will render the mitochondrial D N A effectively nonfunctional.2 According to this hypothesis, the altered base composition of mitochondrial D N A found in cytoplasmic petite mutants is attributable to a slow accumulation of additional errors in the mitochondrial D N A while the petite grows and divides for thousands of generations. To account for the large change in the base composition of the 2 This is not unreasonable in view of the proposed genetic functions of mitochondrial DNA: specification of mitochondrial ribosomes, specification of inner membrane proteins. and specification of regulatory proteins coordinating the contribution of nuclear and mitochondrial DNA to the biosynthesis of mitochondria (see Section XIII) . A nonfunctional ribosomal protein or ribosomal RNA will block mitochondrial protein synthesis that is indispensable for mitochondrial biosynthesis. A nonfunctional membrane protein may prevent normal buildup of the inner membrane and attachment of cytochromes. If the recombination rate is low between mitochondrial DNA’s under conditions of petize induction, single-gene mutations could therefore be lethal to the mitochondrion containing the mutated DNA.

160

P . HORST AND A. M. KROON

mitochondrial D N A of some of these mutants in relation to wild-type mitochondrial D N A , it is probably necessary to assume that large-scale replication errors contribute to the change in mitochondrial D N A . Selection for replic‘1t‘ion efficiency will eventually lead to a homogeneous mitochondrial D N A population. According to this hypothesis the mutation is not caused by the change in buoyant density, but the change in density is the result of the mutation, i.e., the effective loss of the genetic information i n mitochondrial D N A . The hypothesis implies that errors in the replication of mitochondrial D N A occur ;it low frequency in all yeast cells, but that in normal yeast cells the mitochondria containing “good” copies of mitochondrial D N A outgrow the mitochondria containing “bad” copies. Furthermore, the hypothesis predicts that by studying cytoplasmic mutants as soon as they arise, mutant mitochondrial D N A will be found with the same base composition as wild-type mitochondrial D N A . Such mutant mitochondrial D N A will have a high degree of homology with mitochondrial D N A from wild-type cells. Subculturing the mutant for thousands of generations will lead to loss of homology and eventually also to changes in base composition. These predictions are being verified in our laboratory. Since it is not conceivable that proteins with complex biological functions can be coded for by a DNA that contains only A and T, the results of Tecce and co-workers, discussed above, establish that all requirements necessary to make the defective promitochondria of cytoplasmic mutants can be met without any contribution by the mitochondrial D N A itself. Therefore, mitochondrial D N A polymerase and the proteins necessary to make the membrane of the defective promitochondria are coded for by nuclear genes (see also Roodyn and Wilkie, 1968). Whether mitochondrial D N A in other cytoplasmic petites is also completely nonfunctional is not known, but we consider this likely because cytoplasmic petite mutations, never revert, while different cytoplasmic petite mutants do not complement (see discussion after Mounolou et a/., 1968). Petite mutants are known to occur in two types: “neutral” and “suppressive” peti1e.r. In crosses to wild-type cells all daughter cells will be normal if the cross is made with a neutral petite while a variable proportion, depending on the particular mutant chosen, of the daughter cells will be petite if the cross is made with a suppressive petite (see Wilkie, 1964). ,Mounolou et al. (1968) showed that in a cross with a suppressive petite the mitochondrial D N A of the petite daughter cells (95% with this mutant) had the same density (1.696 gm./cm.3) as the D N A of the petite parent, while the mitochondrial D N A of the wildtype daughter cells had the same density as that of the wild-type parent. It is clear, therefore, that when two types of mitochondria are present in a common cytoplasm, one may outgrow the other. How this competition is effected on a molecular level is not known. However, if we accept the hypothesis that the ’

MITOCHONDRIAL D N A

161

mitochondrial D N A of all cytoplasmic petites that have been subcultured for thousands of generations is completely nonfunctional, the difference in behavior between neutral and suppressive petitej presents an interesting paradox., How is it possible that two nonfunctional DNA's behave differently in the same cell? Obviously, a difference in chemical or physical properties of the mitochondrial D N A must be responsible, unless other cytoplasmic genetic determinants are present. There is no obvious systematic difference in the buoyant density of the mitochondrial DNA's of the suppressive and neutral petites studied so far. So, either the base sequence or the size of the mitochondrial D N A must be responsible for the difference in behavior, and the rate of D N A replication must be a decisive factor in the rate of mitochondrial replication and in the competition between different types of mitochondria within one cell. Experimental verification of these ideas has been initiated in our laboratory. Further speculation on this matter is hardly fruitful until more is known about the devices present in normal cells to regulate the number of mitochondria per cell and the rate of mitochondrial multiplication. It may he mentioned in this connection that Mounolou et ul. (1,968) have found that the proportion of mitochondrial D N A in different yeast strains is genetically determined and varies between Iand 15% in diploid cells. Cytoplasmic petites are induced with high efficiency by concentrations of acridine dyes that have no apparent effect on the nuclear D N A of yeast. Two explanations have been advanced to account for the apparent preferential attack of acridines on mitochondrial DNA. (1) Tewari et a/. (1966) suggest that the difference in base composition between mitochondrial and nuclear D N A is responsible, since the association constant of the DNA-acridine complex is higher for a D N A with a low GC content than for a D N A with R high GC content. This explanation was ruled out by the demonstration by Slonimski eb a/. (1968) that ethidium bromide, which binds to D N A with low and high G C content with the same affinity (Waring, 1965), is an even more specific inducer of cytoplasmic petites than the acridine dyes. ( 2 ) At low ethidiuni bromide concentrations, the dye binds more strongly to closed circular duplex D N A than to open circles or h e a r D N A (see Section V,B). In view of this, Slonimski et u/. (1968) have proposed that "mitochondrial D N A in yeast may be natively in the form of superhelical circles and that the changes in the supercoiling after combination with the dye initiate the mitochondriril mutation." Implicit in this proposal is that yeast nuclear D N A does not consist of closed circular duplexes and is therefore much less sensitive to the dye. Two objections may be raised against this proposal. First, there is considerable doubt as to whether any supercoiling is present in closed circular

162

P. BORST AND A. M. KROON

duplex molecules in the intact cell (see Section V,C). Consequently, the difference (relatively small) in affinity for ethidium bromide between open and closed D N A found ipz vitro probably does not exist in the intact cell. Second, the mutagenic action of intercalating dyes such as ethidium bromide is considered to be the result of replication errors (Lerman, 1964). Replication of a circular D N A requires introduction of a swivel in the molecule (see Cairns, 1963), and this swivel relieves the topological restraint on which the preferential binding of ethidium bromide to closed circular duplex D N A is based. We think, therefore, that the preferential interaction of acridine dyes with mitochondrial D N A must be the consequence of a difference in organization between yeast mitochondrial D N A and nuclear DNA, or to a difference in the sensitivity of nuclear and mitochondria1 polymerase to the presence of intercalating dyes within the DNA. Difference in organization may mean various things in this context: For instance, nuclear D N A could be shielded against dye intercalation by the presence of histones or divalent cations, or the concentration of dye in the nucleus could be much lower than in mitochondria because of permeability barriers. Recently, Slonimski et al. (1968) have reported that ethidium bromide also converts nongrowing yeast into cytoplasmic petite mutants, in contrast to the acridines which lead only to mutant daughter cells without affecting the mother cells. The conversion of nongrowing wild-type yeast cells into petites by ethidium bromide followed first-order kinetics after a lag phase of about 5 hours. By extrapolating the linear part of the induction curve back to zero time, it was found that every aerobic nongrowing yeast cell contains about six targets which must be “hit” by ethidium bromide before the cell becomes a petite. This is at least one order of magnitude less than the number of mitochondria in an aerobic yeast cell, and Slonimski et ul. (1968) therefore propose that only a fraction of the mitochondria plays a role in the transmission of the cytoplasmic character. Although the results of Slonimski et al. are very clear-cut, the interpretation they provide is less compelling. It is generally accepted that mutagenesis by intercalating dyes is attributable to copying errors either in replication or in repair (Lerman, 1964). Either of these processes, therefore, has to proceed in starved, nbngrowing yeast cells. Replication in the presence of ethidium bromide leads after two rounds of replication to a 50:50 mixture of hybrid molecules containing one normal and one mutated strand and molecules containing two mutated strands. The hybrids yield normal DNA after removal of the ethidium bromide and therefore the conversion of normal to petite cells represents the loss of the last hybrid D N A molecule. This follows simple first-order kinetics but the intercept represents the number of targets x2, i.e., the target size will be

MITOCHONDRIAL DNA

163

three instead of six. Clearly, however, if recombination and repair are taking place at significant rates, the target size could be much greater, while the target size could be smaller if ethidium bromide only penetrates the cell after a lag. In view of these complications it seems premature to draw any firm conclusions from Slonimski’s ethidium bromide experiment as to the fraction of yeast mitochondria that play a role in the transmission of the cytoplasmic character. If our interpretation of Slonimski’s result is correct, the half-life of 2 hours fo; the conversion of wild-type into petite yeast under nongrowing conditions represents the rate of turnover of mitochondrial D N A under these conditions. This could be experimentally verified rather easily.

X. Recombination of Mitochondria1 DNA The work of Sager and co-workers (see Sager and Ramanis, 1965) has provided evidence that cytopIasmic determinants may recombine in Chlunzydomoms. Although the nature of the determinant was not identified in Sager’s work, it may Recombination of cytoplasmic well be the chloroplast D N A of Chlum~~domoma.r. determinants affecting mitochondrial properties was recently reported by Thomas and Wilkie (1968a) in yeast. In their experiments, they used a series of cytoplasmic yeast mutants in which the biogenesis of the mitochondria is resistant to one or another of the antibiotics erythromycin, spiramycin, or paromomycin. In previous experiments of Thomas and Wilkie, discussed in Section XIII,E, it was shown that the resistance of these mutants is attributable to a change in a cytoplasmic determinant, probably mitochondrial DNA. Crosses between two strains with a different drug-resistance marker gave rise to a very high proportion of clones resistant to neither or to both drugs, as shown in Table X. Thomas and Wilkie conclude that these clones must have arisen from a recombinationa1 event. Several points of interest should be noted. (1) The crosses tabulated in Table X were carried out anaerobically under growth conditions in which no mitochondrial structures were detectable in the cell. This apparently facilitated recombination since very few recombinants were found after crosses between cells grown aerobically. It is, therefore, still doubtful whether or not complete fusion of intact yeast mitochondria can take place. The recombinational event might even involve the (nuclear?) “master copy” of mitochondrial D N A postulated by Wilkie on other grounds (see Section IX) . ( 2 ) In nearly all cases all cells within a clone derived from a zygote, formed under anaerobic conditions, were found to contain mitochondria of one type only. Thomas and Wilkie explain this result by assuming that only one copy of mitochondrial D N A per cell is present. ( 3 ) In most of the crosses studied, there is a significant excess of multiple-

164

P. RORST A N D A . M . KROON

sensitive over multiple-resistant clones, as shown in Table X. Thomas and Wilkie suggest that this results from the recessive character of the mutational change. Recombination of circular DNA leads to a circular dimer, and if these dimers are relatively stable the recombinants will be predominantly "diploid heterotygotes," giving off occasional segregants within a clone. Furthermore, in other cases of drug-resistant ribosomes, recessivity has been seen in heterozygous TABLE X ANALYSISO F CLONESFROM INIXVIULIAL Zycon-s T N VARIOUSCROSSESOF YEAST STRAINSCARRYING DRUGRESISTANCE MARKERS SHOWING CYTOPLASMIC I NH ERITANC c'l Genotype of parentsb

Daughter clones Clone type

Number of clones 25 19 31 5

49 25

14 23 3

4 1

EHSC X ErSN

2

40 152 1

Taken in modified form from Thomas and Wilkie, 1968a. I?j, sensitivity to 10 pg. erythromycin per milliliter; El', resistance to 3 mg. erythromycin per milliliter; S s , sensitivity to 50 pg. spiramycin per milliliter; St', resistance to 2 mg. spiramycin per 'milliliter; Ps, sensitivity to 50 pg. paromomycin per milliliter; PV, resistance to I mg. paromomycin per milliliter IL

*

>

diploids, apparently because the sensitive ribosomes get stuck on the polysonie and block the process of the resistant ribosomes (see Cooper et d.,1967). These most interesting experiments underline the need for detailed facts about the structure, amount, and intracellular localization of yeast mitochondria1 DNA under various conditions. They also focus attention again on the problem of selection within a heterogeneous population of mitochondri'i. If the cell were unabIe to continuously select the mitochondria most suitxble to survival, recombrnation would be useless.

MITOCHONDRIAL DNA

16j

XI. Renaturation Studies with Mitochondria1 DNA The renaturation of D N A , i.e., the formation of an ordered double helix from complementary single strands, was discovered by Marmur and Doty and their co-workers (see review by Marmur et ul., 1963). They demonstrated that the renaturation reaction follows second-order kinetics and that the renatured D N A had the same melting point as the native starting material prior to denaturation, indicating that a perfect double helix was indeed re-formed. In addition, they noted that the rate of renaturation was strongly dependent on the source of the D N A : Viral DNA’s renatured faster than bacterial DNA’s and these renatured faster again than the nuclear DNA’s from eucaryotes. The quantitative aspects of D N A renaturation were studied in more detail by Britten and co-workers (Britten and Waring, 1965; Britten and Kohne, 1966) and Wetmur and Davidson (1968). They showed that under standard conditions of salt, temperature, and D N A fragment size, the second-order renaturntion constant was a linear function of the complexity of the DNA. The complexity of D N A is defined as the number of base pairs in the genome, disregarding repeated sequences. Since the complexity of mitochondrial DNA’s is likely to be small, it is not surprising that mitochondrial D N A from birds (Borst and Ruttenberg, 1966a; Borst et ul., 1967a,b; Dawid and Wolstenholme, 1968a), rodents (Borst and Ruttenberg, 1966a; Borst et ul., 1967a; Corneo et al., 1966; Flamm et ul., 1966; Sinclair and Stevens, 1966; P. Parsons and Simpson, 1967), frog.eggs (Dawid and Wolstenholme, 1968a,b), and yeast (Tewari et al., 1966; Sinclair et ul., 1967a) was found to completely renature with a very high speed, comparable to that of the D N A of the smaller D N A viruses. Although the obvious interpretation of these experiments was that the complexity of mitochondrial D N A is very low, the almost imtantaneous renaturation observed under optimal conditions made it necessary to exclude the alternative interpretation that complete strand separation during denaturation did not occur because of interstrand crosslinks (Geidushek, 1961, 1962) or the presence of a large fraction of component I. This alternative was eliminated by Borst et al. ( 1 967a,b). They used a chick liver mitochondrial D N A preparation containing only component I1 in their renaturation experiments and showed that renituration followed second-order kinetics and that it was strongly salt-dependent. Both results are incompatible with the presence of cross-links which leads to renaturation independent of salt or D N A concentration (cf. Geidushek, 1961, 1962). By comparing the renaturation constant obtained for mitochondrial D N A with the renaturation constants determined by Britten and Kohne (1966) for DNA’s of various complexity, Borst et ul. (1967b) calculated a maximal complexity for mitochondrial D N A of about 12,000 base pairs. In view of the correction factors involved, this value could be in error by 5 0 % . However, the important point is that these

166

P. BORST AND A. M. KROON

experiments set a maximum to the genetic information contained in chick liver mitochondrial DNA, and they strongly suggest that the mitochondrial D N A of chick liver is not only homogeneous in site and base composition but also in base sequence. In view of the fact that mitochondrial D N A in chick liver is packaged in molecules of 15,000 base pairs, it seems reasonable to assume that 15,000 base pairs represent the maximal genetic information available in chick liver mitochondrial DNA. Maximal should be stressed in this context because the experiments of Borst et ul. (1967b) do not exclude redundancy in mitochondrial DNA. An indication that redundancy might be present in mitochondrial D N A from Xenopus eggs was obtained by Dawid and Wolstenholme (1968a). They studied renatured mitochondrial D N A and found a few open circles with a contour length much lower than 5 p and without any sign of single-stranded regions. Although this might be the result of redundancy of Xenopus mitochondrial DNA, Dawid and Wolstenholme ( 1968a) point out that only a very small proportion of the D N A occurred in the form of small circles and “these could be the renaturation product of a different type of D N A which contaminated at a low level the mitochondrial D N A preparations.” Another point that should be stressed is that the matching precision required for renaturation is not complete. Therefore, the renaturation experiments discussed above do not exclude a microheterogeneity in the mitochondrial D N A population attributable to point mutations or small insertions or deletions. Such microheterogeneity can only be excluded by demonstrating its absence in the gene products of mitochondrial DNA. Although quantitative renaturation studies have not been made with mitochondrial DNA’s from sources other than chick liver, there is no reason to doubt that ;he rapid renaturation in all these cases is also a result of the low complexity of these DNA’s3 It will be of great interest, however, to determine by quantitative renaturation studies the genetic information content of mitochondrial D N A from lower organisms, such as yeast, and to extend the rather limited renaturation experiments of Borst et d. (1967b) with mitochondrial D N A of higher organisms. 3 Du Buy et al. (1966) have reported that D N A from mouse brain nuclei renatured to an extent of 20% only and they concluded from this result that mouse mitochondrial D N A is comparatively heterogeneous in base sequence. Since the absence of nuclear D N A in the mitochondrial D N A preparations of D u Buy et al. (1966) was not demonstrated in any way, a more liktly interpretation of their result is that their “mitochondrial” D N A preparations contained 20% mitochondrial and 80% nuclear D N A . This interpretation is supported by two observations. Mouse liver mitochondrial D N A renatures completely (Borst and Ruttenberg, 1966a; Flamin ef d., 1966; Sinclair and Stevens, 1966), and D N A preparations extracted from mouse brain mitochondria contain the same circular molecules in electron micrographs as mitochondria1 D N A from mouse liver (Sinclair et ul., I967b).

MITOCHONDRIAL D N A

167

XII. Evolution of Mitochondrial D N A and the Relation between Mitochondria1 and Nuclear DNA A quantitative analysis of the sequence homology of mitochondrial DNA’s of different organisms and the homology of mitochondrial and nuclear D N A requires quantitative renaturation experiments. Only one study of this kind has been published. Du Buy and Riley (1967) studied the hybridization of C14labeled mouse brain nuclear D N A fragments with either mouse brain “mitochondrial” or nuclear D N A immobilized on nitrocellulose membrane filters. While 23.2% of the nuclear D N A bound to filters with nuclear DNA, 12.2% bound to filters with “mitochondrial” DNA. Although D u Buy and Riley (1967) conclude from these results that 46% of all nuclear base sequences are represented on mitochondrial D N A in mouse brain, we prefer the conclusion (cf. Dawid and Wolstenholme, 1968a,b) that 46% of the “mitochondrial” D N A preparation of the authors consisted of nuclear D N H . The renaturation data of Du Buy et al. (1966) for these D N A preparations even suggest a higher degree of contamination, as pointed out in the previous section. A series of ingenious qualitative hybridization experiments was performed by Dawid and Wolstenholme ( 1968a,b) employing the “concatenation” phenomenon first studied by Britten and Waring (1965). These authors showed that renaturation for very long periods of time leads to the formation of D N A complexes of very high molecular weight because the single-stranded stretches remaining in partially renatured molecules anneal with complementary stretches in other partially renatured molecules. If two DNA’s have sequences in common they will form a common complex. The minimal degree of sequence complementarity required for a common complex is probably identical with the number of base pairs required for the formation of a stable duplex at the temperature of annealing, i.e., about 1 2 base pairs (Niyogi and Thomas, 1967). Completely unrelated DNA’s, such as plant and animal nuclear DNA, do not form a common complex during coannealing (Britten and Waring, 1965). The complexes are detected by analytical CsCl equilibrium gradient centrifugation and since the molecular weight of the complexes is very high, very sharp bands are obtained allowing the distinction of DNA’s with small differences in density. When this method was employed, no coannealing was detected between mitochondrial and nuclear D N A of X . laeuis and Rana pipiens, or between mitochondrial DNA’s of Xetzopm and yeast (Dawid and Wolstenholme, 1968a,b). However, a common complex was found between coannealed Xennpus mitochondrial D N A (native density = 1.702 gm./cni.3) and chick liver mitochondrial D N A (native density = 1.709 gm./cm.“). These experiments exclude a general large-scale homology of mitochondrial and nuclear D N A of the type

168

P . RORST A N D A. M . KROON

envisaged by Du Buy and Riley (1967), but they would be compatible with the presence of one or several master copies of mitochondrial D N A in the nucleus (Dawid and Wolstenholme, 1968a,b). T h e coannealing of mitochondrial D N A from chicks and frogs proves that these two DNA's have at least a sequence of 12 base pairs in common, if we accept the rather extensive evidence for the specificity of the coannealing reaction. This sequence complementarity is unexpected, as these DNA's differ as much as 7 mg./cm." in their buoyant density in CsCI, while they are both considered homogeneous in base Composition in view of the sharpness of their thermal transition profiles (Borst el al., 1967b; Tewari et al., 1766).

XIII. Genetic Function of Mitochondria1 DNA A. INTRODUCTION The possible genetic function of mitochondrial D N A has been studied in five ways. (1) Quantitative D N A - D N A renaturation rates were used by Borst et ul. ( 1967a,b) to determine the information content of mitochondrial DNA. (2) RNA components complementary to mitochondrial D N A were identified in several laboratories and further analysis of these RNA components was used to find out whether mitochondrial D N A codes for ribosomal and/or tRNA and whether mitochondrial mRNA is translated only in the mitochondrial matrix space, or also exported into the cytosol. ( 3 ) Attempts were made to identify the products of mitochondrial protein synthesis it? vivo and it? vitro in the hope that proteins synthesized inside the mitochondrial matrix space are coded for by mitochondrial D N A . (4) The range of mitochondrial enzymes that could be specified by mitochondrial D N A was narrowed down by the localization of the structural gene for cytochrome c on the nuclear D N A of yeast and by the fact that several mitochondrial enzymes are still found in cytoplasmic yeast mutants in which the genetic information of mitochondrial D N A is probably completely lost. ( 5 ) Attempts were made to correlate changes in the amino acid sequence of certain mitochondrial proteins with changes in the base sequence of mitochondrial DNA.

T h e results of the DNA-DNA renaturation studies, which indicate that the maximal information content of chick liver mitochondrial D N A is that contained in a molecule of 15,000 base pairs, have been discussed in Section XI. The experiments bearing on points ( 2 ) - ( 5 ) will be summarized in this section.

MITOCHONDRIAL DNA

1 69

B. DNA-RNA HYBRIDIZATION EXPERIMENTS In principle, DNA-RNA hybridization experiments should be able to resolve the following questions: (1) Are there RNA species present in the cell, either inside or outside the mitochondria, with a base sequence complementary to that of mitochondrial D N A ? Is the mitochondrial complementary R N A mRNA, ribosomal RNA, or tRNA? ( 2 ) Do mitochondria contain RNA species complementary to nuclear D N A ? Are these species unique or are they present both inside and outside the mitochondria S The second question was studied by Humm and Humm (1966) by hybridizing mitochondria1 RNA and nuclear RNA from mouse embryos, labeled for 20 hours with P3?, with mouse nuclear DNA. Both mitochondrial and nuclear RNA combined with nuclear D N A to the same extent, and in competition experiments cold mitochondrial RNA competed even more efficiently than cold nuclear RNA for the sites occupied by PZ-labeled nuclear RNA on the nuclear DNA. Humm and Humm conclude “that at least a part of the mitochondrial RNA has base sequences in common with nuclear RNA.” However, the competition experiments, taken at face value, actually indicate that all major nuclear RNA species labeled in a 20-hour pulse are represented as major R N A components in mitochondria. W e cannot accept this conclusion for two reasons. First, it is highly unlikely that mRNA’s for all extramitochondrial proteins would be present within the mitochondrial matrix space; second, Church and McCarthy (1967) have shown that bulk cytoplasmic RNA competes ineffectively with nuclear RNA for sites on nuclear D N A and that a sizable fraction of nuclear RNA appears not to be present in the cytoplasmic R N A at all. It is possible that the results of Humm and Humm (1966) must be ascribed to false hybridization and false competition, since neither the specificity of the hybridization nor the specificity of the competition was demonstrated by suitable control experiments. However, other artifacts cannot be excluded. The question whether or not RNA copies of mitochondrial D N A are present in the cell was studied in three laboratories. Suyama (1967) isolated two RNA fractions from Tetvuhymenu mitochondria. One had the sedimentation characteristics of tRNA, the other fraction, called pRNA, was found in the pellet after centrifugation of a mitochondrial lysate for 2 hours at ~ 0 0 , 0 0 0x g. The pRNA sedimented as a 1 : l mixture of 18-s and 14-S components through sucrose gradients. pRNA hybridized with niitochondrial D N A to a plateau value of 6.8% RNA/DNA, while the small hybridization of the sRNA fraction with mitochondrial D N A could be accounted for by the presence of pRNA fragments in the sRNA fraction. No competition between postmitochondrial R N A and pRNA for mitochondrial D N A could be demonstrated. Although some binding

170

P. BORST A N D A. M. KROON

of mitochondrial pRNA and tRNA to nuclear D N A was found, the significance of this was considered doubtful in view of the high blanks. Two arguments (see Suyama, 1967) indicate that pRNA is the ribosomal RNA of Tetrahymena mitochondria. pRNA is the main RNA component representing about 70% of the total mitochondrial RNA. pRNA does not compete with the RNA synthesized by Tetrahymena mitochondria in vitro, which is prethe absence of viral D N A in this fraction was not verified. The membrane fracsumably mRNA for mitochondrial proteins (see Suyama and Eyer, 1968). Assuming that pRNA is actually ribosomal RNA, Suyama (1967) calculates that Tetvahymena mitochondrial D N A has a molecular weight of 30 x lo6, provided that the plateau of 6.8% RNA/DNA is correct, that the total molecular weight of the RNA of mitochondrial ribosomes is 2 x 10° daltons and that only one copy of this RNA is present per D N A molecule. In our opinion, Suyama’s experiments provide strong evidence that, at least in Tetrahymena, mitochondrial D N A codes for mitochondrial ribosomal RNA. The existence of RNA fractions in yeast complementary to yeast mitochondrial D N A was studied by Fukuhara (1967, 1968). He showed that RNA from aerobic cells labeled for many generations with P32 hybridized about twice as well to mitochondrial D N A as RNA from anaerobic cells, while cold RNA from anaerobic cells competed less effectively with RNA-P32 from aerobic cells than cold RNA from aerobic cells. N o such differences were observed in hybridizations with nuclear DNA. Fukuhara concludes from these results that preferential transcription of mitochondrial D N A takes place during respiratory adaptation. Maximal hybridization obtained in these experiments was about 2.5% RNA/DNA and no plateau value was reached either in the hybridization or competition experiments. It is, therefore, not possible to conclude that aerobic cells contain RNA fractions complementary to mitochondrial D N A which are absent in anaerobic cells. From a membrane-rich fraction containing mitochondrial marker enzymes, Fukuhara (1967) extracted a metabolically stabIe RNA hybridizing to a maximal level of 1.5% (no plateau reached) with mitochondrial DNA. In sucrose gradients, the RNA complementary to mitochondrial D N A sedimented in a broad band with a peak sedimentation coefficient of 1 2 S. Although these findings are compatible with the hypothesis that the RNA of mitochondrial ribosomes in yeast is specified by mitochondrial DNA, further experiments are necessary to prove this. Recently, Attardi and Attardi (1967) have reported the isolation from HeLa cells of an extramitochondrial RNA fraction which specifically hybridized to a high degree with HeLa cell cytoplasmic DNA. The RNA was associated with extramitochondrial cytoplasmic membranes and after a 30-minute pulse of uridine-H3, about twice as much of the newly synthesized RNA appeared in the

MITOCHONDRIAL DNA

171

region of membrane-associated RNA as in the region of free polysomes, indicating its quantitative importance. The authors suggest that this RNA is mRNA synthesized with mitochondrial D N A as template and exported into the cytoplasm. The obvious inference is that the limited amount of information available in mitochondrial D N A also contributes to the synthesis of extramitochondrial components. Unfortunately, the evidence on which this most interesting conclusion was based is not complete. The “cytoplasmic” D N A used in the hybridization experiments was not identified as mitochondrial D N A by any criterion and the absence of viral D N A in this fraction was not verified. The membrane fraction containing the rapidly labeled RNA W I ~ Sonly characterized by its density in sucrose equilibrium gradients. When cells were homogenized in the absence of Mg.++, the RNA fraction was found at a density of 1.180 gm./cm.3 against a density of the mitochondrial fraction (identified by the A415) of 1.195 g m . / ~ m . ~when ; Mg+ + was present during homogenization, the profiles of &,, A260,and acid-insoluble radioactivity coincided with a peak value of 1.190 gm./cm.3. In our opinion, such a characterization of mammalian cell fractions is not adequate: First, the A,,, of mitochondrial suspensions is largely attributable to nonspecific light-scattering and not to the gamma band of cytochrome c as the authors apparently assume. Second, a density of 1.180 g m . / ~ m .does ~ not agree very well with the density of 1.13 gm./cm.3 reported for rat liver smooth endoplasmic reticulum (D. F. Parsons, 1966). Therefore, even if the band at 1.190-1.195 is really attributable to mitochondria it remains difficult to exclude that the shoulder at 1.180 represents newly synthesized mitochondria with a higher outer membrane to inner membrane ratio than the rest of the mitochondrial population. [The equilibrium density of the pure outer membrane fraction of rat liver mitochondria is 1.13 gm./cm.3 in sucrose (D. F. Parsons, 1966).] This explanation is in agreement with the observation of the Attardis (1967) that the “bulk of the membrane-associated RNA” sedimented already after centrifuging for 10 minutes at 8100 x 6. Sedimentation of the bulk of the microsomes under these conditions would be rather unexpected. Last, it should be noted that the complete and instantaneous inhibition of the incorporation of uridine into the membrane-bound RNA fraction by actinomycin D, while incorporation into polysomal RNA continued, does not support a mitochondrial orlgin of this RNA, in view of the report by Neubert et al. (1968b) that intact mammalian mitochondria are completely impermeable to actinomycin, resulting in a complete resistance to this inhibitor of mitochondrial RNA synthesis in vivo. It c m not be excluded, however, that HeLa cell mitochondria are different in this respect. In view of these discrepancies, two alternative explanations for the results of the Attardis should be seriously considered: (1 ) The rapidly labeled RNA fraction is mRNA complementary to a D N A virus present in the HeLa cell culture.

172

P. BORST A N D A. M. KROON

The D N A of this virus is present in the cytoplasmic D N A preparations used in the hybridization experiments. ( 2 ) The rapidly labeled RNA fraction is present in newly synthesized mitochondria, permeable to actinomycin. As long as these alternatives have not been excluded, the concept of mitochondrial mRNA being exported into the cytosol remains unproved. C. THEPRODUCTO F MITOCHONDRIAL PROTEINSYNTHESIS Amino acid incorporation into protein by isolated mitochondria is inhibited by actinomycin (Kroon, 1965; Neubert et al., 1968b), provided the mitochondria are damaged to render them permeable to the drug. This has led to the hypothesis that mitochondria1 protein synthesis iu ziitro is dependent on the continuous generation of mRNA, synthesized on mitochondrial D N A (Kroon, 1965, 1966a,b). If this hypothesis is correct, identification of the products of protein synthesis by isolated mitochondria will directly provide a list of proteins specified by mitochondrial DNA. Unfortunately, amino acid incorpordtion by isolated mitochondria takes place in the insoluble proteins (Roodyn et al., 1962; Truman, 1964; Wintersberger, 1965; Bronsert and Neupert, 1966; Wheeldon and Lehninger, 1966) associated with the inner membrane (Neupert et al., 1967, 1968), and all attempts to obtain incorporation into well-defined proteins have given equivocal or negative results. Kalf and GrGce (1964) recovered a large fraction of the amino acids incorporated by isolated calf heart mitochondria in a purified “contractile protein” fraction. Since the exisitence of ‘1 contractile protein in mitochondria is highly doubtful (cf. Conover and Biriny, 1966), the significance of Kalf’s results is not clekr. Labeling of protein fractions with the electrophoretic mobilities of the mitochondrial F, ATPase and coupling factor F4 (see review by Pullman and Schatz, 1967) was reported by Work and co-workers (Haldar et d.,1966; Work, 1967, 1968). Inspection of their experimental data reveals, however, that radioactivity was smeared all over the electropherogram. In the absence of a clear-cut fractionation every fraction will be labeled, and the specificity of the labeling remains to be proved. Other investigations have shown that no amino acids are incorporated into either cytochrome c (Roodyn et a!., 1962) or cytochrome aag (Kadenbach, 1968). Although continuing work along these lines may be more successful, two basic objections to this approach at the genetic function of mitochondrial D N A may be mentioned: (1) There is no evidence that complete proteins are synthesized by isolated mitochondria. The chances of identifying incomplete proteins are very small if more than one protein is made. ( 2 ) In view of the limited amount of genetic information contained in mito-

MITOCHONDRIAL DNA

173

chondrial DNA, most of the structural genes for the synthesis of mitochondrial proteins must be present on nuclear DNA. Coordination of the contributions of mitochondrial and nuclear D N A to the synthesis of a mitochondrion will require a system of inducers and repressors. During the isolation of the mitochondria these may be lost or inactivated and the protein synthesized by isolated mitochondria may not be representative for the proteins synthesized under the direction of mitochondrial mRNA in the intact cell. D. IDENTIFICATION O F MITOCHONDRIAL PROTEINS CODEDFOR BY NUCLEAR DNA, OR SYNTHESIZED OUTSIDE T H E MITOCHONDRIA Identification of the mitochondrial components coded for by mitochondrial DNA could in principle also be made by elimination, i.e., by identifying the components specified by nuclear DNA, because it is unlikely that structural genes localized on nuclear D N A are represented in identical form on mitochondrial DNA. The only mitochondrial protein for which this has been done as yet is cytochrome c. Sherman and co-workers ( 1966) have conclusively demonstrated that a mutation in the yeast nuclear gene CY1 leads to a change in the amino acid sequence of iso-1-cytochrome c, the major cytochrome c of yeast. The mRNA for cytochrome c is translated outside the mitochondria, at least in rat liver, because pulse-labeling experiments by Gonzilez-Cadavid and Campbell (1967a,b) have shown that the nascent cytochrome c of rat liver is first found in the microsomal fraction and subsequently transferred to the mitochondria. Similar results were obtained by Freeman et al. (1967) with Krebs ascites tumor cells. Following this approach, Beattie et al. (1966) studied the appearance of labeled amino acids in the soluble and membrane-bound protein fractions of the mitochondria of different rat organs. The rise in specific activity of the soluble proteins was somewhat less rapid than that of membrane-bound proteins, and from this difference Beattie et al. (1966) concluded that the soluble mitochondrial proteins are synthesized on microsomes and then transferred to the mitochondria, while the membrane-bound proteins are synthesized in situ. It is not clear, however, which part of this differential labeling is the result of a difference in turnover of the two protein fractions, a possibility dismissed by Beattie et al. (1966). Moreover, it. must be stressed that the site of synthesis of a mitochondrial protein does not necessarily define the site of its structural gene. The Attardis (1967) have claimed that mitochondrial mRNA is exported into the cytoplasm in HeLa cells, and it can also not be excluded on the evidence now available that nuclear mRNA is translated within the mitochondrial matrix space, as we have pointed out earlier (Borst et al., 1967a). W e conclude that experiments similar to those of Sherman et al. (1966) for cytochrome c may narrow the range of proteins that could be specified by mitochondrial DNA. Study of the intracellular localization of nascent mitochondrial proteins, although less

174

P . BORST A N D A. M. KROON

conclusive than the genetic experiments may also yield useful information if specific proteins are studied.

E. MITOCHONDRIAL ENZYMESFOUNDIN CYTOPLASMIC Petite MUTANTSO F YEAST The recent demonstration that in cytoplasmic mutants of yeast the genetic information contained in mitochondrial D N A is completely lost (see Section IX) has given experimental support to another method for identifying the gene products of mitochondria1 D N A by elimination. Mitochondrial proteins still present in cytoplasmic mutants must be specified by nuclear genes. The existence of organized membrane structures containing typical mitochondrial enzymes in cytoplasmic yeast mutants was first reported by Linnane and Still (1956). Their findings have been extended by Schatz et al. (1963) and others (Mahler et al., 1964; Mackler et al., 1965; Katoh and Sanukida, 1965; Clark-Walker and Linnane, 1967). The membraneous structures consist of a double membrane without the characteristic cristae of aerobic wild-type yeast mitochondria (Linnane and Still, 1956; Yotsuyanagi, 1962a,b; Schatz et al., 1963). The typical mitochondrial enzymes detected include succinate dehydrogenase (Linnane and Still, 1956; Schatz et al., 1963; Mackler et al., 1965; Clark-Walker and Linnane, 1967), antimycin-sensitive NADH-cytochrome c reductase, and D-lactate and Llactate cytochronie c reductases (Mahler et al., 1964; Mackler et al., 1965). Mitochondrial “structural protein” from petite mutants was initially reported (Katoh and Sanukida, 1965) to be identical to that of wild-type yeast. A more detailed investigation (Tuppy and Swetly, 196S), employing polyacrylamide electrophoresis and immunological studies revealed, however, that the structural protein of wild-type yeast consisted of several components, one of which was missing in preparations from a cytoplasmic petite mutant. The ability of yeast structural protein to bind ATP in an atractyloside-sensitive process was also found in yeast mutants. However, the mutant structural protein lost its ability to bind nucleotides when it was extracted in the cold. The authors speculate that the petite mutation causes the loss of a component present in the structural protein fraction and thereby induces cold lability of nucleotide binding. They refer to unpublished experiments of Schatz which indicate that the ATPase (F,) of the mutant yeast mitochondria is cold-labile in situ, while the wild-type enzyme is cold-stable when bound to mitochondria. These experiments show that the mitochondrial structural protein fraction minus one component is present in cytoplasmic yeast mutants. Since structural protein sensu strict0 and the enzymes mentioned above are probably confined to the mitochondria in yeast, it is likely that they are specified by nuclear genes. Similar considerations hold for mitochondrial D N A polymerase and RNA polymerase which were detected in

MITOCHONDRIAL DNA

175

the heavy particle fraction of cytoplasmic yeast mutants by Wintersberger (1968). In addition to the enzymes enumerated above, aconitate hydratase (Schatz et al., 1963), fumarate hydratase (Schatz et al., 1963; Clark-Walker and Linnane, 1967), and malate dehydrogenase (Clark-Walker and Linnane, 1967) have been detected in cytoplasmic petites. Since the presence of two malate dehydrogenase isoenzymes in yeast, one confined to the cytosol, the other predominantly found in the mitochondrial fractions, has been reported by Witt et al. (1966), it is necessary to show that the malate dehydrogenase activity found in mutant yeast is not entirely attributable to the cytosol isoenzyme before the conclusion (cf. Clark-Walker and Linnane, 1967) that this enzyme is specified by nuclear D N A is firmly established. An indication that this conclusion is correct for fumarate hydratase can be found in the observation by Schatz et al. (1963) that the heavy particle fraction from mutant yeast contains this enzyme (specific activity 24% of wild-type control). Moreover, it was recently shown in this laboratory that the aconitate hydratase activity of wild-type yeast has the same distribution in cell fractionation studies as cytochrome oxidase. This suggests that yeast aconitate hydratase is an exclusively mitochondrial enzyme and, as the enzyme is found in cytoplasmic petites, it must be specified by nuclear DNA. Only cytochromes aa3, b, and c1 have been conclusively shown to be absent from rho- cells. The conclusion (cf. Roodyn and Wilkie, 1968; Linnane, 1968) that these cytochromes are therefore coded for by mitochondrial D N A is not warranted, however, since a protein coded for by mitochondrial D N A may exert a very tight control over the synthesis of these cytochromes. The situation may be similar to that observed with succinate dehydrogenase in anaerobic wildtype cells grown in the absence of a source of fatty acids. The repression of succinate dehydrogenase synthesis under these conditions is so effective that less than 1% of the activity found in aerobic cells is present (Lukins et al., 1966). It is noteworthy in this connection that the synthesis of cytochrome aa3 is apparently dependent on the synthesis and function of other cytochromes (Reilly and Sherman, 1965). For instance, cytochrome aa3 synthesis is blocked in yeast strains grown in the presence of antimycin A, which inhibits electron transport between cytochromes b and c without affecting the synthesis of cytochromes b or c (Ycas, 1956). In principle, therefore, the study of mitochondrial proteins in cytop!asmic mutants can only yield information on proteins not specified by mitochondrial DNA. An attempt has been made by Rabinowitz el al. (1968) and Yu el al. (1968) to obtain direct evidence on the nature of the proteins specified by mitochondrial D N A by studying the induction of cytochrome synthesis by oxygen in anaerobic yeast in the presence of cycloheximide. Mitochondria1 protein synthesis is not

176

P. RORST A N D A. M . KROON

affected by this drug, while extramitochondrial protein synthesis is effectively inhibited. Rabinowitz et al. (1968) found no cytochrome induction at all in the presence of cycloheximide, while Yu et al. (1968) detected measurable cytochrome oxidase synthesis at cycloheximide concentrations that completely blocked growth and cytochrome c synthesis. There is some doubt, however, whether or not respiratory adaptation had started in the experiment of Yu et al. (1968) prior to cycloheximide addition during harvesting of the cells (see discussion after Yu et a/., 1968). The study of cytochrome synthesis by spectral analysis of cell suspension represents a simple but insensitive test system. It seems possible that this approach could be exploited more effectively by studying the specific activity of specific mitochondria1 proteins in cells pulse-labeled with a radioactive amino acid after addition of cycloheximide. 1:.

CORRELATION O F CHANGES I N MITOCHONDRIAL PROTEINS WITH CHANGES IN MITOCHONDRIAL DNA

Ideally, proteins specified by mitochondrial DNA should be identified by correlating a change in the amino acid sequence of the protein with a change in the base sequence of mitochondrial DNA. In practice, the mutants required for this analysis may be difficult to select and only two examples have been reported. Woodward and Munkres (1966, 1967) and Munkres and Woodward (1966) studied the amino acid composition of mutants mi-1 and mi-3 of N. crarsn. Both mutants are characterized by respiratory deficiency because of the absence of cytochromes. Both mutations show cytoplasmic inheritance. The structural protein of mi-1 contained one tryptophan residue less and one cysteine residue more than wild-type structural protein, while the structural protein of nzi-3 only contained one tryptophan less than its wild-type counterpart. Woodward and Munkres (1966) explain the pleiotropic character of the mi-1 and mi-3 mutants by assuming that structural protein provides the framework to which all membrane-bound mitochondrial enzymes are attached. A change in this framework as a consequence of an amino acid replacement in the structural protein leads to faulty attachment, or no attachment at all, leading to a nonfunctional mitochondrion. More recently, Woodward and Munkres ( 1967) have extracted structural proteins from cell fractions other than mitochondria. The surprising result obtained was that all cell fractions contained enormous amounts (40Cj, of all cytosol proteins) of structural protein of very similar composition. The structural proteins extracted from nuclei, mitochondria, microsomes, and cytosol were indistinguishable in amino acid composition and immunological behavior, and their peptide maps were very similar. The structural proteins extracted from all cell fractions of mi-1 mutants were found to contain one tryptophan more and one cysteine less than their wild-type counterparts. Woodward and Munkres

MlTOCHONDRlAL D N A

177

(1967) conclude from these results that the structural proteins of all cellular membranes are identical and coded for by mitochondrial DNA. Although the experimental evidence presented by Woodward and Munkres for this view is rather extensive, three points of doubt remain. (1) Recent work in the laboratories of Green (Green and Perdue, 1966) and Racker (Fessenden et al., 1966) suggests that structural protein is not required for the reconstitution of submitochondrial particles that catalyze oxidative phosphorylation, and the nature and function of this protein in mitochondria is not clear at present. ( 2 ) According to Allmann et al. (1967) the structural protein, prepared by the method of Criddfe et a/. (1962), employed by Woodward and Munkres (1966), is not a homogeneous protein and it still contains about 20-2570 contaminating proteins as judged by polyacrylamide gel electrophoresis. It is difficult to see how a meaningful amino acid composition could be obtained with an impure protein. ( 3 ) Tuppy and Swetly (1968) have recently reported that structural protein from 5. cerezisiae mitochondria consisted of several components, one of which was missing in a cytoplasmic petile mutant, in which no functional mitochondrial D N A is thought to be present (Section I X ) . No structural protein, defined as protein able to bind ATP in an atractyloside-sensitive way, could be extracted from cell fractions other than the mitochondria in wild-type yeast. Since it is difficult to imagine that a fundamental aspect of cell physiology, such as the synthesis of membrane proteins, could be arranged differently in related Ascomycetes, the results of Tuppy and Swetly with Saccharomyre.1 are difficult to reconcile with those of Woodward and Munkres (1967) with Neuro.rporn.

A clarification of these three points of doubt will be necessary before we can accept the conclusion that mitochondrial structural protein is coded for by mitochondrial DNA. A different approach was followed by Wilkie and co-workers and Linnane (Wilkie et a/., 1967; Thomas and Wilkie, 1968a,b; Roodyn and Wilkie, 1968; Wilkie, 1968; Linnane, 1968) (see also Section XIV) in their studies with yeast. They isolated a series of mutants resistant to antibiotics such as chloramphenicol, which inhibits mitochondrial protein synthesis (see Kroon, 1965, 1966a; Huang et a/., 1966; Borst e t a/., 1967a; Clark-Walker and Linnane, 1967). In the case of erythromycin resistance, mutants were obtained which showed cytoplasmic inheritance, indicating that resistance was controlled by a gene product of mitochondria1 DNA. In principle, resistance to erythromycin could be the result of: impermeability of the cell membrane or the mitochondrial membrane, the presence of an enzyme inactivating the drug, or a change in the mitochondrial ribosome. To decide among these alternatives Thomas

178

P. RORST AND A. M . K R O O N

and Wilkie (1768b) grew the erythromycin-resistant mutants under strict anaerobiosis in the absence of fatty acids. Under these conditions, mitochondrial profiles disappear completely or nearly completely and the mitochondrial membranes apparently can not be synthesized (see Section IX). Two nuclear erythromycin-resistant mutants lost their resistance completely under these conditions, suggesting that low permeability to erythromycin of either the mitochondrial or the cell membrane was responsible for resistance in these cases. On the other hand, the conclusion that erythromycin resistance in the cytoplasmic mutants is attributable to a change in the mitochondria1 protein-synthesizing system itself, is further supported by the observation that amino acid incorporation by the mitochondria isolated from one of these mutants was also resistant to erythromycin in vitvo (Linnane, 1968). In bacterial systems erythromycin is now thought to act on the ribosome at 'I site close to, but not identical with, that attacked by chloramphenicol (see Cundliffe and McQuillen, 1967). Resistance to these antibiotics is presumably the result of a change in a ribosomal protein. Wilkie's experiments therefore suggest that at least one of the mitochondrial ribosomal proteins is coded for by mitochondrial DNA. The isoldion and characterization of drug-resistant yeast mutants is comparatively easy, and it is to be expected that the analysis of mitochondrial genetics using drug resistance markers will be one of the most promising ways of analyzing the genetic function of mitochondrial DNA available at present (see Wilkie et d.,1967; Wilkie, 1968).

G. CONCLUDING REMARKS It is clear from the experimental results discussed in this section that the outlines of the genetic function of mitochondrial D N A are beginning to emerge. In mammals not more than 15,000 base pairs are available, and it is clear that only a small fraction of mitochondrial components can be specified by these. Suyama's results suggest that in Tetrabymena these components include the rRNA but not the tRNA of mitochondria, while in yeast the genetic experiments of Thomas and Wilkie implicate a ribosomal protein. Taken together these findings suggest that complete mitochondrial ribosomes could be specified by mitochondrial DNA. Ribosomes of E. cnlr contain about 5000 nucleotides (see Stanley and Bock, 1765) and at least 50 different proteins (Moore et al., 1766; Traut e f al., 1967). The regulated synthesis of these components requires at least 30,000 base pairs, unless ribosomal proteins are unusually small, which is very unlikely from the work of Moore et al. (1766) with E . coli. Although this amount of genetic inform'ition may be available in yeast and other lower organisms, it is already twice the amount present in chick liver mitochondria. It is possible that some of the ribosoinal proteins of mitochondria are specified by

MITOCHONDRIAL D N A

179

nuclear genes but speculation in this matter is not very useful until it is known whether or not the rRNA of mammalian mitochondria is also complementary to mitochondrial DNA, as it is in Tetvabytnetza. Other candidates for the role of mitochondrial gene products include mitochondrial inner membrane proteins, similar to one component of the heterogeneous structural protein fraction, cytochromes aa3, b, cl, and unknown extramitochondrial proteins specified by mRNA exported into the cytoplasm. In our opinion none of these is supported, as yet, by conclusive evidence. All other mitochondrial components have to be specified by nuclear genes and although only the structural gene of iso-1-cytochrome c in yeast has been identified with certainty as a nuclear gene, good indirect evidence indicates that in yeast also the structural genes for succinate dehydrogenase, antimycin-sensitive NADH-cytochrome c reductase, D-lactate and L-lactate cytochrome c reductases, aconitate hydratase, mitochondrial fumarate hydratase, mitochondrial RNA polymerase, and mitochondrial D N A polymerase also belong to this class. Possible ways in which these proteins could find their place in the mitochondria have been considered in several recent reviews (Borst et al., 1967a; Roodyn and Wilkie, 1968; Kadenbach, 1968).

XIV. Addendum Since the completion of this review, the information briefly summarized below has become available. The references cited in this Addendum appear at the end of the reference list. The two bands found in alkaline CsCl for a number of mitochondrial DNA’s have been identified as the complementary strands in the case of mitochondrial DNA from human placenta (Curneo et d,, 1968) and from rat and chick liver (Borst and Ruttenberg, 1969). The complementary strands also differ in density in neutral CsCI, but quantitatively aggregate when present in the same gradient. The lighter strand in alkaline CsCl of chick mitochondrial D N A strongly interacts with both poly U and poly IG; the heavier strand exclusively acts as messenger strand in rat liver (Borst and Aaij, 1969). By mixing experiments, Wolstenholme and Dawid ( 1968) have demonstrated that the mitochondrial D N A circles of two urodele amphibians are 15% smaller than those of two anuran amphibians. The earlier conclusion that most of the D N A in Xeizopus eggs is mitochondrial has been criticized by Baltus et al. (1968), who suggest that yolk DNA, which is not related to mitochondrial DNA, represents the butk of egg DNA. Convincing evidence that circularity is not a property of all mitochondrial DNA’s was presented by Suyama and Miura (1968). They showed that the mitochondrial D N A from Tetrahymena consists of a homogeneous population of linear molecules of 17.6 p ZL- 0.08 (S.E.). From sedimentation studies by

I80

P. BORST A N D A. M. K R O O N

Sonenshein and Holt (1968), the molecular weight of slime mold (Physavum) mitochondrial D N A could be in the same order of magnitude. Heterogeneous linear D N A up to 60 p long was obtained by Wolstenholme and Gross (1968) from mitochondria of the red bean, Pba.reol?Ls udgavis. Various proportions of heterogeneous, open circles, varying in size between 1 and 10 p, have been observed in yeast mitochondrial D N A in a number of laboratories but not in this laboratory (Shapiro et al., 1968; Gukrineau et al., 1968; Avers et al., 1968; Bernardi et al., 1968), and the doubts expressed in our article about the reality of these circles have proved unfounded. In addition Shapiro et al. (1968) have suggested that part of the linear molecules have cohesive ends that can interact to produce hydrogen-bonded circles. How the ordered replication and segregation of such a heterogeneous collection of molecules is effected in the intact yeast cells is not clear. Further studies on mitochondrial D N A oligomers have confirmed the absence of circular dimers in normal animal tissues, while more precise measurements now suggest that, in all normal tissues studied, including those of mouse embryos, 10-16% of the total mitochondrial DNA is present in the form of catenated oligomers (Piko et al., 1968; Clayton et a/., 1968; Hudson and Vinograd, 1969). The evasive replicating circles were finally found by Kirschner et al. (1968) by screening a large number of circular D N A molecules 'from rat liver mitochondria. About 1 out of 600 molecules was a replicating circle. This proves that mitochondria1 D N A replicates within the mitochondrion. Synchronized mitochondrial D N A replication, occurring slightly before nuclear D N A synthesis, was observed in Sacchavomyces by Smith et cll. (1968). A mitochondrial D N A polymerase, differing in properties from the nuclear polymerase, was partly purified from rat liver by Meyer and Simpson (1968), whereas a mitochondrial D N A ligase was detected in rat liver in this laboratory. Karol and Simpson (1968) have reported that the deoxyribonucleotide incorporation observed in isolated rat liver mitochondria is attributable to replicative D N A synthesis rather than repair synthesis. Attardi and Attardi ( 1968) have presented additional experiments, which they interpret as support for their conclusion that some mitochondria1 mRNA is translated on extramitochondria1 ribosomes in HeLa cells. Further evidence that the structural protein of Neuvo.rpora mitochondria is not synthesized by the mitochondrial protein-synthesizing system was presented by Sebald et al. ( 1968, 1969). Yeast continues to yield important information on the biogenesis of mitochondria. Further UV induction studies of the cytoplasmic petite mutation in S. cerevisiae by Matoudas and Wilkie (1968) again suggest that only a single heritable unit is present in anaerobic yeast, whereas in aerobic cells the number of genetically effective copies is greater than one but much less than the number

MITOCHONDRIAL DNA

181

of mitochondria present in these cells. On the other hand, studies by Swift and Wolstenholme ( 1969) and Schatz (personal communication) have shown that a large number of mitochondrial profiles containing mitochondrial D N A are present in anaerobic yeast irrespertii)e of the growth roirditions. The absence of mitochondrial profiles in the micrographs of Wallace and Linnane (1964) appears to result from the use of permanganate staining, which does not stain the mitochondria1 profiles if the cells are grown in media low in ergosterol and fatty acids. Linnane and co-workers (Linnane et al., 1968) have also presented extensive studies on the cytoplasmic inheritance of erythromycin resistance. Contrary to Thomas and Wilkie (1968) they conclude that the cytoplasmic factor for erythromycin resistance and the rho factor may not be identical. The relation between mitochondrial RNA and mitochondria1 and nuclear D N A in yeast was analyzed in detail by Wintersberger and Viehhauser (1968). They showed that the ribosomal RNA components of yeast mitochondria specifically hybridized to a plateau of 0.04 pg RNA per microgram of D N A with mitochondrial D N A from wild-type yeast but not with mitochondrial D N A from a cytoplasmic petite mutant, in which the mitochondrial ribosomal RNA components were also missing. In addition, Wintersberger and Viehhauser report that the mitochondrial ribosomal RNA components specifically hybridize to a value well over 0.01 pg RNA per microgram of D N A with yeast nuclear D N A and they conclude that cistrons for mitochondrial ribosomal RNA’s are represented in the nuclear genome. ACKNOWLEDGMENTS W e are grateful to Professor E. C. Slater for advice and help in the preparation of the manuscript; to Dr. J. M. Tager for putting the proofs of the papers read at the RoundTable Discussion on Biochemical Aspects of Mitochondria at our disposal prior to publication; and to several colleagues for allowing us to reproduce figures or tables from their papers. The experimental work of the authors was supported in part by grants from the Netherlands Foundation for Chemical Research (SON) with financial aid from the Netherlands Organization for the Advancement of Pure Research (ZWO) and The Jane Coffin Childs Memorial Fund for Medical Research.

REFERENCES Allmann, D. W., Lauwers. A., and Lenza, G. (1967). I n “Methods in Enzymology” (R. W. Estabrook and M. E. Pullman, eds.), Vol. 10, p. 433. Academic Press, New York. Attardi, B., and Attardi, G. (1967). Pioc. Natl. Arad. Sci. U.S. 58, 1051. Avers, C. J. (1967). Proc. Natl. Arud. Sri. U.S. 58, 620. Avers, C. J., Pfeffer. C. R., and Rancourt, M. W. (1965). J . Barteriol. 90, 481. Bailey, E., Taylor. C. B., and Bartley, W . (1967). Biorhem. J . 104, 1026. Bartley, W., and Tustanoff, E. R. (1966). Bioc-hem. J. 99, 599. Bass, R., and Neubert, D. (1966). Federation European Biochem. Sor., 3rd Meeting; W ~ ~ s aAbstr., w p. 168. Bauer, W., and Vinograd, J. (1968). J . M o l . Biol. 33, 141.

I82

P . BORST A N D A . M. KROON

Beattie, D. S., Basford. R. E.. and Koritz, S. B. (1966). Biochemistry 5, 926. Becker, A,, Lyn, G., Gefter, M., and Hurwitz, J. (1967). Pror. Natl. A d . Sci. U S . 58, 1996. Bode, V. C. (1967). J . Mot. Biol, 26, 125. Bode, V. C., and McHattie, L. A. (1968). J. Mol. Biol. 32, 673. Borst. P. (1961). In “11th Yearbook for Cancer Research and Fight against Cancer in the Netherlands” (0.Miihlbock, ed.) p. 227. DeBussy, Amsterdam. Borst, P. (1963). I n “Funktionelle und morphologische Organisation der Zelle” (P. Karlson, ed.), p. 137. Springer, Berlin. Borst, P., and Ruttenberg, G . J. C. M. (1966a). Biorhim. Biophys. Acta 114, 6.15. Borst, P., and Ruttenberg, G . J. C. M. (1966b). In “Regulation of Metabolic Processes in Mitochondria” (J. M. Tager, S. Papa, E. Quagliariello, and E. C. Slater, eds.), BBA Library, Vol. 7, p. 454. Elsevier, Amsterdam. Borst, P., Kroon, A. M., and Ruttenberg, G . J. C. M. (1967a). In “Genetic Elements, Properties and Function” ( D . Shugar, ed.), p. 81. Academic Press and Polish Scientific Publishers, London and Warsaw. Borst, P.. Ruttenberg. G. J. C. M., and Kronn, A. M. ( 1 967b). Bior-him. Biojlhys. A ~ t a 149, 110. Borst, P.. Van Bruggen, Ii. F. J., Ruttenberg, G . J. C. M.. and Kroon, A. M . ( 1 9 6 7 ~ ) . Biorhim. Biophys. Arta 149, 156. Borst. P., Van Bruggen, E. F. J., and Ruttenberg, G . J. C. M. (1968). I n “Round-Table Discussion on Biochemical Aspects of the Biogenesis of Mitochondria” (E. C. Slater, J. M. Tager, S. Papa, and E. Quagliariello, eds.), p. 51. Adriatica Editrice, Bari. Breidenbach, R. W., Castelfranco, P., and Criddle. R. S. (1967). Plant Physiol. 42, 1035. Brewer, E. N.. D e Vries, A,, and Rusch, H. P. (1967). Biorhim. Bi0phy.r. Acta 145, 686. Britten, R. J., and Kohne, D. E. (1966). Carnegie Inst. Wash. Yeavbook 65, 78. Britten, R. J., and Waring, M . (1965). Carnegie Inst. Wash. Yearbook 64,316. Bronsert, U., and Neupert, W. (1966). In “Regulation of Metabolic Processes in Mitochondria” (J. M. Tager, S. Papa, E. Quagliariello, and E, C. Slater, eds.), BBA Library, Vol. 7, p. 426. Elsevier, Amsterdam. Burgoyne, L. A,, and Symotis, R. H. (1966). Biorhim. Biophys. Arta 129, 502. Cairns, J. (1963). Cold Spriizg Harbor Symp. Quant. Biol. 28, 43. Cairns, J. (1916). J. M o l . Biol. 15, 372. Carnevali, F.,Piperno, G., and Tecce, G . (1966). Atti Arc-ad. Nazl. Linrei, Rend. Classe S1.i. Fir. Mat. Nut. 41, 194. Chang, L. O., and Looney, W. B. (1966). Intern. I. Radiation B i d . 12, 187. Chun, E. H. L., Vaughan, N . H., Jr.. and Rich, A. (1963). J . Mol. Biol. 7, 130. Church, R.. and McCarthy, B. J. (1967). Pror. Nad. Acad. Sri. U.S. 58, 1548. Clark-Walker, G. D., and Linnane. A. W. (1967). I . Cell B i d . 34, 1. Clayton, D. A., and Vinograd, J. (1967). Nature 216, 652. Conover, T . E., and Biriny, M. (1966). Biochim. Biophys. Arta 127, 235. Cooper, D., Banthorpe, D . V.. and Wilkie, D . (1967). J. Mol. Biol. 26. 347. Corneo, G.,Moore, C., Sanadi, D . R., Grossman, L. J., and Marmur, J. (1966). Srienie 151, 687. Cornea, G., GineIIi, E., and Polli, E , (1967). 1. Mol. Biol. 23, 619. Counts, W . B., and Flamm, W. G. (1966). Biorhim. Biophys. Acta 114, 628. Crawford, L. V., and Waring, M. J. (1967). J. Mol. Biol. 25, 23. Criddle, R. S., Bock, R. M., Green, D . E., and Tisdale, H. D . (1962). Biochemktry 1. 827. Cummins, J. E., Rusch, H. P., and Evans, T. E. (1967). I . Mol. B i d . 23, 281.

MITOCHONDRIAL DNA

183

Cundliffe, li., and McQuillen, K. (1967). J. Mol. Biol. 30, 137. Curtis, P. J., and Smellie, R. M. S. (1966). Biorhem. /. 98, 818. Curtis, P. J., Burdon, M. G., and Smellie, R. M. S. (1966). Biorhem. J. 98, 813. Cuzner, M. L., Davison, A. N., and Gregson, N. A. (1966). Biorhem. J. 101, 618. Dawid, I. B. (1965). J. M o l . B i d . 12, 581. Dawid, I. B. (1966). Pror. Natl. Arad. Sri. U.S. 56, 269. Dawid, I. B., and Wolstenholme, D. R. (1967). J. Mof. Biol. 28, 233. Dawid, I. B., and Wolstenholme, D. R. (1968a). Biophyr. J. 8, 65. Dawid, I. B., and Wolstenholme, D . R. (1968b). I n “Round-Table Discussion on Biochemical Aspects of the Biogenesis of Mitochondria” (E. C. Slater, J. M . Tager, S. Papa, and E. Quagliariello, eds.), p. 83. Adriatica Editrice, Bari. De Deken, R. H. (1966a). J. Gee. Microbiol. 44, 149. De Deken, R. H. (1966b). J. Gen. Microbiol. 44, 157. De Duve, C., and Baudhuin, P. (1966). Physiol. Rev. 46, 323. Du Buy, H. G., and Riley, F. L. (1967). Pror. Natl. Acud. Sri. U.S. 57, 790. Du Buy, H. G., Mattern, C. T. F., and Riley, F. L. (1966). Biorhim. Biophys. Arra 123, 298. lidelman, M., Schiff, J. A., and Epstein, H. T . (1965). J. Mol. Biol. 11, 769. Edelman, M., Epstein, H. T., and Schiff, J. A. (1966). J. Mol. Biol. 17, 463. Evans, T. E. (1966). Biochem. Biophys. Res. Commun. 22, 678. Fessenden, J. M., Dannenberg, M. A., and Racker, E. (1966). Biorhem. Biophys. Res. Commuti. 25, 54. Flamm, W . G., Bond, H . E., Burr, H. l!., and Band, S. B. (1966). Biorhim. Biophys. Arta 123, 652. Fletcher, M. J., and Sanadi, D. R. (1961). Biochim. Biophys. Arta 51, 356. Freeman. K. B., Haldar, D., and Work, T . S. (1967). Biorhrm. J. 105, 947. Freifelder, D., and Kleinschmidt, A. K. (1965). J. Mol. B i d . 14, 271. Fukuhara, H . (1967). Pror. Nutl. Arad. Sci. U S . 58, 1065. Fukuhara, H . (1968). I n “Round-Table Discussion on Biochemical Aspects on the Biogenesis of Mitochondria” (E. C. Slater, J. M. Tager. S. Papa, and E. Quagliariello, eds.), p. 303. Adriatica Editrice, Bari. Geidushek, E. P. (1961). Pvor. Null. A d . Sri. U.S. 47, 950. Geidushek, E. P. (1962). J. Mol. B i d . 4, 467. Gellert, M. (1967). Pror. Natl. Acad. Sci. U.S. 57, 1/18. Gibor, A,, and Granick, S. (1964). Science 145, 890. Gonzilez-Cadavid, N. F., and Campbell, P. N. ( 1967a). Biochem. J . 105, 427. Gonzilez-Cadavid, N. F., and Campbell. P. N. (1967b). Biorhem. J., 105, 443. Granick, S., and Gibor, A. (1967). Prugr. Nnrleic Arid Res. Mol. Biol. 6, 161. Green, B. R., and Gordon, M. P. (1966): Srienre 152, 1071. Green, B. R., and Gordon, M. P. (1967). Bioi-him. Biophys. Acta 145, 378. Green, D. E., and Perdue, J. F. (1966). Pror. Natl. Arad. Sci. U S . 55, 1295. Greenawalt, J. W., Hall, D. O., and Wallis, 0. C. (1967). In “Methods in Enzymology” (R. W. Estabrook and M. E. Pullman. eds.), Vol. 10, p. 142. Academic Press, New York. Gross, N. J., Getz, G . S., and Rabinowitz, M. (1968). In “Round-Table Discussion on Biochemical Aspects on the Biogenesis of Mitochondria” (E. C. Slater, J. M. Tager. S. Papa, and E. Quagliariello, eds.), p. 161. Adriatica Editrice, Bari. Guttes, E., and Guttes. S. (1964). Srienre 145, 1057. Haldar, D., Freeman, K. B., and Work, T . S. (1966). Nature 211, 9.

184

P. BORST A N D A. M . KROON

Heldt, H. W., and Klingenberg, M. (1965). Biorhem. 2. 343, 433. Hradecna, Z., and Szybalski, W. ( 1967). Virology 32, 633. Huang, M., Biggs, D. R., Clark-Walker, G . D., and Linnane, A. W. (1966). Biorhim. Biophys. Acta 114, 434. Huberman, J. A., and Riggs, A. D. (1968). J. Mol. Bol. 32, 327. Hudson, B., and Vinograd, J . (1967). Nature 216, 647. Humm, D. G., and Humm, J. H. (1966). Pror. Nail. Arad. Sri. US. 55, 114. Jansz. H. S., Baas, P. P., Pouwels, P. H., van Bruggen, E. F. J., and Oldenziel, H. (1968). 1. Mul. B i d . 32, 159. Jayaraman, J., Cotman, C . , Mahler, H. R., and Sharp, C. W. (1966). ,411-h. Biochem. Biophys. 116, 224. Jinks, J. L. (1964). “Extrachromosomal Inheritance.” Prentice-Hall, Englewood Cliffs, New Jersey. Kadenbach, B. (1968). I n “Round-Table Discussion on Biochemical Aspects on the Biogenesis of Mitochondria” (E. C. Slater, J. M. Tager, S. Papa, and E. Quagliariello, eds.), p. 415. Adriatica Editrice, Bari. eds.), p. 415. Adriatica Editrice, Bari. Kalf. G. F., and Gri.ce, M. A. (1964). Bjochem. Biophys. Res. Commun. 17. 674. Kalf, G.F., and GrPce, M. A. (1966). J. B i d . Chem. 241, 1019. Katoh, T., and Sanukida, S. (1965). Biochern. Biophys. Res. Commun. 21, 373. Kemp, A.,Jr., and Groot, G . S. P. (1967). Biorhinz. Biophys. Acta 143, 628. Kirk, J. T. 0. (1967). J. Mol. B i d . 28, 171. Kislev, N.,Swift, H., and Bogorad, L. (1965). J, Cell. Bid. 25, 327. Klingenberg, M., and Pfaff, E. (1966). I n “Regulation of Metabolic Processes in Mitochondria” (J. M. Tager, S. Papa, E. Quagliariello, and E. C. Slater, eds.), BBA Library, Vol. 7, p. 180. Elsevier, Amsterdam. Koch, J., and Stokstad, E. L. R. (1967). European J. Bimhrm. 3, 1. Kroon, A. M. (1965). Biochirn. Biophys. Acta 108, 275. Kroon, A. M. (1966a). “Protein Synthesis in Mitochondria,” M.D. Thesis, Llniv. of Amsterdam. Drukkerij Noordholland, Hoorn. Kroon, A. M. ( l 9 6 6 b ) . I n “Regulation of Metabolic Processes in Mitochondria” ( J . M. Tager, S. Papa, E. Quagliariello, and E. C. Slater, eds.), BBA Library, Vol. 7, p. 397. Elsevier, Amsterdam. Kroon, A. M., and Aaij, C. (1968). In “Round-Table Discussion on Biochemical Aspects on the Biogenesis of Mitochondria” (E. C. Slater, J. M. Tager, S. Papa, and €1. Quagliariello, eds.) , p. 207. Adriatica Editrice, Bari. Kroon, A.M., Borst, P., van Bruggen, E. F. J., and Ruttenberg, G . J. C. M . (1966). pror.. Natl. Arad. S r i . US.56, 1836. Kubinski, H.,Opara-Kubinska, Z., and Szybalski, W. (1966). J. Mol. Biol. 20, 313. Lark, K.G. (1966). Barteriol. Rev. 30, 3. Larsson, A,, and Reichard, P. (1967). Progr. Nurl. Acid Res. 7, 3 0 3 . Lehninger, A. L. (1964). “The Mitochondrion.” Benjamin, New York. Lerman, L. S. (1961). J. Mol. B i d . 3, 18. Lerman, L. S. (1964). J. Cellular Comp. Physiol. 64, Suppl. 1, p. 1. Linnane, A. W. (1965). In “Oxidases and Related Redox Systems’’ (T. E. King, H . S. Mason, and M. Morrison, eds.). P. 11, p. 1102. Wiley, New York. Linnane, A. W. (1968). I n “Round-Table Discussion on Biochemical Aspects on the Biogenesis of hritochondria” ( E . C. Slater, J. M. Tager, S. Papa, and E. Quagliariello, eds.), p. 333 Axitralian J. Sci., 107, 165. Adriatica Editrice, Bari.

MITOCHONDRML DNA

185

Linnane, A. W., and Still, J. L. (1956). Australian J. Sci. 19, 165. Little, J. W., Zimmerman, S. B., Oshinsky, C . K., and Gellert, M. (1967). Pror. Natl. Arud. Sri. U.S. 58, 2004. Luck, D . J. L. (1965). Am. Nutuvalist 99, 241. Luck, D . J. L. (1966). In “Probleme der biologischen Reduplikation” (P. Sitte, ed.), p. 314. Springer, Berlin. Luck, D . J. L., and Reich, E . (196-1). Pvoc. Null. Arud. Sri. U S . 52, 931. Lukins, H . B., Tham, S. H., Wallace, P. G., and Linnane, A. W. (1966). Bioc-hem. Biophys. Res. Commun. 23, 363. Mackler. B., Douglas, H . C . . Will, S., Hawthorne, D. C . , and Mahler, H . R. (1965). Biochemistry 4, 2016. Mahler, H. R., Mackler, B., Grandchamp. S., and Slonimski, P. P. (1964). Biochemistry 3, 668. Marmur, J., and Cordes, S. (1963). In “Informational Macromolecules” ( H . J. Vogel, V. Bryson, and J. 0. Lampen, eds.), p. 79. Academic Press, New York. Marmur, J., and Doty. P. (1962). J. Mol. Biol. 5, 109. Marmur, J., Rownd, R., and Schildkraut, C . L. (1963). Prop. Nurleir Arid Res. 1, 231. Marver, H. S., Collins, A., Tschudy, D. P., and Rechcigl, M., Jr. (1966). J . B i d . Chenr. 241, 4323. Mattoon, J. R., and Sherman, F. (1966). J. Biol. Chem. 241, 1330. Merker, H . J., Herbst, R., and Kloss, K. (1967). Z . Zellfo~.r~-h. Mikroskop. Anai. 86, 139. Meselson, M., and Stahl, F. W . (1958). Pror. Nail. A d . Sci. U S . 44. 671. Moore, P. B., Traut, R. R., Noller, H., Pearson, P., and Delius, H . (1966). J. Mol. B i d . 31, 4-11. Mounolou, J. C., Jakob, H., and Slonimski, P. P. (1966). Biorhem. Biophys. Res. Commun. 24, 218. Mounolou, J. C., Jakob, H., and Slonimski, P. P. (1968). I n “Round-Table Discussion on Biochemical Aspects on the Biogenesis of Mitochondria” ( E . C . Slater. J. M. Tager, S. Papa, and E. Quagliariello, eds.), p. 473. Adriatica Editrice, Bari. Moustacchi, E., and Williamson, D . H. (1966). Bioc-hem. Biophys. Res. Commun. 23, 56. Munkres, K. D., and Woodward, D. 0. (1966). Pror. Natl. At.ad. 51-i.U S . 55, 1217. Nass, M. M. K. (1966). Pror. Natl. Arad. Sri. U S . 56, 1215. Nass, M. M. K. (1967). In “Organizational Biosynthesis” (H. J. Vogel. J. 0. Lampen, and V. Bryson, eds.), p. 503. Academic Press, New York. Nass. M. M. K. (1968). In “Round-Table Discussion on Biochemical Aspects on the Biogenesis of Mitochondria” ( E . C. Slater, J. M. Tager. S. Papa, and E. Quagliariello, eds.). Adriatica Editrice, Bari. In press. Nass, M. M. K., Nass. S., and Afzelius. €3. A. (1965). E x p / l . Cell Res. 37, 516. Nass, S. (1967). Biochim. Biophys. Aria 145, 60. Nass, S.. Nass, M. M. K., and Hennix, U. (1965). Biochim. Biophyv. Artu 95, ,126. Negrotti, T., and Wilkie, D. (1968). Biorhirn. Biophys. Arta 153, 341. Neubert, D., Oberdisse, E., and Bass, R. (1968a). I n “Round-Table Discussion on Biochemical Aspects on the Biogenesis of Mitochondria” ( E . C . Slater, J. M. Tager, S. Papa, and E. Quagliariello, eds.), p. 103. Adriatica Editrice, Bari. Neubert, D., Helge, H.. and Merker, H . J. (1968b). I n “Round-Table Discussion on Biochemical Aspects on the Biogenesis of Mitochondria” (E. C. Slater, J. M. Tager, S.Papa, and E. Quagliariello, eds.), p. 251. Adriatica Editrice, Bari.

I86

P. RORST A N D A. M. KROON

Neupert, W., Brdiczka, D., and Biicher, T . (1967). Biocheni. Biophys. Res. Commurr. 27, 488. Neupert, W., Brdiczka, D., and Sebald, W . (1968). In “Round-Table Discussion on Biochemical Aspects on the Biogenesis of Mitochondria” (E. C. Slater, J. M. Tager, S. Papa, and E. Quagliariello, eds.), p. 395. Adriatica Editrice, Bari. Niyogi, S. K., and Thomas, C. A,, Jr. (1967). Biochem. Biophys. Res. Commun. 26, 51. Ogur, M., Hickler, S., Lindegren, G., and Lindegren, C. C. (1952). Arch. Biochem. Biophys. 40, 175. Ohnishi, T., KrBger, R. A,, Heldt, H . W., Pfaff, E., and Klingenberg, M. (1967). Europeatz J . Biorhem. 1, 301. Parsons, D. F. (1966). Proc. Can. Catzcer. Re.r. Cotif., 7, 193. Parsons, J. A. (1965). J. Cell Biol. 25, 641. Parsons, J. A,, and Dickson, R. C. (1965). J . Cell B i d . 27, 77A. Parsons, P., and Simpson, M. V. (1967). Science 155, 91. Parsons, P., and Simpson, M. V. (1968). In “Round-Table Discussion on Biochemical Aspects on the Biogenesis of Mitochondria” (E. C. Slater, J. M. Tager, S. Papa, and E. Quagliariello, eds.), p. 171. Adriatica Editrice, Bari. Piko, L., Tyler, A,, and Vinograd, J. (1967). Biol. Bull. 132, 68. Polakis, E. S., and Bartley, W. (1965). Biochem. J. 97, 284. Polakis, E. S., Bartley, W., and Meek, G . A . (1964). Biochem. 1. 90, 369. Polakis, E. S.. Bartley, W., and Meek, G. A. (1965). Biochem. J. 97, 298. Pouwels, P. H., Jansz, H . S., van Rotterdam, J., and Cohen, J. A. (1966). Biochim. Biophys. Ar/a 119, 289. Pouwels, P. H., Knijnenburg, C. M., van Rotterdam, J., Cohen, J. A,, and Jansz, H. S. (1968). J . Mol. Biol. 32, 169. Pullman, M. E.,and Schatz, G. (1967). Anti. Re{*.Biorhem. 36, 539. Purvis, J. L., and Lowenstein, J. M. (1961). J. B i d . Chem. 236, 2794. Rabinowitz, M., Getz, G. S., and Swift, H . (1968). lrr “Round-Table Discussion on Biochemical Aspects on the Biogenesis of Mitochondria” (E. C. Slater, J. M. Tager, S. Papa, and E. Quagliariello, eds.), p. 155. Adriatica Editrice, Bari. Rabinowitz, M., Sinclair, J., De Salle, L., Haselkorn, R., and Swift, H. H . (1965). PYOC-. Natl. Acad. Sci. US.53, 1126. Radloff, R., Bauer, W., and Vinograd, J. (1967). Proc. Natl. Acad. Sci. U.S. 57, 1514. Ray, D . S., and Hanawalt, P. C. (1964). J. Mol. Biol. 9, 812. Ray, D. S., and Hanawalt, P. C. (1965). J. Mol. Biol. 11, 760. Reich, E., and Luck, D. J. L. (1966). Pror. Natl. Arad. Sci. U.S. 55, 1600. Reilly, C., and Sherman, F. (1965). Biochim. Biophys. Aria 95, 640. Rhaese, H. J., and Freese, E. (1968). Biochim. Biophys. Acta 155, 476. Roodyn, D. B., and Wilkie, D . (1968). “The Biogenesis of Mitochondria.” Methuen, London. Roodyn, D. B., Suttie, J. W., and Work, T. S. (1962). Biochem. J. 83, 29. Roth, T. F., and Helinski, D . R. (1967). Pvoc. Natl. Acad. Sci. U S . 58, 650. Rush, M. G., and Warner, R. C. (1967). Pror. Natl. A d . Sri. U S . 58, 2372. Rush, M . G., Kleinschmidt, A . K., Hellmann, W . , and Warner, R. C. (1967). Proc. Natl. Arad. Sri. U.S. 58, 1676. Ruttenberg, G. J. C. M., and Borst, P. (1968). Federation European Biochem. Sor. j t h Meeting, Prague Abstr.. 329. Ruttenberg, G . J. C. M., Smit, E. M., Borst, P., and van Bruggen, E . F. J. (1968). Bjochim. Biophys. Acta 157, 429.

MITOCHONDRIAL DNA

187

Saccone, C., Gadaleta, M. N., and Gimigliano, A. F. (1968). In “Round-Table Discussion on Biochemical Aspects on the Biogenesis of Mitochondria” (E. C. Slater, J. M. Tager, S. Papa, and E. Quagliariello, eds.), p. 265. Adriatica Editrice, Bari. Sager, R., and Ramanis, 2. (1965). Proc. Nail. Acad. Sci. U.S. 53, 1053. Schatz, G. (1965). Biochim. Biophys. Aria 96, 342. Schatz, G., Tuppy, H., and Klima, J. (1963). Z . Naturforsch. 18b. 145. Schatz, G., Haslbrunner, E., and Tuppy, H. (1964a). Monatsh. Chem. 95, 1135. Schatz, G., Haslbrunner, E., and Tuppy, H. (1964b). Biorhem. Biuphys. Re.1. Commun. 15, 127. Schildkraut, C. L., Marmur, J., and Doty, P. (1962). f . Mol. B i d . 4, 430. Schildkraut, C. L., Richardson, C. C., and Kornberg, A. (1964). J. Mol. B i d . 9. 21. Schrnieder, M., and Neubert, D. (1966). Arch. Pharmakol. Expil. Pathol. 255, 68. Schneider, W. C., and Kuff, E. L. (1965). Pror. Natl. Ac.ad. Sci. U S . 54, 1650. Sherman, F., Stewart, J. W., Margoliash, E.,Parker, J., and Campbell, W. (1966). Pror. Nail. Acad. Sci. U.S. 55, 1498. Shipp, W. S., Kieras, K. J., and Haselkorn, R. (1965). Pror. Natl. Arad. Sci. U.S. 54, 207. Sinclair, J. H., (1966). Ph.D. Thesis, Univ. of Chicago, Chicago, Illinois. Sinclair, J. H., and Stevens, B. J. (1966). Pror. Nail. Arad. Sci. U.S. 56, 508. Sinclair, J. H., Stevens, B. J., Sanghavi, P., and Rabinowitz, M. (1967a). Science 156, 1234. Sinclair, J. H., Stevens, B. J., Gross, N., and Rabinowitz, M . (1967b). Biorhim. Biophys. Aria 145, 528. Slonimski, P. P. (1956). Proc. 3rd Iiitertz. Corigr. Biochem., B r u s e h , 1911i p. 242. Slonimski, P. P., Perodin, G., and Croft, J. H . (1968). Biochem. Biophys. Res. Commun. 30, 232. Somlo, M., and Fukuhara, H. ( 1965). Biochem. Biophyr. R e f . Commun. 19, 587. Stanley, W. M., Jr., and Bock, R. M. (1965). Biochemistry 4, 1302. Studier, F. W. (1965). J. Mol. Biol. 11, 371. Sueoka, N. (1961). J. M o l . Biol. 3, 31. Suyama, Y.(1966). Biochemistry 5, 2214. Suyama, Y . (1967). Biochemistry 6, 2829. Suyama, Y., and Bonner, W. D., Jr. (1966). Plant Physiol. 41, 383. Suyarna, Y., and Eyer, J. (1968). J. B i d . Chem. 243, 320. Suyama, Y., and Preer, J. R., Jr. (1965). Genetirs 52, 1051. Swartz. M., Trautner, T., and Kornberg, A. (1962). J. B i d . Chem. 237, 1961. Swift, H. (1965). A m . Nururalist 99, 201. Swift. H., Rabinowitz, M., and Getz, G. (1968a). In “Round-Table Discussion on Biochemical Aspects on the Biogenesis of Mitochondria” (E. C . Slater, J. M. Tager, S. Papa, and E. Quagliariello, eds.), p. 3. Adriatica Editrice, Bari. Swift, H., Sinclair, J. H., Stevens, B. J., Rabinowitz, M., and Gross, N . (1968b). I n “Round-Table Discussion on Biochemical Aspects on the Biogenesis of Mitochondria” (E. C. Slater, J. M. Tager, S. Papa, and E. Quagliariello, eds.), p. 71. Adriatica Editrice, Bari. Slater, E. C., Tager, J. M., Papa, S., and Quagliariello, E. (1968). “Round-Table Discussion on Biochemical Aspects of the Biogenesis of Mitochondria.” Adriatica Editrice, Bari. Tamm, C.. Shapiro, H. S., Lipshitz, R., and Chargaff, E. (1953). J. Biol. Chrm. 203, 673. Taylor, J. H. (1968). J. Mol. B i d . 31, 579.

188

P. BORST A N D A. M . KROON

Ter Schegget, J., and Borst, P. (1968). Unpublished observations. Tewari, K. K., and Wildman, S. G. (1966). Science, 153. 1269. Tewari, K. K., Votsch, W., Mahler, H. R., and Mackler, B. (1966). J. Mol. Biol. 20, 453. Thomas, C. A., Jr. (1966). J. Gee. Physiol. 49, 143. Thomas, D.Y . , and Wilkie, D. (1968a). Biorhem. Biophys. Res. Commun. 30, 368. Thomas, D. Y . , and Wilkie, D. (1968b). Genet. Res. 11, 444. Traut, R. R., Moore, P. B., Delius, H., Noller, H., and Tissisres, A. (1967). Pror-. Natl. Acad. Sri. U.S. 57, 1294. Truman, D . E. S. (1964). Biochem. J. 91, 59. Tuppy, H., and Swetly, P. (1968). Biochim. Biophys. Acta 153, 293. Tuppy, H., and Wintersberger, E. (1966). In “Probleme der biologischen Reduplikation” (P. Sitte, ed.), p. 325. Springer, Berlin. Van Bruggen, E. F. J., Borst, P., Ruttenberg, G. J. C. M., Gruber, M., and Kroon, A. M. (1966). Biochim. Biophys. Aria 119,437. Van Bruggen, E. F. J., Runner, C. M., Borst, P., Ruttenberg, G. J. C. M., Kroon, A. M., and Schuurmans Stekhoven, F. M. A. H. (1968). Biochim. Biophys. Arta 161, 402. Vinograd, J., and Hearst, J. E. (1962). “Progress in the Chemistry of Organic Natural Products,” p. 372. Springer, Berlin. Vinograd, J.. and Lebowitz, J. (1966). J. Gen. Physiol. 49, 103. Vinograd, J., Lebowitz, J., Radloff, R., Watson, R.. and Laipis, P. (1965). Pror. Natl. A d . Sri. U.S. 53, 1104. Vinograd, J., Lebowitz, J., and Watson, R. (1968). J. Mol. Biol. 33, 173. Wallace, P.G., and Linnane, A . W . (1964). Nature 201, 1191. Wang, J. C . (1969). J. Mol. Biol. In press. Wang, J. C . , Baumgarten, D., and Olivera, B. M. (1967). Pror. Natl. Acad. Sci. U.S. 58, 1852. Waring. M. J. (1965). J. Mol. Biol. 13, 269. Weil, R., and Vinograd, J. (1963). Proc. Natl. At-ud. Sc-i. U.S. 50, 730. Wells, R., and Birnstiel, M. L. (1967). Biochem. J. 105, 53P. Wells, R. D., and Blair, J. E. (1967). J. M o l . Biol. 27, 273. Wmner, C. E. (1967). Advaz. Enzymoi. 29, 321. Wetmur, J. G . . and Davidson, N. (1968). J. Mol. Biol. 31, 349, Wheeldon, L. W . , and Lehninger, A. L. (1966). Biochemistry 5, 3 5 3 3 . Whitfeld, P. R., and Spencer, D. (1968). Biorhim. Biophys. Acta 157, 333. Wilkie, D . (1963). J. M o l . Biol. 7, 527. Wilkie, D. ( 1964). “The Cytoplasm in Heredity,” Methuen Monograph Biol. Subjects. Wiley, New York. Wilkie, D. (1968). Z ~ J “Round-Table Discussion on Biochemical Aspects of the Biogenesis of Mitochondria” ( E . C. Slater, J. M. Tager, $. Papa, and E. Quagliariello, eds.), p. 457. Adriatica Editrice, Bari. Wilkie, D., Saunders, G. W., and Linnane, A. W . (1967). Gene/. Res. 10, 199. Wintersberger, E. (1965). Biochem. Z . 341, 409. Wintersberger, E. (1966). Biochem. Biophys. Res. Comrnuri. 25, 1. Wintersberger, E. (1968). I n “Round-Table Discussion on Biochemical Aspects of the Biogenesis of Mitochondria” (E. C . Slater, J. M. Tager, S. Papa, and E. Quagliariello, eds.), p. 189. Adriatica Editrice, Bari. Witt, J., Kronau, R., and Holzer, H. (1966). Bioshirn. Biophys. Acta 128,63. Wolstenholme, D.R., and Dawid, I . B. (1967). Chromosoma 20, 445.

MITOCHONDRIAL DNA

189

Woodward, D . O., and Munkres, K. D. (1966). Proc-. Natl. A[-ad.Sci. U S . 55, 872. Woodward, D . O., and Munkres, K. D. ( 1967) Ztz “Organizational Biosynthesis” ( H . J. Vogel, J. 0. Lampen, and V. Bryson. eds.), p. 489. Academic Press, New York. Work, T. S. (1967). Biochem. J. 105, 38P. Work, T. S. (1968). In “Round-Table Discussion on Biochemical Aspects of the Biogenesis of Mitochondria” ( E . C. Slater, J. M. Tager. S. Papa, and E. Quagliariello, eds.), p. 367. Adriatica Editrice, Bari. Ycas, M. (1956). Exptl. Cell Res. 11, 1. Yoshikawa, H . (1967). Pror-. Nutl. A L - u Sr-i. ~ . U S . 58, 312. Yotsuyanagi, Y . (1962a). 1. Ullrarlrur-z. Re.r. 7, 121. U C 7, ~ .141. Yotsuyanagi, Y. (1962b). 1. U ~ U J ~ VRes. Yu, R., Lukins, H . B., and Linnane, A . W . (1968) i r z “Round-Table Discussion on Biochemical Aspects of the Biogenesis of Mitochondria” ( E . C. Slater, J. M. Taper, S. Papa, and E. Quagliariello, eds.). p. 359. Adriatica Editrice, Bari.

SUPPLEMENTARY REFERENCES Attardi, B., and Attardi, G. (1968). Pror.. Natl. Ar-ad. Si-i. U.S. 61. 261. Avers, C. J. (1968). Pvor-. Natl. Acad. S r i . U.S. 61, 90. Baltus, E., Hanocq-Quertier, J., and Brachet, J. (1968). Pror. Null. Acad. Sci. U.S. 61, 469. Bernardi, G., Carnevali, F., NicolaieR. A., Piperno, G., and Tecce, G. (1968). J , Mol. Biol. 37, 493. Borst, P., and Aaji. C. (1969). Biochem. Biophys. Res. Corunrun. 34, 358. Borst, P., and Ruttenberg, G. J. C. M. (1969). Federatiotz European Biochrnt. Sor.. 6th. Meeting, Madrid, Abstr. In press. Clayton, D. A. ( 1968). Nature 220, 976. Corneo, G. (1968). 1.Mol. Biol. 36, 419. GuPrineau, M., Grandchamp, C.. Yotsuyanaji, Y., and Slonimski, P. P. (1968). Comp/. Rend. 266, 1884, 2000. Hudson, B., and Vinograd, J. (1969). Nature 221, 3 3 2 . Karol, M. H., and Simpson, M. V. (1968). Science 162, 470. Kirschner, R. H.. Wolstenholme, D. R., and Gross, N.J. (1968). Proc. Nad. Acad. Sri. U.S. 60, 1466. Linnane, A. W. (1968). Prur-. Nntl. Arad. Sci. U S . 59, 903. 1288. Maroudas, N . G., and Wilkie, D. (1968). Biorhirn. Biophys. Acid 166, 681. Meyer, R. R.. and Simpson, M. V. (1968). Pror. Na/I. A d . Sci. U S . 61, 130. Piko. L. (1968). Pro[-. Natl. Acad. Sci. U.S. 59, 838. Sebald, W., Biicher, T., Olbrich, B., and Kaudewitz, F. (1968). FEBS Letters 1, 235. Sebald, W., Hofstotter, T., Hacker, D., and Biicher, T. (1969). FEBS Letterr 2, 177. Shapiro. L., Grossman, L. I., Marmur, J., and Kleinschmidt, A. K. (1968). J . Mol. B i d . 33, 907. Smith, D., Tauro, D.. Schweizer, E., and Halvorson, H . 0. (1968). Proc. Natl. h a d . Sci. U.S. 60, 936. Sonenshein, G. E., and Holt, C. E. (1968). Biorhem. Biophys. Res. Commun. 33, 361. Suyama, Y., and Miura, K. (1968). Proc. Natl. Acad. Sci. U S . 60, 2 3 5 . Swift, H., and Wolstenholme, D. R. (1969). “Handbook of Molecular Cytology,” ( A . Lima-de-Faria,ed.) North-Holland Publ., Amsterdam. In press.

190

P . BORST AND A. M. KROON

Thomas, D. Y., and WiIkie, D. (1968). Biorhem. l3jophy.r. Res. Cornnzun. 30, 368. Wallace, P. G., and Linnane, A. W . (1964). Nature 201, 1191. Wintersberger, E., and Viehhauser, G. L. (1968). Nature 220, 699. Wolstenholme, D. R., and Dawid, I. B. (1968). J. Cell B i d . 39, 222. Wolstenholme, D. R., and Gross, N. J. (1968). Proc. Nut/. Acud. Sci. U.S. 61, 245.

Metabolism of Enucleated Cells KONRADKECK Department of Biological Srienrer, University of Avizona. T U C ~ OArizona ?I, I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. Initiation of the Anucleate State . . . . . . . . . . . . . . . . . . . . . . A. Physical Enucleation B. Inhibition of RNA Syn 111. Quantitation of mRNA . . A . Introductory Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Direct Methods . . . . . . C. Relationship between mRNA and Polysomes . . . . . . . D. Relationship between mRNA and Protein Synthesis . . . . IV. Decay of mRNA and Protein Synthesis in Anucleate Cells A. Procaryotic Organisms B. Eucaryotic Organisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . V. Nature of mRNA Decay . . . . . . . . . . . . . . . . . . . . . . . . . . . . A . Enzymic Degradation . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Initiation of Decay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

191 192 102 193 I Oh 106

196 198 201 208 208 212 222 222 221 225 225

I. Introduction Studies on the metabolic activities of enucleated cells are often directed toward an understanding of the interactions between the cell nucleus and the cytoplasm. The subject has been discussed from this viewpoint in several review articles (Hammerling et ul., 1959; Hammerling, 1963; Prescott, 1960a; Brachet, 1961). The present article deviates from this conceptual approach and focuses instead on the anucleate state per se. The term “anucleate” is defined here in the most general sense o f the word and will be applied to all cellular systems in which the flow of genetic information from nuclear genes to the cytoplasm has been interrupted. The anucleate condition can therefore be initiated by physical enucleation and by naturally occurring nuclear degeneration, as well as by chemical inhibition of nuclear RNA synthesis. The liberal interpretation of the term “enucleation” avoids a restriction of the text to a few cells, often atypical, that are amenable to microsurgical enucleation, and in addition justifies the inclusion of procaryotic organisms which offer a wide spectrum of pertinent information on this subject. Investigation of the anucleate state, as defined above, comprises in essence the characterization of metabolic changes occurring in cells that gradually exhaust their supply of genetic information of nuclear origin. Depending on the lifetime of individual messenger species and on the stability of vital proteins persisting 191

192

KONRAD KECK

after the complete decay of their respective mRNA, a breakdown of the cellular organization sooner or later occurs. This simplified model ignores several additional parameters, viz., the replenishment of cytoplasmic ribosomes and tRNA by the nucleus, the contribution by the nucleus of products other than RNA, and the existence of independent genetic elements in the cytoplasm. Because of the very low turnover rate of rRNA and the main structure of tRNA (Section II,B,l), these RNA components only become limiting in anucleate cells that contain exceedingly long-lived mRNA. Further, little is known about the direct contribution of DNA-containing cytoplasmic organelles to the total protein synthesis in the cytoplasm. The chloroplast system has been eliminated from consideration in this article by restricting the discussion mainly to heterotrophic organisms, and only a few proteins seem to be encoded in mitochondria1 D N A (Roodyn et d.,1762; Woodward and Munkres, 1766; Kadenbach, 1967). In spite of its somewhat artificial nature, an anucleate cellular system can provide valuable information concerning posttranscriptional regulation of metabolism in general, and of protein synthesis in particular. The rate of synthesis of a given protein in the intact cell depends on the amounts of available mRNA, which in turn are governed by the rate of synthesis, controlled at the gene level, and the rate of decay. The latter varies greatly among individual mRNA species for reasons that are not presently understood. The anucleate system lends itself ideally to a study of mRNA decay and the relationship between mRNA levels and the rate of protein synthesis. The system is equally suitable for the investigation of translational control mechanisms of a protein-specific, or nonspecific nature, without the added complexity of superimposed transcriptional regulation. Some facets of the anucleate metabolism bear also on cellular differentiation. Certain metazoan cells become naturally enucleated during their terminal stage of differentiation. This anucleate phase then represents the ultimate level of the acquisition of a specialized function. In a wider sense, every differentiated cell may be considered anucleate with respect to a number of repressed genes. The time interval between the beginning of gene repression and its subsequent phenotypic expression depends upon the mean life of the respective mRNA pools. On the other hand, gene activation during development may remain without immediate metabolic consequences because of translational repression, or because of the storage of the messengers in "masked" form.

11. Initiation of the Anucleate State ENUCLEATION A. PHYSICAL Microsurgical enucleation is the most reliable method for removal of the nuclear genome from the cell. In addition, it permits the precise timing of the

METABOLISM OF ENUCLEATED CELLS

193

interruption of the mRNA flow into the cytoplasm. In suitable cells enucleation can be combined with the reimplantation of another nucleus containing the same or a different genome. With the help of interspecific nuclear transplantations the experimenter is in a position to change only a small part of the genome and to study the anucleate condition for individual genes without, at least in theory, interfering with the basic functions of the cell. If allelic differences exist in the molecular structure of a given protein, the depletion of the original protein-specific messenger pool, and the buildup of the homologous new pool can be followed directly in an otherwise undisturbed cell. This experimental approach was first tried with the unicellular alga Acetabulavia (Keck, 1960, 1961; Clauss, 1962 ; Schweiger et al., 1967). Unfortunately, not many cell types are suitable for routine microsurgicaI enucleation. With small cells insurmountable difficulties arise when relatively large quantities of enucleated cells are needed for biochemical analysis. In rare instances the large-scale production of enucleated cells, or cell fragmnts, can be achieved by collective treatment of cell populations. A well-known example is the enucleation of sea urchin eggs by centrifugation (Harvey, 1956). Enucleation is not restricted to eucaryotic cells; anucleate fragments termed “mini cells” have also been obtained from an abnormally budding strain of Esrheiichia r o l i (Adler et nl., 1967). The enucleation operation practically always evokes side effects, the consequences of which are difficult to assess. There is a more-or-less pronounced traumatic reaction, perhaps connected with a temporary disturbance of the cell‘s permeability, and an unavoidable loss of a certain portion of the cytoplasm. Further indirect effects might result from the preferential localization of organelles or metabolic products in the removed portion of the cytoplasm. Effects of this kind would be more pronounced in highly polar cells. B. INHIBITIONO F RNA SYNTHESIS 1. InhibitorJ

The antibiotic actinomycin D ( A D ) inhibits the DNA-dependent RNA synthesis specifically and at relatively very low concentrations by binding to the guanine bases in the minor groove of double helical D N A in the B-configuration, thereby blocking RNA polymerase activity (Kirk, 1960; Reich et af., 1961; Hurwitt et al., 1962; Goldberg eta]., 1962; Kahan et d., 1963; Hamilton et al., 1963; Reich, 1964). At appropriate concentrations the inhibition of the de nozw synthesis of all RNA species is complete, but the turnover of the terminal CpCpA- group of tRNA continues at such concentrations (Merits, 1962; Tamaoki and Mueller, 1962; Eason et al., 1963; Franklin, 1963). The inhibition of DNA-catalyzed RNA synthesis in eucaryotic cells should

194

KONRAD KECK

mimic the effect of enucleation on the metabolic activity of the cytoplasm. This makes AD a valuable tool for the study of anucleate metabolism in all cell types that are not suitable for direct enucleation including, in a wider sense of the word, enucleation of procaryotic organisms. Since the DNA-specific chemical and biological action of AD is well understood and has been recently reviewed in detail (Reich and Goldberg, 1964), the following section is restricted to a discussion of nonspecific effects of the drug as well as to the differences between true enucleation and the consequences of AD inhibition of RNA synthesis. It is often difficult to distinguish between indirect effects of AD, i.e., effects on the cell metabolism that ultimately can be traced back to a block in the RNA synthesis, and the nonspecific effects, which are the consequence of reactions of AD with components other than DNA. Nonspecificity of action of AD is sometimes inferred from the observation that a particular effect is manifested in the cells long before the overall rate of protein synthesis is significantly diminished as a result of mRNA decay. However, there is still the possibility that a few mRNA species have a much shorterthan-average life and soon fail to support the synthesis of vital proteins. Nonetheless, some of the observed effects cannot easily be explained in this way. For instance, the inhibition of respiration and anaerobic glycolysis in human leukemic leukocytes is evoked by AD but not by puromycin, a potent inhibitor of protein synthesis (Laszlo et ul., 1966), and in sarcoma ascites cells the inhibition of protein synthesis by AD can either be prevented or, once it is established, reversed by the addition of glucose to the medium (Honig and Rabinowitz, 1965). Nonspecific toxicity of AD is also indicated when concentrations over and above those needed for the complete suppression of RNA synthesis produce additional biochemical lesions in the cells, such as, for instance, an accelerated decline of the rate of protein synthesis (Soeiro and Amos, 1966). Another example of interference of AD with protein synthesis, apparently unrelated to messenger decay, was found in rat heart in which the rate of in vivo protein synthesis declined more rapidly than the level of polysomes. A defect in the ribosomes was suggested since ribosomes isolated from AD-treated tissue responded much less to stimulation with polyuridylic acid (poly-U) than control ribosomes (Earl and Korner, 1966). In contrast to nonspecific effects, some of the indirect effects of AD can very well also be expected to be expressed in physically enucleated cells. This is true, for instance, of unstable gene-controlled repressors which operate at the translational level (Section III,DJ), or the possible stabilization of mRNA because of a higher frequency of ribosome attachment during mRNA depletion (Trakatellis et ul., 1965b). The observed acceleration of RNA breakdown in B u r i l h szlbtilis in the presence of AD was originally interpreted as a nonspecific effect (Acs et ul., 1963). Subsequently, similar effects in eucatyotic cells and in other

METABOLISM O F ENUCLEATED CELLS

195

bacteria have been interpreted differently, viz., that interference of AD with the completion of partially synthesized RNA molecules renders them sensitive to nuclease attack (Girard e f al., 1964; Zimmerman and Levinthal, 1967). If the latter assumption is correct, accelerated RNA decay must be classified as an indirect effect of AD, but nevertheless one that would not be found after microsurgical enucleation of cells since unfinished RNA would be confined to the cell nucleus. In eucaryotic cells the possibility exists that transitory mRNA in the nucleus can still enter the cytoplasm after the complete inhibition of RNA synthesis by AD thereby postponing the moment of effective enucleation. Although only limited information is available on this subject, it seems that AD interfers in an unknown way with the transport of mRNA across the nuclear membrane (Girard et al., 1964, 1965). Proflavin is another compound that binds to D N A (DeMars et al., 1953) and thereby inhibits the enzymic synthesis of RNA in uivo as well as in uitro (Hurwitz et ul., 1962). Concentrations of proflavin that efficiently inhibit the DNA-primed RNA synthesis in an iiz oitro system from bacteria have no inhibitory effect on the poly-U-directed phenylalanine incorporation in the same system and therefore do not seem to interfere with protein synthesis per se (Woese et a/., 1963). Both proflavin and dinitrophenol have been used to determine the rate of mRNA decay in E. coli after the in uivo inhibition of RNA synthesis (Woese et ul., 1963). The usefulness of dinitrophenol for this purpose was recently questioned by Friesen (1966), who cited experimental evidence for a nonspecific effect of this compound resulting in enhanced breakdown of mRNA and stable RNA. 2. Defective mRNA

As an alternative to inhibiting mRNA synthesis, the cell can be made to produce defective mRNA by administering certain purine or pyrimidine analogs. The analogs are incorporated into RNA in place of the corresponding natural base and thereby affect the functional properties of RNA. As an example, fluorouracil is incorporated into bacterial RNA (Horowitz and Chargaff, 1959) and causes the formation of inactive enzyme proteins, presumably as a result of translational errors arising from the presence of the abnormal base in mRNA (Naono and Gros, 1960; Gros et al., 1961a; Gros and Naono, 1961; Nakada and Magasanik, 1964). The above interpretation of the effects of fluorouracil on protein synthesis is, however, open for reappraisal, since it was recently demonstrated by Horowitz and Kohlmeier ( 1967) that the fluorouracil-initiated synthesis of inactive 0-galactosidase in E . coli occurred only in the presence of readily catabolizable-substrates, e.g., glycerol, while active enzyme was synthesized during fluorouracil treatment in the absence of catabolic repression.

196

KONRAD KECK

3. Starvatzon f o r R N A P i e r u n o r

The availability of cell strains that are auxotrophic for nucleic acid precursors offers the possibility of inhibiting RNA synthesis by starving the cells of the required precursor. There is, however, a significant time lag between the removal of the required precursor and the cessation of R N A synthesis. This lag is in part attributable to the time necessary to deplete the intracellular pool of the precursor and in part to recycling of the precursor from mRNA breakdown products into newly synthesized RNA. The recycling period terminates when most of the available precursor has become incorporated into stable RNA; its length depends on the cellular levels of mRNA and on the rate of stable R N A synthesis. The latter, of course, is a function of the growth rate of the cells. The lag period can introduce significant differences between the actual decay time of mRNA and the observed decay time.

111. Quantitation of mRNA A. INTRODUCTORY REMARKS At the present time there is no method available that permits the routine quantitative analysis of individual, gene-specific mRNA species. It has, however, been possible with rather unique systems to isolate R N A fractions that which contain only one, or very few, mRNA species, such as the messenger for the polypeptide gramicidin S (Hall et al., 1965), the messengers for hemoglobin (Marbaix et ul., 1966; Chantrenne et ul., 1967), and messengers of the luc operon of E. coli (Hayashi et al., 1963). Consequently, most of our knowledge concerning mRNA metabolism was gained from investigations on heterogeneous populations of molecules which may include hundreds of mRNA species differing in molecular weight, base composition, and in functional life. Adding to this the fact that mRNA comprises but a small proportion of the total cellular RNA, a few percent at best, one can easily appreciate the experimental difficulties inherent in this type of research. It is not surprising, therefore, to find significant dissimilarities among experimental data obtained from the same biological system by techniques based on different properties and functions o f mRNA. It seems proper, therefore, to include here a brief survey of techniques.

B. DIRECTMETHODS 1.

Unstable R N A

Messenger RNA is generally characterized by a high rate of turnover in contradistinction to ribosom'il RNA (rRNA), which is relatively stable in both procaryotic (Davern and Meselson, 1960; Meselson et ul., 1964) and eucaryotic organisms (Rake and Graham, 1962; Loeb et ul., 1965; Hadjiolov, 1966), and

METABOLISM O F ENUCLEATED CELLS

197

transfer RNA (tRNA) whose turnover is confined to the terminal pCpCpA group (Franklin, 1963; Merits, 1962; Tamaoki and Mueller, 1962; Eason et al., 1963). Therefore, exposure of the cells to radioactive RNA precursors for periods of time that are short compared to the mean life of mRNA moIecules permits the preferential labeling of mRNA. Provided with a radioactive tag, the mRNA fraction can then be isolated and further characterized by base ratio analysis (Volkin and Astrachan, 1956), sucrose density gradient sedimentation (Nomura et aL, 1960), or column chromatography (Ellem and Sheridan, 1964; Yoshikawa et d., 1964; Yoshikawa-Fukada et al., 1965; Ellem, 1966). The decay rate of mRNA can be determined from the time-dependent loss of acidinsoluble label after the inhibition of RNA synthesis. In exponentially growing cells the relative cellular amounts of mRNA can be computed from changes in the distribution of label between stable and unstable RNA (Levinthal et al., 1962), as well as from the labeling kinetics of the precursor pool (Salser et al., 1968), as outlined in Section IV,A,I. The common occurrence of long-lived mRNA in eucaryotic cells renders the above-mentioned procedures unreliable, and assay techniques must be used that are based on other properties of mRNA. 2.

Stimilating Activity

This method takes advantage of the functional properties of mRNA, viz., its capacity to stimulate the incorporation of labeled amino acids into acid-insoluble material in a complete in vitro system for protein synthesis (Nirenberg and Matthaei, 1961; Tissicres and Hopkins, 1961). Purified rRNA has a very low stimulating activity, or “template activity,” while mRNA is very active (Barondes et al., 1962; Brawerman et al., 1963; Hoagland and Asconas, 1963; DiGirolamo rt d.,1964). The stimulating capacity of a given nucleic acid species depends most likely on the absence of a secondary structure. Thus, while double-stranded virus RNA is inactive, the same RNA after heat denaturation, or the native single-stranded RNA of tobacco mosaic virus (TMV), are both active (Miura and Muto, 1966). Ribosomal RNA and tRNA have a very low stimulating activity in their native state, but elicit greatly increased activities after heat destruction of their secondary structure (Holland et d., 1966), and even denatured D N A displays template activity in the in vitro system (McCarthy and Holland, 1965). The fact that rRNA is methylated but mRNA apparently is not (Moore, 1966) does not seem to explain the differences in their template activities since methyl-deficient rRNA isolated from s o called “relaxed” particles of methionine-starved E . coli has the same low stimulating activity as fully methylated rRNA isolated from “relaxed” particles of arginine- or histidine-starved cells (Manor and Haselkorn, 1967; Sypherd, 1967). The molecular weight of the test RNA does not appear to be too critical as long as it remains above a certain value. The stimulating activity of TMV RNA, which is often used as a

198

KONRAD KECK

“standard” for this assay, does not change when its molecular weight is reduced from the normal value to ca. 400,000 by thermal degradation. Further reduction of the molecular weight from 300,000 to approximately 75,000 results in a lower stimulating activity. The latter value apparently reflects the minimum chain length required for significant template activity in vitro (Boedtker and Stumpp, 1964). Therefore, even fragments of a messenger molecule could function independently in this test. This idea might explain the observation that the amounts of template RNA seem to increase slightly after the irt v h administration of AD (Kennell, 1964). The specificity of the iiz vztro stimulation of protein synthesis is still questionable in some cases. For instance, a nonspecific stimulation of protein synthesis, controlled by endogeneous messenger, is suggested for a cell-free E . coli system to which reticulocyte RNA has been added; the synthesized proteins were found to resemble bacterial proteins more closely than hemoglobin (Drach and Lingrel, 1966). 3 . Moleczllar Hybridizcltiotz

The mRNA content of a given radioactively labeled RNA preparation can be estimated by molecular hybridization with homologous D N A (Hall and Spiegelman, 1961). Differentiation between mRNA on the one hand, and stable RNA species on the other hand, is based upon the finding that commonly only a very small portion of the DNA, less than I%, code for rRNA (Yankofsky and Spiegelman, 1962, 1963) and tRNA (Giacomoni and Spiegelman, 1962; Goodman and Rich, 1962). In some cells somewhat higher multiplicities of the genetic loci for rRNA seem to exist (Matsuda and Siegel, 1967). In any event, the hybridization of labeled rRNA or tRNA can be further suppressed by the addition of an excess of the respective homologous unlabeled RNA species. The RNA-DNA hybrids, formed in solution, can be separated from uncomplexed RNA by various procedures such as CsCl equilibrium centrifugation (Hall and Spiegelman, 1961; Hayashi and Spiegelman, 1961), MAK-column chromatography (Hayashi et ul., 1965), or fiItration through nitrocellulose filters (Nygaard and Hall, 1963, 1964). The denatured D N A can also be immobilized for the hybrid formation on agar (Bolton and McCarthy, 1962) or membrane filters (Gillespie and Spiegelman, 1965). The background level of nonspecifically bound RNA can be greatly reduced by treatment of the complexes with ribonuclease (Gillespie and Spiegelman, 1965). C. RELATIONSHIP BETWEEN mRNA

AND

POLYSOMES

Cytoplasmic mRNA is associated with ribosomes forming functional aggregates of various sizes, termed polyribosomes or polysomes (Korner and Munro, 1963; Penman et a/., 1963; Staeheh et a/., 1963b; Wettstein et a/., 1963). An

METABOLISM O F ENUCLEATED CE1.I.S

199

RNA fraction differing from rRNA has been isolated from polysomes and found to have properties that are characteristic of mRNA (Penman et d., 1963; Munro and Korner, 1964; Munro et al., 1964; Burny and Marbaix, 1965). As expected on theoretical grounds, larger polysomes generally contain heavy messenger, and small aggregates light messenger (Staehelin et al., 1.964; Trakatellis et al., 1964). However, the distribution is not always clear-cut, and, for example, in rapidly growing HeLa cells up to 50% of the heavy messenger was recovered from light polysomes indicating that some of the heavy mRNA strands carry fewer than their maximum number of ribosomes (Latham and Darnell, 1965). According to the currently accepted model for the translation process, the “tape mechanism” (Gierer, 1963; Gilbert, 1963; Warner et al., 1963; Watson, 1963), the ribosomes, or their subunits, attach to the 5’-terminus of the messenger strand and, in proceeding to the 3’-terminus (Salas et al., 1965; Thach et al., 1965; Terzaghi et al., 1966), translate the genetic message into the proper amino acid sequence beginning with the N-terminal end of the polypeptide chain (Bishop et al., 1960; Dintzis, 1961). At the 3’-end of the mRNA molecule, both the finished polypeptide chain and the associated ribosome are released. Polysomes thus represent the operational unit linking the growing polypeptide chain with the mRNA molecule. Consequently, polysome structure and function affords two approaches to mRNA quantitation. One approach concerns the quantitative relationship between the cellular levels and the “size” spectrum of polysomes on the one hand, and the rate of total protein synthesis on the other hand. Understanding this relationship would permit us to extrapolate our findings with bulk mRNA to individual protein-specific messengers. This possibility is discussed in Section II1,D. The second aspect relates to the quantitation of cytoplasmic mRNA in functional form. Assuming that each polysomal aggregate carries only one cistronspecific messenger strand (polycistronic messengers are not considered here), then the number of mRNA molecules in each class of polysomes would be proportional to the total number of ribosomes in this class divided by the number of ribosomes per aggregate. Estimation of the amounts of cytoplasmic messenger by this method are only meaningful if the following conditions are met: (1) The yield of polysomes is high and reproducible; ( 2 ) there is no extensive degradation or aggregation of the polysomes; (3) the class of monomers in a preparation can definitely be assigned a role as either free ribosomes or ribosomes bound to one mRNA strand; (4) no substantial amount of cytoplasmic mRNA exists in free or “masked” form. Although condition (1) may raise problems with certain materials, the yield of free or membrane-bound polysomes can be estimated (cf. Blobel and Potter, 1967a,b), and techniques then improved until this condition is met. Condition (2) is more difficult to achieve. Endonucleolytic attacks on the messenger strand,

200

KONRAD KECK

the “backbone” of a polysome, are very likely to occur during the isolation and fractionation procedure. Large polysomes are thereby cleaved into two or more smaller ones resulting in an overestimation of the relative amount of mRNA. With certain material endonuclease activity is difficult to control; fortunately though, extensive cleavage of aggregates can be detected after iiz vizw saturation labeling of the nascent proteins. As predictable from the “tape mechanism” of translation, the average polypeptide label per ribosome increases with increasing length of the messenger strand if we assume approximately equal spacing of the ribosomes along the strand. Consequently, the specific activity of undegraded polysomes (label in nascent protein per unit RNA) increases in a very characteristic manner from small to large aggregates (Noll et d., 1963; Kuff and Roberts, 1967). After extensive degradation all the polysome classes have approximately the same specific activity because of random distribution of long and short unfinished polypeptide chains among the polysome fragments (Warner et d l . , 1963). There is good experimental evidence that in certain cell types polysomes associate and form aggregates of higher order. Such polysome clusters contain more than one mRNA strand and therefore the messenger content of the entire polysome population would be underestimated. Polysome clusters have been observed in certain specialized cells synthesizing mainly collagen (Kretsinger et d., 1964). Treatment with the enzyme collagenase, but not with ribonucleas, was found to cleave the clusters to units of smaller size indicating that originally several polysomes were held together by bonds extending between nascent peptide chains (Goldberg and Green, 1967). In another example cells from stimulated lymph nodes were found to contain a distinct class of polysomes, the so-called “immune peak” which are resistant to mild ribonuclease treatment. The administration of puromycin, which should cause the release of nascent peptide chains, does not bring about the breakdown of these polysomes, indicating that they are probably not bound together solely by the nascent peptides (Manner et al., 1965). Resistance to ribonuclease was also reported for heart muscle polysomes which sediment as large aggregates and are thought to be held together by nascent protein (Rabinowitz et al., 1964). Serious objections could be raised against the estimation of the cytoplasmic mRNA levels from the polysome sedimentation profiles on the basis of conditions (3) and ( 4 ) . Single ribosomes attached to natural messengers can carry out protein synthesis in vitro (Munro et al., 1964; Dreyfus and Schapira, 1966) and are capable of releasing complete proteins, a criterion for the translation of intact messenger (Lamfrom and Knopf, 1964, 1965). Whether or not the attachment of the ribosomes to mRNA strands already existed iiz vivn or occurred during or after the fractionation is difficult to ascertain. Single ribosomes, attached to short fragments of mRNA, often result as artifacts from

METABOLISM OF ENUCLEATED CELLS

201

endonucleolytic or mechanical breakdown of polysomes. Such fragments elicit amino acid-incorporating activity it1 v h o , but the "run-off ' synthesis does not lead to the release of polypeptides (No11 et af., 1963; Staehelin et d., 1963b; Zimmerman, 1963). When labeled amino acids are administered i j z viva, however, the isolated monomers do not carry labeled peptides, suggesting that monomer ribosomes are not involved in protein synthesis in zivo (Noll et ul., 1963; Penman et al., 1963; Zimmerman, 1963). This does not rule out the possibility, of course, that monomers are nevertheless attached, even if in inactive form, to mRNA in Z ~ V O .Indeed, it has been reported that mRNA is conserved, most likely in association with single ribosomes, when polysomes dissociate under certain physiological conditions. Upon the recovery of cells from such a condition, polysomes re-form in the absence of de novo RNA synthesis. Reversible polysome dissociation can be induced in Chang liver cells by the omission of glutamine from the culture medium (Eliasson et a/.,1967), and in rat liver by the feeding of tryptophan-deficient diet (Fleck ef al., 1965). Interestingly, amino acid starvation of reticulocytes is without this effect (Burka and Marks, 1964). Inhibitors that interfere with the energy metabolism of the cell, such as fluoride, dinitrophenol, cyanide, or iodoacetate, produce reversible polysonie dissociation, just as anaerobiosis (Marks et nl., 1965; Coconi et nl., 1966; Lin et al., 1966). The loss of polysomes in rat liver after the administration of ethionine was originally thought to be caused by the inhibition of m R N A synthesis resulting from the lowering of the cellular ATP level (Villa-Trevino et al., 1964). Reinvestigation of the effect has clearly demonstrated the conservation of mRNA after the disappearance o f the polysomes (Stewart and Farber, 1967). The conserved messenger strand seems to remain associated with single ribosomes after the breakdown of the polysomes. Evidence was presented that monomer ribosomes, isolated from sodium fluoride-treated reticulocytes, still cont;iined the information for hemoglobin synthesis (Lin et al., 1966). Furthermore, 9-S RNA, presumably the messenger for hemogIobin, could be recovered after polysonie dissociation from the 80-S pellet (Lebleu et al., 1967). Physiological conditions that permit partial or complete dissociation of polysomes, even with ensuing conservation of mRNA, would certainly invalidate an estimation of the mRNA content from the cellular level of polysomes. Whether or not :t11 mRNA remains attached to single ribosomes is of little concern here since obviously not all of the ribosomes could carry one messenger strand.

D. RELATIONSHIP BETWEEN mRNA AND PROTEIN SYNTHESIS 1. Theoretical Cotiside~atiotzs The estimation of the cytoplasmic messenger levels by any of the abovediscussed methods is restricted to bulk messenger or, at best, tu large hetero-

202

KONRAD KECK

geneous classes of messengers. However, a theoretical model can be devised that permits the indirect determination of the relative cellular levels, as well as the half-lives, of individual messenger species from the kinetics of synthesis of the respective proteins. W e assume that at any given time after enucleation the rate of synthesis of a given protein is related to the cellular amount of its specific messenger by the following expression :

where dP/dt is the rate of protein synthesis, M is the amount of mRNA, and k,, is the rate constant of protein synthesis. Assuming, furthermore, that the decay of mRNA follows first-order kinetics, then

where k,, is the decay constant for mRNA. Thus, the amount of mRNA remaining after time M

=Moe-bMt

/

is

(3)

where rM, is the initial cellular amount of mRNA. Substitution of Eq. ( 3 ) into E l . ( 1 ) yields

The integrated form of this equation, with the condition that P = 0 when t = 0, is

Considering Eq. ( 5 ) as t approaches infinity and rearranging, we arrive at the following approximation :

Thus the cellular amount of a given mRNA species at the time of enucleation is proportional to the total amount of the corresponding protein synthesized after enucleation. Since the rate of protein synthesis is proportional to the cellular level of mRNA, the half-life of this mRNA equals the time period during which the rate of protein synthesis decreases to one-half its original value. In practice the half-life can simply be obtained from a semilogarithmic

METABOLISM O F E N U C L E A T E D CELLS

203

plot of the experimentally obtained rates of protein synthesis against time after enucleation. This model can easily be expanded to accommodate a special hypthetical case in which an inactive pool of messenger is drawn upon to replenish the decaying functional messenger, keeping the active fraction at a constant steady-state level until all of the inactive form is consumed. The residual amount of the active fraction would then again decay exponentially. O n the basis of Eq. ( 2 ) we can write -

A M = kl,briM"At

(7)

where iM, represents the initial steady-state level of the active fraction during the "holding period" At, and AM the amount of mRNA consumed during this period. Likewise, according to Eq. ( I ) the amount of protein AP synthesized during the holding period At is AP = kspiMOAt (8) where ksl, again is the rate constant of protein synthesis. Solving for M,, and substituting into Eq. ( 7 ) we obtain h*t Ap -AM = __

(9)

kSP

Adding the amount o f mRNA degraded during the holding period [Eq.(9) 1 to the amount lost by the ensuing exponential decay 1Eq. ( 6 ) ] we obtain the total amount of messenger M,l.,,t: MTot

=

4 II hl ~

kSP

(AP

+ P)

(10)

where AP is the protein synthesized during the holding period, and P the amount of protein synthesized during the exponential phase of the messenger decay. For practical purposes a time period covering a few half-lives can be infinite time. considered sufficient to approximate _The model presented is based on certain restrictive conditions. Thus, it has been assumed that during the period of anucleate protein synthesis mRNA is the only limiting factor and that the concentrations of the required enzymes, tRNA, amino acids, energy donors, and other components of the machinery for protein synthesis do not change sufficiently to affect protein synthesis. Aside from general defects, however, there exist a number of experimental data that indicate that the postulated relationship between the amount of messenger and the rate of protein synthesis, specified by the factor k S P in Eq. (1) may not remain constant over the entire anucleate period. In particular, three possible causes for a deviation from the ideal system should be considered here: (1) The

2 04

KONRAD KECK

existence of cellular control mechanisms that operate at the translational level; ( 2 ) the effect of the increasing cellular pool of monomer ribosomes on the frequency of polypeptide chain initiation; and ( 3 ) the possible occurrence of inactive ribosomes in polysomes. In all three cases the degree of interference, if any, is likely to change during the phase of messenger depletion. Another condition tacitly included in Eq. (1) is the stability of the proteins. Although the turnover rate of many proteins is negligible compared to that of their RNA template, there are examples in which the average life of a protein molecule is shorter than that of its messenger, requiring the addition of a term for the protein decay in Eq. ( 1 ) . 2.

Trnizslational Controls

The existence of translational control mechanisms of both protein-specific or nonspecific nature, has been postulated for a number of systems. Protein nonspecific regulation of protein synthesis is inferred from experimentally inducible changes in the functional efficiency of polysome preparations in rdtro. Altered polysome efficiency was observed either in response to certain physiological stresses imposed on cells or animals iiz viva, or in response to the addition of particular fractions of cell homogenates to an in oitro system for protein synthesis. Liver polysomes isolated from rats that had been fed a protein-free diet were found to sustain much lower levels of amino acid incorporation in vitvo than polysomes from protein-fed animals. Differences in the messenger content or in the composition of the supernatant fraction were discounted as responsible factors because the experimental and the control preparations contained the same ratio of polysomes to single ribosomes, and both supernatant fractions proved equally active (Von der Decken, 1967). Starvation of mouse ascites tumor cells produces similar effects. The i7z vitvo activity of ribosomal preparations from starved cells is repressed, while after brief recovery of the cells in supplemented medium polysomes can be obtained that support much higher levels of protein synthesis in vitro (Kerr et nl., 1966). The recovery phenomenon is not prevented by AD treatment and therefore probably does not depend on de ) m ’ o RNA synthesis. Moreover, the release and subsequent reattachment of pre-existing mRNA to ribosomes can be discounted as the responsible factors since isolated RNA from active as well as from inactive ribosomal preparations has the same stimulating activity in a reticulocyte iiz vitru system and thus apparently contains the sitme amounts of mRNA (Kerr et d.,1966). Subcellular fractions of unknown composition have been described that stimulate protein synthesis in homologous cell-free systems, both in rat liver (Mizrahi, 1965j and in reticulocytes (Beard and Armentrout, 1967j . These preparations, referred to as “fraction X,” were obtained from postribosomal supernatants.

METABOLISM OF ENUCLEATED CELLS

205

The identity of the responsible factor with messengerlike RNA, tRNA, or activating enzymes was ruled out. The reticulocyte factor is probably a protein and might be involved in the initiation of polypeptide chains (Beard and Armentrout, 1967), perhaps in connection with the formation of N-formylated amino acids (Clark and Marcker, 1966). Similar conclusions were also reached for the rat liver factor on the basis of the observed factor-induced acceleration in the rate at which 8 0 3 ribosomes become associated with polysomes (Mizrahi, 1965). Membrane-associated ribosomes in normal and regenerating rat liver were found to contain different amounts of a heat-labile factor of unknown chemical nature which inhibits amino acid incorporation in vitro. The factor can he released from the membranes by sonication and does not appear to react by destruction of mRNA, by interference with mRNA binding to ribosomes, or by the release of nascent protein from ribosomes (Hoagland et al., 1964). Protein-specific regulation at the translational level is thought to involve unstable repressor molecules, the synthesis of which is initiated at the transcriptional level. In rat liver such a repressor seems to be responsible for the decline of tryptophan pyrrolase and tyrosine-a-ketoglutarate transaminase synthesis which normally follows the hormone-stimulated elevation of these enzymes (Garren el d., 1964). In the rat specific repression of hepatic tyrosine transaminase synthesis, but not of total liver protein synthesis, can also be invoked by administration of stressing agents (Kenney and Albritton, 1965). In both cases enzyme repression can be blocked by treatment o f the cells with AD, suggesting the transcriptional initiation of the repressor synthesis. Both the enzymes seem to be synthesized on stable RNA templates since AD has no short-term effect on the basal rate of synthesis of the enzymes. The latter observation excludes the possibility that the cellular site of repression is at the gene level. A system with very similar properties was described by Eliasson (1967a,b). The translation of Chang liver cell arginase is thought to be controlled by a metabolic repressor although the messenger for this enzyme apparently is very stable. As with rat liver, the arginase repression can be inhibited with AD (Eliasson, 1967a,b). -3.

Freynenry of Chaiir Izitia~ioiz

Under steady-state conditions o f protein synthesis the number of polypeptide chains released per polysome and unit time equals the number of new chains initiated. The rate of chain initiation, in turn, should equal the frequency of ribosome attachment to the messenger strand provided each ribosome is active in protein synthesis. For a given translational speed there is perhaps an upper limit to the number of ribosomes that can attach per unit time, imposed by the requirement for a minimum distance between adjacent ribosomes. During the anucleate phase the breakdown of polysomes eventually leads to an increase in the proportion of free ribosomes. According to the law of mass action

206

KONRAD KECK

one might expect that an elevated pool of ribosomes results in a higher attachment rate of ribosomes thereby enhancing the efficiency of the remaining polysome population (cf. Williamson and Schweet, 1964). Experimental evidence seems to support this prediction in some but not all of the systems investigated. Trakatellis et al. (1965a,b) observed in two different cell types (reticulocytes and mammary carcinoma cells) that during the progressive depletion of mRNA the in vivo rate of protein synthesis declined slower than the level of polysomes. Similar nonparallel changes, but in reversed order, were observed during the recovery of reticulocytes from the sodium fluoride-induced dissociation of polysomes. The rate of protein synthesis reached control values long before the normal level of polysomes was restored (Coconi et al.] 1966; Marks et al., 1965 ) . The presence of inactive polysomes was ruled out because the specific activity of the polysomes (label in nascent protein per ribosome) did not change during the recovery from fluoride poisoning. However, alternative explanations of the phenomenon can be given (Coconi et al., 1966). In rat liver, on the other hand, the polysome level and the rate of protein synthesis iiz vivo, corrected for the specific activity of the cellular amino acid pool, both declined with similar rates after the administration of AD (Wilson and Hoagland, 1967). Since presumably only polysomes containing less than the saturating number of ribosomes could respond to an elevated free ribosome pool, the observed inconsistencies could simply reflect differences in the degree of polysome saturation at the time of enucleation. The ribosome content of polysomes, as indicated by their “size” spectrum, appears to depend on physiological conditions of the cells. Under normal conditions adult rat liver polysomes seem to carry a nearmaximum number of ribosomes (Staehelin et ale, 1964). Prolonged fasting of rats causes a shift of the polysome size distribution to smaller values (Webb et nl., 1966), while force-feeding of a threonine-deficient diet enhances the in vivo protein synthesis in rat liver and results in heavier polysomes (Sidransky et al., 1964). Polysomes in exponentially growing cells, as in HeLa cells for instance, carry less than the maximum number of ribosomes (Latham and Darnell, 1965). The absence of a shift in the size spectrum of polysomes, on the other hand, does not necessarily indicate the constancy of the rate of ribosome attachment, since a concomitant change in the translational speed could offset the change in polysome size.

4. Imctive RibosomeJ Polysome efficiency would be reduced if ribosome attachment to the mRNA strand were not always coupled with the initiation of a polypeptide chain. Sever31 observations support the idea that polysomes may contain a varying number of ribosomes that are not engaged in protein synthesis. Polysomes in maturing reticulocytes gradually lose their in uiuo efficiency for protein syn-

METABOLISM OF ENUCLEATED CELLS

207

thesis, most likely because of increasing proportions of inactive ribosomes. This is borne out by the finding that polysomes in aging reticulocytes carry diminishing amounts of nascent protein (Marks et al., 1963a; Glowacki and Millette, 1965) and that poly-U stimulation of such ribosomal preparations resulted in lower rates of phenylalanine incorporation (Rowley and Morris, 1967). Polysome preparations isolated from yeast that had been sampled at different points in their growth phase, differed in their endogenous and poly-U-stimulated capacity to incorporate amino acids into protein. The defect was localized in the ribosomes and apparently was not caused by differences in the mRNA content of the preparations (Dietz and Simpson, 1964). It appears that inactive ribosomes can be preferentially detached from rat liver polysomes by lowering of the magnesium ion concentration (Munro et al., 1964), but whether or not this treatment can be used for the estimation of the proportions of inactive ribosomes in other preparations remains to be seen. Further work is needed before it is meaningful to speculate on the general occurrence of inactive ribosomes in polysome populations, or on the change of their proportions during the anucleate phase. 5 . Proteiii Degradation

The instability of some proteins comes close to, or even exceeds, that of their respective template RNA (Lin and Knox, 1958; Feigelson et al., 1959; Kenney, 1967; Peterkofsky and Tomkins, 1967). If we assume that protein degradation proceeds with first-order kinetics and, furthermore, that the decay factor of proteins remains unchanged throughout the entire anucleate phase, we can correct for protein degradation by introducing another term to Eq. (1 ) ClP dt

-- - k,,.M - kDpP

where kDP represents the rate constant of protein decay. In many cases such a correction cannot be applied. The relative decay rates of some proteins were found to vary in response to i9t viva changes in the substrate levels (Schimke et al., 1965) and in response to dietary stimuli (Schimke, 1964). Even genetic factors seem to exist that control protein degradation, expressed in the form of time- and tissue-specific patterns. For instance, the enzyme UDP-galactose-polysaccharide transferase is degraded during a specific stage of development of the slime mold Dictyortelitlm (Sussman, 1965). The process seems genetically programmed and initiated via a system that can be inhibited by AD, and perhaps involves the synthesis of proteolytic enzymes (Sussman and Sussman, 1965), Evidence for the genetic control of the decay of mouse liver catalase was presented by Rechcigl and Heston (1967).

208

KONRAD KECK

IV. Decay of mRNA and Protein Synthesis in Anucleate Cells ORGANISMS A. PROCARYOTIC 1. Uiisiuble R N A

Bacteria contain an unstable R N A fraction which is preferentially labeled after brief incubation of the cells with radioactive RNA precursors. This rapidly labeled RNA has been characterized by base ratio analysis and sedimentation, as well as by molecular hybridization with homologous DNA, and is thought to consist mainly of mRNA (Astrachan and Fisher, 1961; Gros et al., 1961a,b; Hayashi and Spiegelman, 1961). The turnover rate of this RNA was originally determined by pulse labeling and subsequent “chasing out” of the label with an excess of cold precursor, hut more recent data rtn its breakdown kinetics have been obtained by following the loss of acid-insoluble label after the inhibition of further R N A synthesis by AD (Levinthal et al., 1962). Even after very brief labeling periods the decay of rapidly labeled RNA in B. rubtilk begins without delay after AD administration and follows closely a first-order kinetics with a mean decay time (half-life/ln 2 ) of approximately 2 minutes. (Levinthal et ul., 1962). The data strongly imply that the decay of mRNA molecules is a random process rather than the result of a functional aging of the molecules. The ADinitiated loss of acid-insoluble R N A label is paralleled by the disappearance of a labeled R N A component with a mean sedimentation constant of approximately 15 S and, in addition, by a decrease in the rate of valine incorporation into proteins, the latter with a decay time of ca. 3-4 minutes (Levinthal et al., 1962). When R N A synthesis in E. coli cells, which are normally not sensitive to AD, was inhibited by dinitrophenol or proflavin a similar close relationship between the decay of rapidly labeled RNA and the declining rate of amino acid incorporation into protein was observed (Woese et al., 1963). Since the possibility existed that the rate of mRNA degradation in the absence of de ~ O V OR N A synthesis might not reflect the true messenger life, perhaps because of indirect or nonspecific effects of the inhibitors employed, efforts were made to determine the rate of mRNA turnover under steady-state conditions of RNA synthesis. Cells of E. coli were pulsed with radioactive azaguanine, and the subsequent recycling of label from degraded into newly synthesized RNA was prevented by the addition of an excess of cold guanine. The half-life of mRNA as obtained by this method was not significantly different from that obtained after AD inhibition of RNA synthesis (Chantrenne, 1965). The decay time of labile RNA under steady-state conditions of mRNA turnover can also be calculated from the time delay in the labeling of the cellular guanosine triphosphate (GTP) pool after the administration of radioactive guanine. The degradation of initially unlabeled mRNA continues to supply cold guanine to the cellular pool thereby increasing the time necessary to “wash” all of the

hlETABOLISM OF ENUCLEATED (.ELLS

2 09

cold precursor into stable rRNA. Although this method weighs more heavily the longer-lived mRNA species, an upper limit of 3 minutes was calculated f o r the decay time of labile RNA of B. snbtilis at 37OC., and ca. 4 minutes for E. coli at 30OC. (Salser et al., 1968). The existence of mRNA with lifetimes significantly longer than 1-2 minutes has been recognized after careful analysis of the RNA decay kinetics in bacteria. Leive (1965b) reports that in E. coli the degradation of pulse-labeled RNA can be graphically represented in a semilogarithmic plot by two intersecting lines with different slopes. The more rapidly decaying component had a half-life of 11/2 minutes, while the other component with a slower decay rate had a half-life of 16 minutes or more. When the synthesis of R N A in a uracil auxotrophic strain of E. coli is inhibited by uracil starvation, one finds that the in vitrostimulating activity of the isolated RNA, presuinably reflecting its mRNA content, decays in a biphasic mode; half-lives of 5 minutes and 42 minutes at 25°C. were calculated after correction of the measured values for recycling uracil (Forchhammer and Kjeldgaard, 1967). In Bacillus megnterium the presence of mRNA fractions with half-lives of 4 minutes and 10 minutes was concluded I incorporation into proteins from the declining rates of the i n Z ~ Z Y phenylalanine after treatment of the cells with AD (Yudkin, 1965). In addition, there is good experimental evidence for the existence of bacterial messengers with extremely long lives; these will be discussed later. The relative proportion of unstable RNA in bacteria can be derived from the time-dependent distribution of radioactive precursors between stable and unstable RNA after saturation of the mRNA pool with label (Levinthal et al., 1962). The cellular amounts of unstable R N A can also be computed from the kinetics of the GTP pool IabeIing as outlined above (Salser et al., 1968). Reported values are 1.5-376 for E. coli (Leive, 1965b; Mangiarotti and Schlessinger, 1967; Salser et al., 1968) and 7.6-9.0% for B. .rubtilis (Levinthal et al., 1962; Zimmerman and Levinthal, 1967; Salser et ul., 1968). 2.

Polysome Levels

A high percentage of bacterial mRNA seems to be associated with ribosomes (Mangiarotti and Schlessinger, 1967), and one should therefore expect that the cellular level of polysomes fairly accurately reflects the relative amounts of -cytoplasmic messenger and, furthermore, that the breakdown of mRNA is closely paralleled by a corresponding disappearance of polysomes. Investigation of polysome-associated mRNA revealed, however, that unexpected differences existed between the mean life of unstahle RNA and that of polysonial mRNA. In E. coli the chemical mean life of mRNA in polysomes, presumably identical with its functional lifetime, was estimated to extend over a period of 11-12 minutes (Mangiarotti and Schlessinger, 1967). These values were computed

210

KONRAD KECK

from the labeling kinetics of polysomal mRNA under steady-state conditions; they are significantly larger than those obtained from the rate of degradation of unstable RNA. Similar results were obtained with B. meguterizm. The decay of pulse-labeled RNA and the loss of hybridizable RNA both proceeded with a half-life of less than 1 minute at 37OC. after AD administration, while polysoma1 mRNA has a half-life of 3-4 minutes (Schaechter et ul., 1965). One of the possible explanations for the discrepancies of the values for mRNA half-life rests on the assumption that a significant portion of the decaying RNA consists of incompletely synthesized molecules. The unfinished RNA molecules might have been arrested in an unprotected state by the action of AD and thus left vulnerable to ribonuclease attack (Schaechter and McQuillen, 1966; Zimmerman and Levinthal, 1967). After pulse-labeling periods which are short compared to the estimated transcription time of 1-2 minutes for the average messenger molecule (Alpers and Tomkins, 1965; Goldstein et al., 1965; Leive, 1965a), a significant portion of the total incorporated label would be contained in unfinished RNA molecules and therefore preferentially degraded. This interpretation might be valid for some, but not all, of the rapidly decaying RNA, since studies with inducible enzymes have definitely shown that the functional lifetimes of at least some of the bacterial messengers are in the order of 1-2 minutes. 3. lndztcible Enzymes

Because of the uniqueness of microbial systems it is possible to initiate, or stop within a few seconds, the transcription of individual operons without thereby significantly interfering with the overall metabolism of the cell (Jacob and Monod, 1961). Enzyme induction is immediately followed by the buildup of a protein-specific mRNA pool prior to the appearance of active enzyme (Pardee and Prestidge, 1961). The presence of a specific mRNA pool is indicated by the acquisition of an “enzyme-forming capacity,” which can be quantitatively defined as the total amount of enzyme protein eventually synthesized after the removal of the inducer (Kepes, 1963; Hartwell and Magasanik, 1963, 1964; Kepes and Beguin, 1966). Under the conditions outlined in Section 111, D,1, [ Eq. (6) 1, this amount of protein is proportional to the cellular level of messenger at the time of deinduction. In support of this assumption it was observed that the rise in enzyme-forming capacity after induction closely follows a 100(1 - e - - k t ) kinetics which is characteristic for a compound synthesized at a constant rate (ignoring the exponential growth of a bacterial culture) and decays with first-order kinetics. It is therefore tempting to equate the enzymeforming capacity with the relative cellular levels of mRNA at the time of deinduction, but for the following reasons it might be more appropriate to also

METABOLISM O F ENUCLEATED CELLS

211

include in the mRNA pool the already initiated but still unfinished messenger molecules. In bacteria, ribosomes apparently attach to the messenger strand while it IS still in association with its D N A template at the growing point (Schaechter and McQuillen, 1966; Bremer and Konrad, 1964; Byrne et d., 1964; Alpers and Tomkins, 1965; Naono et al., 1966; Das et al., 1967; Revel and Gros, 1967). Ribosome attachment to unfinished mRNA also seems to take place in the case of messengers for inducible enzymes (Kepes and Beguin, 1966; Kepes, 1967; Leive and Kollin, 1967). There is even some evidence for the concurrent transcription of several messenger strands from a given D N A template, i.e., new strands are initiated prior to the completion of one, or several, of the preceding strands (Zimmerman and Levinthal, 1967). Therefore, if deinduction blocks the initiation of new mRNA molecules, all of the partly synthesized strands will still be completed (Alpers and Tomkins, 1965) and the steady-state concentration of functional messenger should not be affected by deinduction until the last initiated strand is released from the D N A template. AD app‘irently acts differently on this system. By combining with DNA, AD not only inhibits further initiations but also prevents the completion of already initiated and partly completed mRNA strands. The outlined difference between AD administration and deinduction is clearly demonstrated by experimental data. Addition of AD within 2% minutes after the induction of p-galactosidase in E. coli at 30°C. totally suppresses enzyme formation (Leive, 1965a). This time span is probably necessary for the completion and release of the first initiated mRNA molecule. Removal of the galactosidase inducer, even within a fraction of 1 minute after the induction, does not interfere with the completion of the entire induction process and the so-called “elementary wave” which leads to the production of a small amount of enzyme protein (Kepes and Beguin, 1966; Kepes, 1967). Other experimental data, obtained with p-galactosidase in E. coli as well as with histidase in B. mbtilis, confirm this concept. The exponential decline in the rate of enzyme synthesis begins after a short lag when deinduction is initiated by quick dilution of the inducer, but no lag is found after AD treatment. In addition, slightly higher amounts of enzyme are obtained after deinduction as compared to AD inhibition (Hartwell and Magasanik, 1964; Kaempfer and Magasanik, 1967; Leive and Kollin, 1967). The messengers of both these inducible enzymes are very short-lived. The pgalactosidase messenger in E. coli decays with a half-life of 1.3-2.5 minutes at 30°C. (Nakada and Magasanik, 1964; Leive, 1965b; Kepes, 1967), and the histidase messenger in B. szlbtilis with a half-life of ca. 2.5 minutes at 37°C. (Hartwell and Magasanik, 1963). Since in both cases the presence of inducer has no effect on the decay rate of

212

KONRAD KECK

the respective messengers, and since the messenger decay rates after deinduction and after AD administration are not significantly different, we can conclude that the inhibition of RNA synthesis in itself has no effect on the degradation of functional messengers.

4. Long-Lizied Messetzgers The synthesis of the enzyme penicillinase continues for 30-40 minutes after AD treatment of the fully induced cells of an inducible strain of Bacillus cereus. In the corresponding constitutive strain, the RNA template for the same enzyme is even more stable with a lifetime of ca. 2 hours (Pollock, 1963). Increased mRNA stability of the constitutive penicillinase, as compared to its inducible counterpart, was also reported for BacillziJ licheniformis. Yudkin (1966) found that synthesis of the maximally induced penicillinase ceased 5 minutes after the addition of AD, while in the constitutive strain (which differed from the former by a single mutation in the regulator gene) penicillinase synthesis continued for 20 minutes. Nevertheless, both strains of B. licheiziformis responded identically to AD with a biphasic decay of the rate of total protein synthesis. These observations indicate that messenger decay could be subject to gene-specific controls. Bacterial messengers with extremely long lives also seem to exist. In B. cereus the process of sporuhtion requires specific messengers which are synthesized approximately 4 hours prior to the onset of sporulation. The analogs azaguanine and fluorouracil, administered at this time, prevent sporulation. The same drugs, as well as AD, do not prevent further synthesis of spore proteins when added to the culture at the time of beginning sporulation, although chloramphenicol treatment at this stage inhibits spore formation (Rosas del Valle and Aronson, 1062; Aronson and Rosas del Valle, 1964). The synthesis of flagellin, which is the main protein component of bacterial flagella and does not contain tryptophan, is not affected for a considerable time period when R N A synthesis is strongly suppressed by tryptophan or uracil starvation of a stringent strain of B. .in/htili~(Martinez, 1966) or Salmouella typhiimriwn (McClatchy and Rickenberg, 1967). In other experiments the synthesis of R N A in S. typhirnzirizmz was inhibited by AD; nevertheless, there was no effect on leucine incorporation into flagellin for a period of 90 minutes, although both the synthesis of P-galactosidase, coded by a lclc episome, and the synthesis of total protein were immediately inhibited under these conditions (McClatchy and Rickenberg, 1967).

B. ELICARYOTIC ORGANISMS Umtuble K N A The mRNA of eucaryotic cells in general is characterized by a much longer functional life as compared to bacterial mRNA, ranging from approximately 1 1.

METABOLISM O F ENUCLEATED C E L L S

213

hour to many days. Direct determination of the turnover rates of long-lived messenger is very difficultand indirect estimates based on protein synthesis are often required. A relatively unstable fraction was found in HeLa cells. Following a labeling period of 30 minutes as much as one-third of the acid-insoluble label was lost froin the cells within 8 minutes after A D administration. In sucrose gradients most of this unstable RNA sedimented in the 35- to 4 0 3 region together with ribosomal precursor RNA. The same region also contained R N A fractions that gave a high stimulating activity in the E. coli system im vitro and, furthermore, anneal efficiently with HeLa cell D N A (Schemer et nl., 1963). As with bacteria, the possibility must be considered that the unstable RNA consists in part of unfinished and unprotected molecules that carry a significant portion of the acidinsoluble label after brief incubation with radioactive precursors. Some of the labile RNA in HeLa cells might be identical with the rapidly degraded nuclear RNA described by Harris and Watts (1962). Cytoplasmic mRNA in HeLa cell polysoines, on the other hand, proved more stable. After 3 hours of A D treatment only 5 0 % of the polysomes were degraded. Isolated polysomal mRNA sedimented as a heterugeneous fraction in sucrose gradient, with a peak in the 10-S. region, and had a base composition similar to that of D N A (Penman et al., 1963). Trakatellis et nl. (1965b) noted significant differences in the half-life of mRNA in mammary carcinoma cells, depending on the method of deterinination. According to the rate of polysome breakdown, the loss of labeled RNA in the 5- to 2 0 3 region, and the decrease in amino acid-incorporating activity of ribosomal preparations in z G t w , mRNA seemed to decay with a half-life of ca. 4 hours. A much shorter half-life of only 30 minutes was derived from the labeling kinetics of polysvmal mRNA under steady-state conditions. On the other hand, a messenger half-life in excess of 4 hours was computed from the decline in rate of protein synthesis iiz vivo after treatment with AD. The authors suggest that mRNA life is extended under nonsteady-state conditions, perhaps as an indirect result of an increasing rate of ribosome attachment to inRNA during the later phase of polysome degradation. The statistically increased occupancy of the 5’-terminus of mRNA by attaching ribosomes might offer better protection from exonuclease attack (Trakatellis et d., 1965b). Extensive work on mRNA life has been carried out on rat liver. AD must be administered to the animals in relatively high doses in order to efficiently inhibit RNA synthesis. Under such experimental conditions the level of liver poIysomes decreases by 30-8070 within 4-8 hours (Staehelin et nl., 1963a). Revel et al. (1964) also noted the loss of polysomes but could not confirm the irz vivo disaggregation of polysomes on the basis of electron-optical observations. They concluded that the polysome breakdown occurred during the isolation, perhaps caused by the indirect effect of AD. It was later shown, however, that ordered

2 14

KONRAD KECK

arrays of ribosomes attached to the endoplasmic reticulum of the cell persist after mRNA degradation in vivo and that, therefore, their presence cannot be used as evidence for the presence of functional polysomes (Blobel and Potter, 1967a). The extreme stability of the majority of rat liver messengers, with a functional life of at least 40 hours, was postulated by Revel and Hiatt (1964) on the basis of sustained protein synthesis after a single injection of AD. Applied to bulk messenger, these values are probably too high since liver cells begin to recover from the effects of A D approximately 14-17 hours after the administration of a single dose of the drug (Schwartz et al., 1965). The decay of rat liver mRNA over extended periods of time was also studied by Wilson and Hoagland (1967). A second injection of AD was administered in their experiments to prevent the recovery of the cells. A biphasic decay of rat liver mRNA, with half-lives of ca. 3 hours and 80 hours, was derived from the slopes of the semilogarithmic plots of polysome levels versus time. The rate of amino acid incorporation in z h o , corretced for the specific radioactivity of the amino acids in the cellular pool, declined concomitant with the level of polysomes. A rapidly labeled RNA fraction, which sedimented in sucrose gradients with a 17-S. peak and has aG C/A U ratio of 0.8, also decayed parallel to the polysome level. The long-lived polysome population appears to be mainly responsible for the synthesis of albumin, the main export protein of liver (Wilson el al., 1967). The existence of polysomes in rat liver that is refractory to decay were also recognized by Korner and Munro (1963), Staehelin et al. ( 1963a), Villa-Trevino et al. ( 1964), and Kwan and Webb (1967).

+

+

2 . Protein-Specific Differences in Template Life

Differences in the half-life of individual protein-specific RNA templates are of great interest because they might reveal new and novel mechanisms for the regulation of protein levels in the cell. Differences in messenger stability occur in unicellular organisms as well as in multicellular organisms; in the latter they seem to be expressed in the form of time and tissue-specific patterns as part of cellular differentiation. The life span of a given mRNA species can only be determined indirectly from the anucleate synthesis of the respective protein, and the results are therefore subject to ambiguities, as outlined in Section II1,D. Early work with the unicellular alga Acetabularia has shown that the synthesis of several enzymes became affected at characteristically different times after enucleation, ranging from 1 to 3 weeks (Baltus, 1955; Keck and Clauss, 1958; Clauss, 1959). Since basic metabolic processes were not inhibited during this period the termination of enzyme synthesis was thought to be caused by the depletion of protein-specific messengers rather than by general biochemical lesions (Keck, 1965). Marchis-Mouren and Cozzone (1966) determined the messenger life for six enzymes in rat pancreas. The enzyme proteins were pulse-

METABOLISM O F ENUCLEATED CELLS

21 5

labeled for 10 minutes at various times after AD injections and then isolated in partially pure form. The amount of label incorporated into a given enzyme protein represented the instantaneous rate of synthesis at the time of sampling, since it was found that the specific radioactivity of the particular amino acid pool in the cells did not significantly change after AD administration, and protein degradation could be ignored because of the brevity of the pulse. The RNA templates for three basic enzyme proteins proved to be considerably more stable than the templates for three acidic proteins; the respective half-lives were 8 hours and 3 hours. No correlation was found between the length of a given messenger strand, as reflected in the molecular weight of the corresponding protein, and its stability. The pulse-labeling experiments were also extended to rat liver, which, in general, contained templates of greater stability than pancreas (Cozzone and Marchis-Mouren, 1967). Nevertheless, mRNA that codes for basic proteins was again found to be more stable than mRNA coding for acidic proteins. It was proposed that the net charge of the nascent protein influenced the rate of messenger degradation. The functional lifetimes of the messengers for two export proteins of liver were investigated by John and Miller (1966). The production of serum albumin and fibrinogen by the isolated and perfused rat liver was measured chemically and serologically over an 8-hour period after the infusion of AD. Rapid inhibition of protein synthesis by puromycin ruled out the presence of significant amounts of preformed proteins; the absence of nonspecific toxicity was indicated by the normal rates of urea synthesis and of changes in a-amino nitrogen. Rat serum albumin synthesis declined with a half-life of 2-4 hours, and fibrinogen synthesis with a half-life of 1y2-2hours. The decline in the rate of lysine-C14 incorporation into total liver protein, which is indicative of the average half-life for all liver protein templates, gave values of 3-4 hours, but the existence of templates with significantly longer lives was also suggested. Inducible enzymes in rat liver were investigated by Pitot et al. (1965). The enzymes serine dehydrase, ornithine transaminase, and tyrosine transaminase were induced by feeding casein hydrolyzate to protein-fasted rats. At various times after the induction, a second inducing stimulus was administered together with, or without, AD. The following template lifetimes-defined here as the finite period of time after AD administration during which the system supports enzyme synthesis-were obtained: G 8 hours for serine dehydrase, 18-24 hours for ornithine transaminase, and less than 3 hours for tyrosine transaminase. The uninduced base level of one of these enzymes, viz., tyrosine transaminase, seems to be maintained by longer-lived templates (Pitot, 1964). Results of preliminary experiments, also reported in the paper, indicated template lifetimes of over 2 weeks for tryptophan pyrrolase, and less than 3 hours for thymidine kinase. It is not likely that enzymically inactive precursor proteins existed for any of the

216

KONRAD KECK

investigated enzymes since induction of the enzymes could be inhibited by puromycin (Pitot et al., 1965). In contrast to the preceding discussion, Bloom et ul. (1965) suggest that the intracellular milieu might determine the lifetime of RNA templates in a given cell type and that the finding of greatly different messenger lives in tissues reflected the heterogeneity of cell populations rather than genuine differences within a given cell type. They found in support of their ideas that the ADinitiated exponential decay of hydroxyproline-CI4 incorporation, representing collagen synthesis in their fibroblast culture, and of proline-Cl4 incorporation, representing all noncollagen protein synthesis, both proceeded with the same half-life of ca. 3 hours. These experiments do not disprove, however, that a small proportion of the proteins has templates with significantly longer or shorter lifetimes. 3. Lorig-Lived iMe.rJenger in Eiuaryoter

Metazoan cells in certain stages of development and differentiation contain messengers of extreme longevity, i.e., a life in excess of 1 2 hours. Evidence for the existence of such mRNA species is mostly indirect and either inferred from the continued synthesis of proteins, or from the persistence of a certain population of polysomes in the absence of RNA synthesis. The most characteristic examples are cells that have acquired a highly specialized function during their terminal stage of differentiation, which is often followed by the natural degeneration of the cell nucleus. Quite frequently the functional specialization is restricted to the synthesis of large amounts of one or very few proteins. Messengers of extreme longevity are produced during mammalian erythrocytic development. According to cytochemical tests, there is no RNA synthesis after the basophilic erythroblast stage, while hemoglobin synthesis is most pronounced at later stages, notably in erythrocytes (Grass0 et al., 1963). The time interval between the terminal period of RNA synthesis and entrance into the reticulocyte stage is approximately 40 hours, as determined by pulse labeling of in vivo maturing cells (DeBellis et aL, 1964). Danon et ul. (1965) cited 48 hours as the time necessary for the completion of the developmental process after inhibition of the formation of new cells by AD. Corroborating experiments with in ziitro incubated erythrocytes have conclusively demonstrated that reticulocytes do not synthesize significant amounts of RNA (Marks et ul., 1962; Burny and Chantrenne, 1964), although the cells maintain high levels of d e novo hemoglobin synthesis under thes conditions (Kruh and Borsook, 1956; Borsook et al., 1957). In highly specialized cells in the lens of the eye the synthesis of lens protein is maintained by stable messengers. The tissue-specific localization of long-lived messenger was demonstrated in the 12-day chick lens by radioautographic studies.

METABOLISM O F ENUCLEATED CELLS

2 17

Amino acid incorporation continued in the highly differentiated lens core for at least 8 hours after A D treatment, while protein formation was not detectable in the epithelial cells (Reeder and Bell, 1965). T h e presence of long-lived and short-lived m R N A in the 14-day chick lens was concluded from the biphasic decay of polysomes, with half-lives of 3 hours and over 30 hours, after inhibition of R N A synthesis. The functional capacity of the stable polysome population in the lens was established hy pulse labeling with amino acids (Scott and Bell, 1965). Messenger with a half-life of at least 30 hours was also found in the calf lens (Spector and Kinoshita, 1965). There the synthesis of a-, 1.1-, and y-crystallins takes place during the transformation of lens epithelial cells into fiber cells. While the synthesis of crystallins during the early stage of differentiation of epithelial cells into elongated fiber cells is still sensitive to AD, a stabilization of the messengers occurs during their terminal differentiation into cortex fiber cells (Papaconstantinou et al., 1964, 1966; Papaconstantinou, 1967; Stewart and Papaconstantinou, 1967). A very stable population of polysomes, presumably engaged in the synthesis of feather keratin, was found in the skin and feather buds of chick embryos (Humphreys et d.,1964a,b). T h e stable population of polysomes can survive 12 hours of incubation with AD. In the 15-day-old skin, at a time when deposition of keratin normally takes place, stable polysomes become more prevalent and rapidly incorporate labeled amino acids into nascent protein (Humphreys et al., 1964b). Subsequent investigations have provided evidence that the appearance o f an inactive class of tetramer polysomes in embryonic chick skin was the result of an artifact produced by exposure o f the cells to low temperature (Humphreys and Bell, 1967) and not the expression of “masked” messengers (Spirin, 1966). Short- and long-lived messengers seem to support protein synthesis in incubated lamb thyroid slices. T h e synthesis of some proteins is sensitive to AD, but the labeling of thyroglobulin continued for 5-21 hours at rates that were not significantly different from the control rates (Seed and Goldberg, 1963). Another example of highly specialized cells are the blood platelets, which continue to incorporate amino acids into protein for at least 72 hours during i71 ~ i t m incubation, even though the platelets do not contain measurable amounts of D N A and therefore probably cannot synthesize m R N A during their normal life span of 3-8 days (Booyse and Rafelson, 1967).

4. Masked iMe.iseizger The concept of a masked or inactive messenger was originally developed as one possible explanation for the striking onset of protein synthesis in sea urchin eggs after fertilization o r parthenogenetic activation under conditions that preclude de nova R N A synthesis, such as pretreatment with A D (Gross and Cou-

218

KONRAD KECK

sineau, 1963, 1964) or enucleation (Tyler, 1962, 1963; Brachet et a/., 1963; Denny, 1963). The presence of significant amounts of mRNA in unfertilized sea urchin eggs was later confirmed by two different techniques. Egg RNA preparations were shown to elicit high template activity in in vitro systems (Maggio et al., 1964; Slater and Spiegelman, 1966a,b) and to hybridize efficiently with homologous D N A (GliSin et al., 1966; Whiteley et al., 1966). A series of investigations on the nature of the biosynthetic block in the unfertilized egg (see reviews by Spirin, 1966; Nemer, 1967; Tyler, 1967) led to the realization that the suppression of protein synthesis could not be caused by deficiencies in the general machinery of protein synthesis, nor could it be entirely the result of an incompetence of egg ribosomes. Mdjor experimental support for this view came from the finding that protein synthesis could be evoked in an in vitro system from unfertilized eggs by the addition of synthetic polynucleotides (Nemer, 1962; Tyler, 1962; Wilt and H u h , 1962; Nemer and Bard, 1963). The decisive parameter for the suppression of protein synthesis thus appears to be the unavailability of mRNA, although partial impairment of ribosome function cannot be ruled out. The latter possibility is borne out by the observation that mild pretreatment of egg ribosomes with trypsin further enhances their response to synthetic or homologous mRNA in the in vitro system (Monroy et al., 1965). Proteins also seem to be involved in the masking of maternal mRNA in the unfertilized egg. In untreated homogenates from sea urchin eggs mRNA sediments relatively rapidly and can be located in the 12,000 x 8-pellet by virtue of its template activity. After mild trypsin treatment of the homogenate, however, mRNA appears in the supernatant fraction and in the ribosome fraction, perhaps initiating the spontaneous formation of polysomes (Mano and Nagano, 1966). Another form of presumably nonfunctional cytoplasmic mRNA was discovered in early embryos of the loach (Belitsina et al., 1964; Spirin et al., 1964) and of the sea urchin (Spirin and Nemer, 1965). This RNA is combined with protein and forms discrete classes of particles with sedimentation coefficients from 20 to 70 S. It hybridizes efficiently with homologous D N A (Spirin and Nemer, 1965; Infante and Nemer, 1968) and elicits high template activity (Spirin et al., 1964). A distinct correlation exists between the sedimentation value of the particles and that of their RNA component (Nemer and Infante, 1965). The mRNA-containing particles can be characterized by density equilibrium centrifugation after fixation with formaldehyde (Spirin et al., 1966). The buoyant density of the particles from sea urchin embryos is generally lower than that of ribosomal subunits, but mRNA particles of relatively high density also exist. However, mild ribonuclease treatment of these heavy particles results in a decrease of their buoyant density, a response not shared by ribosomal subunits that are attached to mRNA strands (Infante and Nemer, 1968). The mRNA particles have been termed “informosomes” by Spirin et a/.

METABOLISM OF ENUCLEATED CELLS

219

(1964) and are though to represent a transitory stage of newly synthesized mRNA, which perhaps offers protection to mRNA during its transport from the the nucleus to the cytoplasm and during its subsequent storage in masked form (Spirin, 1966). In earIy sea urchin embryos newly synthesized messenger is also found in the so-called “light” polysomes which, contrary to the “heavy” polysomes, do not incorporate labeled amino acids into nascent protein and thus appear to be inactive in protein synthesis (Spirin and Nemer, 1965; Infante and Nemer, 1967). The eventual fate of the informosomes and their possible relationship to nonfunctional polysomes in early embryos are presently unknown. Attempts have been made to apply the concept of messengers in masked form to a number of cell systems, notably those containing long-lived messengers (Spirin, 1966; Tyler, 1967). Whether or not a long-lived messenger indeed passes through a temporary stage as a masked form rests on the experimental proof that there is a significant time lag between the buildup of a specific messenger pool and the onset of synthesis of the respective protein. In some of the examples cited the existence of such a lag has not been unequivocally established and other interpretations of the phenomena are possible. Thus, in B . cereuj the AD-sensitive period for spore formation extends between the culture age of 8 and 9y4 hours, yet morphological alterations which eventually lead to spore formation begin at 12 hours. The proteins involved in sporulation could have been synthesized, however, in advance of the structural differentiation and, in fact, the required complement of spore proteins must have been acquired prior to 12 hours by a portion of the cells since the addition of chloramphenicol to the culture at this time permits the completion of the spomlation process in 10% of the cells (Rosas del Valle and Aronson, 1962). In the chick embryo the lag between the end of the AD-sensitive phase and the appearance of hemoglobin could also have been caused, according to Wilt (1965) by a delay in the availability of substrate for the synthesis of the prosthetic group. The role of heme in the initiation of new globin chains and in the control of hemoglobin synthesis during maturation of the cells has been discussed by Zucker and Schulman (1968) and Schulman ( 1968), respectively. The appearance of inactive tetramer polysomes just prior to keratin synthesis in embryonic chick feathers has been shown, as mentioned earlier in this review to be caused by an isolation artifact (Humphreys and Bell, 1967). And last, the delayed increase in phosphatase activity in enucleated cells of Acetubalaria (Spencer and Harris, 1964) might be the result of an unmasking of cytoplasmic mRNA but could also indicate chloroplastic control of this enzyme. j.

Enucleated Cells

The unicellular marine alga Acetabularia has often been cited as an organism with extremely long-lived messengers. Originally this concept was applied to species-specific “morphogenetic substances” of unknown chemical nature, which

220

KONRAD KECK

were thought to be synthesized by the cell nucleus and remain active in the anucleate cytoplasm for many weeks (Hammerling, 1953). Morphogenesis in anucleate cells is accompanied by a net increase of several hundred percent in the amount of total protein (Vanderhaeghe, 1954; Brachet et al., 1955; Clauss; 1958; Hammerling et al., 1959). Total protein synthesis ceases approximately 3 weeks after enucleation at a time when basic metabolic processes such as photosynthesis or respiration are hardly affected (Chantrenne-Von Halteren and Brachet, 1952 ; Hammerling et al., 1959). Furthermore, since low-molecular weight metabolites and protein precursors are still abundant at this time (Clauss and Keck, 1959; Bremer et al., 1962), the relatively early cessation of protein synthesis, as well as the even more restricted synthesis of individual enzymes (cf. Section IV,B,2), is probably not caused by general metabolic lesions but rather by the depletion of cytoplasmic RNA templates (Keck, 1965). It is very difficult to estimate the contribution of genetically independent cytoplasmic systems, particularly chloroplasts, to anucleate protein synthesis in Acetabularia. The presence of DNA in Acetabularia chloroplasts was recognized by Baltus and Brachet (1963) and by Gibor and Izawa (1963); the synthesis of chloroplastic RNA was demonstrated by Naora et al. (1960), Schweiger and Berger (1964), and Goffeau and Brachet (1965). Evidence for the occurrence of chloroplastic polysomes in cells of higher plants was provided by Stutz and No11 (1967). Several attempts were made to establish in Acetabularia the nuclear origin of messengers for various enzyme proteins. The discovery of species-specific molecular forms of acid phosphatase in Acetabularia (Keck, 1960) offered the possibility of establishing the localization of its structural gene by nuclear transplantation experiments. Buffer extracts from each of three investigated species contained one electrophoretically distinct phosphatase type. The phosphatase type of one species, Acicdaria Schenckii, proved to be convertible to the Acetabidaria mediterranea type via unknown reactions at the molecular level. The conversion process occurred in vivo in a variety of cellular graft combinations between the two species and after the injection of A. mediterranea cytoplasm into nucleate or anucleate Acirzl&a cells. The conversion process could also be initiated in vitro in a mixture of the respective homogenates (Keck, 1961; Keck and Choules, 1963). It was demonstrated by repeated amputation of cytoplasm from hybrid cells that the ultimate electrophoretic character of this enzyme was determined by the remaining cell nucleus (Keck, 1961). Tripplett et al (1965) later discovered additional phosphatase types in A. mediterranea after detergent treatment of the homogenates and studied their specific activities after enucleation. The activity of one of the phosphatase types, as well as the total phosphatase activity of the homogenate, was found to increase rapidly after the twelfth day, implying chloroplastic control of the enzymes. In another species, Acetabularia rrenulata, phosphatase activity, measured at pH 5 .O, increased steadily

METABOLISM OF ENUCLEATED CELLS

221

in anucleate cells over a period o f 3 weeks (Spencer and Harris, 1964). Contrary to these findings, Keck and Clauss (1958) and Keck (1961) observed that in A. mediterrama the rate of acid phosphatase synthesis continued for only 1 week at control levels after the enucleation, followed by a subsequent slow decline. This residual synthetic capacity is conserved when protein synthesis is inhibited by starving anucleate cells in the dark for periods of up to 3 weeks and is fully expressed upon reillumination (Keck, 1965; Schlapfer and Keck, 1964 unpublished observations). Although slightly different culture techniques (Keck, 1964) and assays were used, the reason for the discrepinq in results is not readily understood. The nuclear control of several species-specific malic dehydrogenase types, or isozymes,” in Acetabidaria was postulated by Schweiger et al. (1967). Strictly speaking, however, here as well as with acid phosphatase it was only demonstrated that the ultimate electrophoretic mobility of the proteins is determined by the nucleus and the possibility remains that nuclear control is restricted to secondary structural modifications of the enzyme protein, perhaps similar to the ones reported for cholinesterase (Svensmark, 196I ; Augustinson and Ekedahl, 1962). There is good experimental evidence for the occurrence of structural modifications in the case of acid phosphatase (Keck and Choules, 1963) and even in the case of malic dehydrogenase a similar conversion may have occurred and remained undetected. Also relevant are observations (Schweiger et al., 1967) that the cytoplasmic enzyme type mysteriously disappears after the implantation of the species-foreign nucleus, while the same enzyme type persists in the anucleate cell serving as the control. Unlike Aretabl/larja anucleate halves of Amoeba protezis do not seem to synthesize RNA under rigorously controlled conditions. Earlier reports to the contrary (Plaut, 1958; Plaut and Rustad, 1957, 1959) were probably attributable in part to the presence of ingested bacteria. Prolonged starvation of the amebas prior to surgery greatly reduces the incorporation of labeled RNA precursors (Prescott, 1959), although an unequivocal answer to the question was not obtained. Another difficulty is posed hy the occurrence of self-duplicating “ D N A bodies” in A . protezfs cytoplasm which might be endosymbionts (Wolstenholme and Plaut, 1964; Cummins and Plaut, 1964). This theory was recently substantiated by the discovery of infectious strains of bacteria which live parasitically in Amoeba dircoides (Jeon and Lorch, 1967) . Anucleate halves of Aranthnmneba, a small amoeba which can be cultured in sterile nutrient medium, do not incorporate any detectable amounts of labeled RNA precursors and probably cannot synthesize RNA in the absence of the cell nucleus (Prescott, 1960b). Similarly, the absence of measurable cytoplasmic RNA synthesis was concluded from experiments with microsurgically enucleated human amnion cells (Goldstein et al., 1960) and enucleated Tetruhymena cells (Prescott, 1962). Only a few studies have been carried out on the residual protein synthesis of 1 < .

222

KONRAD KECK

physically enucleated heterotrophic cells. Anucleate fragments of human amnion cells continue to incorporate amino acids into protein at control rates for a period of 20-30 hours (Goldstein et al., 1960). Protein synthesis continues for several hours in amacronucleate paramecia which maintain a very low level (23% of controls) of RNA synthesis (Kimball and Prescott, 1964). The proteinsynthesizing capacity of anucleate Tetrabymenu is relatively low; the incorporation of histidine into proteins declines 50% within 30 minutes after the enucleation operation (Prescott, 1962).

V. Nature of mRNA Decay A. ENZYMICDEGRADATION The nature of mRNA degradation is of considerable interest because it might shed light on the molecular events that determine the mean life of individual mRNA species. It might be useful to distinguish at first between the inactivation of a mRNA molecule and its subsequent gross destruction (Kivity-Vogel and Elson, 1967). The latter process is undoubtedly carried out by RNA-hydrolyzing enzymes. In E. coli, the only system that has been investigated in detail, three enzymes have been found that hydrolyze RNA: RNase I, a potassium-dependent phosphodiesterase (RNase 11) , and polynucleotide phosphorylase. Of these enzymes, RNase I is the least likely candidate for the postulated function. The end products of RNase digestion, the 3’( 2’) -nucleoside monophosphates have not been found among the in vitro breakdown products of artificial messenger (Barondes and Nirenberg, 1962; Spahr and Schlessinger, 1963) or natural messenger (Andoh et ul., 1963a). The it2 vivo degradation of mRNA by RNase I can be ruled out on the basis of experiments with RNase I-less mutants of E. coli. Although the RNase levels of these strains were less than 1% of the wildtype level, normal decay rates were recorded for rapidly labeled RNA (Gesteland, 1966) and for the messenger of p-galactosidase (Kivity-Vogel and Elson, 1967). Similar conclusions were also reached from experiments with E . cnli spheroplasts that had lost their RNase I activity (Artman and Engelbert, 1965). Polynucleotide phosphorylase, an exonucIeoIytic enzyme, attacks polyribonucleotides from the 3’-hydroxy end of the chain in the presence of orthophosphate (Lehman, 1963). The end products of this reaction, the 5’-ribonucleoside diphosphates, have been identified among the products of mRNA decay in some of the in vitro systems (Sekiguchi and Cohen, 1963; Andoh et ul., 1963a,b). In other systems, polynucleotide phosphorylase activity is not detectable. Enzyme(s) in the supernatant fraction of E . coli hydrolyze the mRNA of ribosomal preparations in the complete absence of orthophosphate and without the concomitant accumulation of 5’-nucleoside diphosphates (Spahr and Schlessinger, 1963). Nucleases other than polynucleoside phosphorylase are apparently bound

METABOLISM O F ENUCLEATED CELLS

223

to the mRNA-ribosome complex and degrade mRNA during the incubation of these complexes in phosphate-free buffer. The end products of the enzymic reaction were unfortunately not identified in these experiments (Artman and Engelbert, 1964). It was furthermore shown that the in vivo inactivation of the (3-galactosidase messenger is not impaired in a mutant of E. coli that is deficient in polynucleotide phosphorylase (Kivity-Vogel and Elson, 1967). W e can conclude from the cited experimental evidence that the presence of this enzyme is required neither for inactivation, nor for the gross destruction of mRNA, although the participation of this enzyme in degradative processes is, of course, not ruled out. The involvement of a phosphodiesterase in RNA hydrolysis has been suggested by a number of investigators (Spahr and Schlessinger, 1963; Sekiguchi and Cohen, 1963; Spahr, 1964). The phosphodiesterase from E . coli has been purified 600-fold and was found to require the presence of potassium ions and divalent ions (magnesium) for its activity. The diesterase specifically hydrolyzes single-stranded RNA to 5’-nucleoside monophosphates and does not attack RNA in the helical configuration (Singer and Tolbert, 1965). In cell homogenates this enzyme is bound to some extent to ribosomes and could therefore be present in washed ribosome-mRNA preparations. Whether or not any of the described nucleases is actually responsible for the in vivo breakdown of mRNA is still problematic. On theoretical grounds it has been suggested (cf. Kepes, 1967) that mRNA destruction is exonucleolytic and proceeds in direction from the 5’- to the ?/-end of the messenger strand, thus being equidirectional to the transcription as well as translation process. The wave of degradation of a given strand could thus closely follow the last ribosome. Such a system would prevent the formation of abnormal proteins during mRNA destruction.

B. INITIATIONOF DECAY The knowledge that one or a combination of several enzymes is responsible for the in vivo breakdown of mRNA to reutilizable end products does not in itself explain the initiating event. The often confirmed exponential rate of messenger inactivation points to the randomness of this event. Its frequency of occurrence differs greatly among individual messenger species of a cell, all of which are presumably exposed to the same concentration of nucleolytic enzymes. Thus, additional parameters must exist that determine the decay rate of individual messenger species in the cell. In spite of the close functional relationship between mRNA and protein synthesis there is no experimental support of the notion that the life of a messenger molecule is limited by functional “wear and tear.” This conclusion is mainly based on the observation that certain agents specifically inhibit the synthesis of proteins without retarding the breakdown of

224

KONRAD KECK

mRNA. Puromycin, for instance, inhibits protein synthesis without extending the average messenger life in B. subtilis (Fan et nl., 1964), or the life of (3galactosidase messenger in E. coli (Nakada and Fan, 1964). The amino acid analog methyltryptophan is another inhibitor of protein synthesis in E. coli that does not affect messenger decay. The decay rate of the pulse-induced (3-galactosidase-forming capacity in the presence of methyltryptophan is similar to that of control cells (Kepes, 1963). On the other hand, a considerable increase in the mean life of mRNA results from the inhibition of protein synthesis by chloramphenicol in E. colz ( Fqriesen, ’ 1966; Woese et nl., 1963; Forchhammer and Kjeldgaard, 1967) and in B . subtilis (Fan et al., 1964). Blocking of protein synthesis by amino acid starvation similarly retards the decay of the stimulating activity in E . coli R N A (Forchhammer and Kjeldgaard, 1967). Exposing cells of B. subtilis to an anaerobic environment significantly retards the decay of rapidly labeled RNA; protein and RNA synthesis are aIso inhibited under such conditions. Puromycin, added during the anaerobic phase, counteracts the protection of labile RNA by anaerobiosis (Fan et al., 1964). Comparable results were also obtained with (3-galactosidasemessenger in E. coli (Nakada and Fan, 1964). It can be concluded from the examples cited that it is not the absence o f protein synthesis per se, but some of the accompanying circumstances that can offer protection of the mRNA against degradation. Inhibition of protein synthesis by puromycin, for instance, involves the release of unfinished polypeptide chains (Allen and Zamecnik, 1962; Morris et al., 1962; Gilbert, 1963) and the loss of ribosomes from the messenger strand (Marks et al., 1963b). These reactions most likely expose the nonfunctional messenger to nuclease attack, while “freezing” of the translational process by chloramphenicol (Das et al., 1966) would tend to preserve the messenger because of its protected state. The meaningful interpretation of these experiments is complicated by the fact that neither the identity of the degrading enzyme(s) nor the nature of the hydrolytic attack on the polynucleotide chain is presently understood. The normal relationship between mRNA decay and the rate of protein synthesis is not altered by shifts in temperature. The Arrhenius-type plot of the mean decay time of rapidly labeled RNA, of the rate of decline of the leucine incorporation, and of the growth constant of B. subtilis all gave the same slope between the temperature interval from $10” to 40°C. (Fan et al., 1964). Therefore, the temperature coefficient for mRNA degradation is identical with the coefficient for protein synthesis and that of balanced growth in general. This correlation is also borne out by P-galactosidase induction experiments with E. coli. Although at 25°C. the mRNA decay is significantly slower than at 4OoC., the same total amount of galactosidase is produced in both cases after pulse induction (Kepes and Beguin, 1966). consequently, the temperature coefficient for mRNA breakdown must be very close to that of peptide chain growth.

METABOLISM O F ENUCLEATED CELLS

225

VI. Conclusions Investigation of the anucleate state in pro- and eucaryotic cells has provided strong evidence that various protein-specific messengers of a cell differ greatly in their mean functional life. Admittedly, the average life of individual messenger species has been determined indirectly from the anucleate rate of protein synthesis and several factors must be considered that could be responsible for significant deviations from the “ideal” relationship between mRNA and protein synthesis. Nevertheless, it can be assumed that the majority of these factors affect all proteins to more or less the same degree. The observed differences between various proteins in the decline of their anucleate synthesis can thus be interpreted as a reflection of true differences in the mean life of the respective messengers, even though strict proportionality may not prevail. The concurrent existence of short- and long-lived messengers in the same cell type rules out the possibility that mRNA life is solely determined by the intracellular milieu, unless one postulates the localization of some messenger species in separate subcellular compartments, e.g., messenger in membrane-associated polysomes versus messenger in “free” polysomes. Such a model would require additional mechanisms for the selection of protein-specific messengers for their respective compartments. More attractive is the hypothesis that the statistical life span of a given messenger is determined by the structure of mRNA per se and therefore encoded in DNA. Structural differences at or near the 5’-terminus of the mRNA strand might very well control the accessibility of the molecule to the attachment of exonucleases. The concept of a genetically determined messenger life receives support from the observation that a single mutation in the operator region of a repressible gene results in a significant change in the mean life of the messenger for the respective enzyme (Yudkin, 1966). Genetic determination of the decay rate of mRNA includes the additional indirect control over the rate of protein synthesis via the cellular steady-state levels of the corresponding mRNA species. More experimental work is needed, however, before this or any other model can be seriously considered. REFERENCES Acs, G., Reich, E., and Valanju, S. (1963). Bioi-him. Biophy.r. Artu 76, 68-79. Adler, H. I., Fisher, W. D., Cohen, A., and Hardigree, A. A. (1967). Pror. Nutl. Arud. S1.i. U.S. 57, 321-326. Allen, D. W., and Zamecnik. P. C. (1962). Biochim. BiophyJ. Artu 55, 865-874. Alpers, D. H., and Tomkins, G. M . (1965). Pror. Natl. Acud. Sri. U S . 53, 797-803. Andoh, T., Natori, S., and Mizuno, D. (1963a). Biorhim. Biophyf. Actu 76, 477-479. Andoh, T., Natori, S., and Mizuno, D. (1963b). J. Biorhem. ( T o k y o ) 54, 339-348. Aronson, A. I., and Rosas del Valle, M. R. (1964). Biochim. Biophys. Acla 87, 267-276. Artman, M., and Engelbert, H. (1964). Biorhim. Biophyr. Artu 80, 517-520. Artman, M., and Engelbert, H. (1965). Biorhirn. Biophys. Actu 95. 687-690.

226

KONRAD KECK

Astrachan, L., and Fisher, T . N . (1961). Federation Pror. 20, 359-3551. Augustinson, K., and Ekedahl, G . (1962). Biorhim. Biophys. Acta 56, 392-393. Baltus, E. (1955). Congr. Intern. Bior-him., 3 8 , Brussels p. 76. Baltus, I:., and Brachet, J. (1963). Biorhim. Biophys. Arta 76, 490-492. Barondes, S. H., and Nirenberg, M. W. (1962). Science 138, 810-813. Barondes, S. H., Dingman, C. W., and Sporn, M. P. (1962). Nature 196, 145-147. Beard, N. S., Jr., and Armrntrout, S. A. (1967). Pror. NatL Arad. Sri. U S . 58, 750-757. Belitsina, N . V., Ajtkhozhin, M. A., Gavrilova, L. P., and Spirin, A. S. (1964). Biukhimiya 29, 363-374. Bishop, J., Leahy, J., and Schweet, R. (1960). Pror. Natl. Acad. Sci. U.S. 46, 1030-1038. Blobel, G., and Potter, V. R. (1967a). 1. Mol. Biol. 26, 279-292. Blobel, G., and Potter, V. R. (1967h). J. Mol. Biol. 28, 539-542. Bloom, S., Goldberg, B., and Green, H. (1965). Biorhem. Biophys. Res. Commun. 19, 317-321. Boedtker, H., and Stumpp, S. (1964). Biorhim. Biophys. Acta 87, 247-252. Bolton, E. T., and McCarthy, B. J. (1962). Pror. Natl. Arad. Sci. U.S. 48, 1390-1397. Booyse, F. M., and Rafelson, M. E. (1967). Biorhirn. Biophys. Arta 145, 188-190. Borsook, H., Fisher, E. H., and Keighley, G . (1957). J . Biol. Chem. 229, 1059-1070. Brachet, J. (1961). In "The Cell" (J. Brachet and A. E. Mirsky, eds.), Val. 2, pp. 771841. Academic Press, New York. Brachet, J., Chantrenne, H., and Vanderhaeghe, F. (1955). BioL.him. Biophys. Arta 18, 544-563. Brachet, J., Ficq, A., and Tencer, R. (1963). Expil. Cell Re.r. 32, 168-170. Brawerman, G., Gold, L., and Eisenstadt, J. (1963). Proc. Natl. Arad. Sci. US.50, 630638. Bremer, H. J., Schweiger, H. G., and Schweiger, E. (1962). Biorhirn. Biophys. Arta 56, 380-382. Bremer, H., and Konrad, M. W. (1964). Pror. Natl. Arad. Sri. U S . 51, 801-808. Burka, E. R., and Marks, P. A. (1964). I. M o l . Biol. 9, 439-451. Burny, A., and Chantrenne, H. (1964). Biochim. Biophys. Acta 80, 31-38. Burny, A,, and Marbaix, G. (1965). Biorhim. Biophys. Acta 103, 409-417. Byrne, R., Levin, J. G., Bladen, H. A., and Nirenberg, M. W. (1964). Proc. Natl. Acad. Sri. U.S. 52, 140-148. Chantrenne, H. (1965). Biorhim. Biophys. Arta 95, 351-353. Chantrenne, H., Burny, A,, and Marbaix, G. (1967). Progr. Nurleir Arid Res. Mol. B i d . 7, 173-194. Chantrenne-Von Halteren, M. B., and Brachet, J. (1952). Arch. Intern. Physiol. 60, 187. Clark, B. F. C., and Marcker, K. A. (1966). Nature 211, 378-380. Clauss, H. (1958). P h t a 52, 334-350. Clauss, H. (1939). P h t a 52, 534-542. Clauss, H. (1962). Naturwissen.rrhaften 49, 523-524. Clauss, H., and Keck, K. (1959). Planta 52, 543-553. Coconi, F. M., Bank, A., and Marks, P. A., (1966). J. M u l . Biol. 19, 525-540. Cozzone, A,, and Marchis-Mouren, G . (1967). Biochemistry 6, 3911-3917. Cummins, J. E., and Plaut, W. (1964). Biorhim. Biophys. Arta 80, 19-30. Danon, D., Zehavi-Willner, T., and Berman, G . R. (1965). Proi.. Natl. Arud. Sri. U.S. 54, 873-879. Das, H. K., Goldstein, A,, and Kanner, L. C. (1966). Mol. Pharmacol. 2, 158-170. Das, H. K., Goldstein, A., and Lowney, L. I. (1967). J. Mol. Biol. 24, 231-245.

METABOLISM O F ENUCLEATED CELLS

227

Davern, C. I., and Meselson, M. (1960). J . Mol. B i d . 2, 153-160. DeBellis, R. H., Gluck, N., and Marks, P. A. (1964). J . Cliw. Inzfe.rt. 43, 1329-1337. DeMars, R. I., Luria, S. E., Fisher, H., and Levinthal, C . (1953). Ann. Imt. Pasteur 84, 113. Denny, P. C. ( 1963). Ant. Zoologist 3, 505-506. Dietz, G . W., and Simpson, M. V . (1964). Federation P l o r . 23, 219-219. DiGirolamo, A., Henshaw, E. C., and Hiatt, H . H. (1964). J. Mol. B i d . 8, '479-488. Dintzis, H. M. (1961). Pror. Nail. Arad. S1.I. U.S. 47, 247-261. Drach, J. C., and Lingrel, J. B. (1966). Bioc-him. Biuphyx. At-ta 129, 178-185. Dreyfus, J. C . , and Schapira, G . (1966). Biochim. Biophjs. Acts 129, 601-612. Earl, D. C. N., and Korner, A. (1966). Arch. Bior-hem. Biophys. 115, 437-444. Eason, R., Cline, M. J.. and Smellie, R. M. S. (1963). J. B i d . Chem. 238, 3978-3984. Eliasson, E. E. (1967a). Biorhem. Biophys. Rer. Commun. 27, 661-667. Eliasson, E. E. (1967b). Exptl. Cell Res. 48, 1-17. Eliasson, E. E., Bauer, G . E., and Hultin, T. (1967). J . Cell Biol. 33, 287-297. Ellem, K. A. 0. (1966). J . M o l . Biol. 20, 283-305. Ellem, K. A. O., and Sheridan, J. W. (1964). Biochem. Biophys. Res. Commun. 16, 5 0 5 510. Fan, D. P., Higa, A,, and Levinthal, C. (1964). J. Mol. Biol. 8, 210-222. Feigelson, P., Dashman, T., and Margolis, F. (1959). Arch. Biochem. Biophys. 85, 478482. Fleck, A,, Shepherd, J.. and Munro, H. N . (1965). Srienie 150, 628-629. Forchhammer. J., and Kjeldgaard, N. 0. (1967). J . Mol. B i d . 24,459-470. Franklin, R. M. (1963). Biorhim. Biophys. Actn 72, 555-565. Friesen, J. D . (1966). J . Mol. B i d . 20, 559-573. Garren, L. D., Howell, R. R., Tomkins, G . M., and Crocco, R. M. (1964). Pro[-. Nad. Sri. U.S. 52, 1121-1129. Gesteland, R. F. (1966). J . Mol. Biol. 16, 67-84. Giacomoni, D., and Spiegelman, S. (1962). Science 138, 1328-1331. Gibor, A,, and Itawa, M. (1963). Proc. Natl. Arad. Sri. U.S. 50, 1164-1169. Gierer, A. (1963). J . Mol. Biol. 6, 148-157. Gilbert, W. (1963). J. Mol. Biol. 6, 371-388. Gillespie, D., and Spiegelman, S. (1965). J. Mol. B i d . 12, 829-812. Girard, M., Penman, S., and Darnell. J. E. (1964). Pror. Natl. Acud. Sri. U S . 51, 205211. Girard. M., Latham, H., Penman, S., and Darnell, J. E. (1965). J. Mol. Biol. 11, 187201. GliSin, V. R., GliSin. M. V.. and Doty. P. (1966). Pro[-. Nut/. Acid. Sii. U.S. 56, 285289. Glowacki, E. R., and Millette. R. L. (1965). J . Mol. Biol. 11. 116-127. Goffeau, A,, and Brachet. J. (1965). Biochim. Biophys. A[-/a95, 302-313. Goldberg, B., and Green, H. (1967). J . Mol. B i d . 26. 1-18. Goldberg, I. H., Rabinowitz, M., and Reich. I<. (1962). PFOC.Nutl. Arad. SI-i. U.S. 48, 2094-2 I o I . Goldstein, A,, Kirschbaum. J . B., and Roman. A . (1965). Pvor. Nud. Acud. Sci. U S . 54, 1669-1675. Goldstein, L.. Micou, J., and Crocker, T. T. (1960). Biorhim. Biophys. Acta 45. 82-86. Goodman, H. M., and Rich, A . (1962). Pror. Natl. Acad. Sci. U.S. 48, 2101-2109.

228

KONRAD KECK

Grasso, J. A., Woodard, J. W . , and Swift, H . (1963). Pror. Natl. Arad. Sci. U.S. 50, 134-140. Gros, F.,and Naono, S. (1961). I?? “Protein Biosynthesis” (R. J. C. Harris, ed.), pp. 195-205. Academic Press, New York. Gros, F.. Gilbert, W., Hiatt, H. H., Attardi. G., Spahr. P. F., and Watson, J. D. (1961a). Cold Spring Harbor Synip. Quant. B i d . 26, 111-132. Gros, F., Hiatt, H., Gilbert, W., Kurland. C. G., Risebrough, R. W., and Watson, J. D. (196lb). Nature 190, 581-585. Gross, P. R., and Cousineau, G. H. (1963). Bioihem. Biophys. Res. Commun. 10. 321326. Gross, P. R., and Cousineau, G . H . (1964). E x [ ) / / . Cell Re.r. 33, 368-395. Hadjiolov, A. A. (1966). Biochim. Biophys. Acla 119, 54-556. Hammerling, J. (1953). lnteru. Reii. C y t d . 2, 475-498. Hfmmerling, J. (1963). Atin. Rev. Plant Yhy.riol. 14, 65-92. Hammerling, J., Clauss, H., Keck, K., Richter, G., and Werz, G . (1959). E x p / l . Cell Red. Suppl. 6, 210-226. Hall, B. D., and Spiegelman, S. (1961). P w i . Natl. Arad. Sci. U.S. 47, 137-146. Hall, J. B., Sedat, J. W., Adiga, P. R., Uemura, I., and Winnick, T . (1965). J . M d , Riol. 12, 162-173. Hamilton, L. D., Fuller, W . , and Reich, E. (1963). Nature 198, 538-540. Harris, H., and Watts, J. W. (1962). Proc. Roy. Sor. (Lortdo?/) B156, 109-121. Hartwell, L. H., and Magasanik, B. (1963). J . Mol. B i d . 7, 401-420. Hartwell. L. H.. and Magasanik, 8. (1964). J . Mol. Biol. 10, 105-1 19. Harvey, E. B. ( 1956). “The American Arbacia and Other Sea llrchins.” Princeton Ilniv. Press, Princeton, New Jersey. Hayashi, M., and Spiegelman, S. (1961). Pror. Natl. Arad. S1.i. U S . 47, 1564-1580. Hayashi, M., Spiegelman, S., Franklin, N . C., and Luria, S. I<. (1963). Pror. Natl. Arad. Sri. U.S. 49, 729-736. Hayashi, M. N., Hayashi, M., and Spiegelrnan, S. (1965). Biophys. J. 5, 231-241;. Hoagland, M.B., and Asconas, B. A. (1963). Proc. Natl. Arad. Sci. U S . 49, 130-17. Hoagland, M. B., Scornik, 0. A,, and Pfefferkorn, L. C. (1964). Yroc. Natl. Arad. S1.i. U.S. 51, 1184-1191. Holland, J. J., Buck, C. A., and McCarthy, B. J. (1966). Biorhernistry 5, 358-365. Honig, G.R., and Rabinowitz, M. (1965). Science 149, 1504-1506. Horowitz, J., and Chargaff, E. (1959). Nature 184, 1213-1215. Horowitz, J., and Kohlmeier, V. (1967). Biorhim. Biophys. Aria 142, 208-218. Humphreys, T., and Bell, E. (1967). Biorhem. Biophys. Res. Commun. 27, 443-447. Hurnphreys, T., Penman, S., and Bell, E. (1964a). Biophys. Sor-., 8th Ann, Meetjn
METABOLISM O F ENUCLEATED CELLS

229

Kaempfer, R. 0. R., and Magasanik, B. (1967). J . M o l . Biol. 27, 475-494. Kahan, E., Kahan, F. M., and Hunvitz. J. (1963). J . B i d . Chrm. 238, 2491-2497. Keck, K. (1960). Biorhrm. Biophys. Res. Commun. 3, 56-61. Keck, K. (1961). Ann. N . Y . Arad. Sri. 94, 741-752. Keck, K. (1964). I n “Methods in Cell Physiology” (D. M. Prescott, ~ d . ) Vol. , 1, pp. 189-213. Academic Press, New York. Keck. K. (1965). Pror. 16th Intern. Congr. Zool., Washington. D.C., 1963 3, 203-207. Keck, K., and Choules, E. A . (1963). J . Cell Biol. 18, 459-469. Krck, K., and Clauss, H . (1958). Botun. Gaz. 120. .li-49. Kmnell, D. (1964). J . Mol. B b l . 9, 789-800. Krnnry, F. T. (1967). Science 156, 525-528. Kenney, F. T., and Albritton. W. L. (1965). Pror. Natl. Ai.ad. S c i . U.S. 54. 1693-1698 Kepes, A. (1963). Biochim. Biophy.r. A r t a 76, 293-309. Kepes. A . (1967). Bioihitn. Biophys. Arta 138, 107-121. Kepes, A , , and Beguin, S. (1966). Biochim. Biophys. Acta 123. 516-560. Kerr, I. M., Cohen, N., and Work, T. S. (1966). Biuchem. J . 98, 826-835. Kimball, R. F., and Prescott, D. M. (1964). J. Cell B i d . 21, ,496497. Kirk, J. (1960). Biorhim. Biophys. Acra 42, 167-169. Kivity-Vogel, T., and Elson, D. (1967). Biurhim. Biophys. Arta 138, 66-75. Korner, A., and Munro, A . J. (1963). Biorhrnz. Biophys. Res. Commun. 11, 235-238. Kretsinger, R. H., Manner, G.. Gould, B. S., and Rich, A. (1964). Nature 202, 438-441. Kruh, J., and Borsook, H . (1956). J . Biol. Chrm. 220, 905-915. Kuff, E. L., and Roberts, N . E. (1967). J . Mol. Biol. 26, 211-225. Kwan, S. W . , and Webb. T. E. (1967). J . Biol. Chem. 242, 5542-5548. Lamfrom, H., and Knopf, P. M. (1964). J . M o l . Biol. 9, 558-575. Lamfrom, H., and Knopf, P. M. (1965). J . Mol. B i d . 11, 589-599. Laszlo, J., Miller, D. S., McCarthy. K. S., and Hochstein, P. (1966). Science 151, 10071010.

Latham, H., and Darnell. J. 1;. (1965). 1. Mol. B i d . 14, 1-12. Lebleu, B., Huez, G., Burny, A,, and Marbaix, G. (1967). Biorhim. Biophys. Arta 138, 186-188.

Lehman, I. R. (1963). Progr. Nurleir Acid Res. Mol. B i d . 2, 83-123. Leive, L. (1965a). Biochem. Bio1Jhy.r. Rrs. Commun. 20, 321-327. Leive, L. (1965b). J. Mol. B i d . 13, 862-875. Leive, L., and Kollin, V. (1967). J . Mol. Biol. 24, 247-259. Levinthal, C., Keynan, A., and Higa, A. (1962). Proc. NatI. Ai.ad. Sri. U.S. 48, 16311638.

Lin, E. C . C., and Knox. W. E. (1958). J . Biof. Chem. 233, 3186-1189. Lin, S., Mosteller, R. D.. and Hardesty. B. (1966). J . Mol. B i d . 21, 51-69. Loeb, J. N.. Howell, R. R., and Tomkins. G. M. (1965). Scirncr 149, 1093-1095. McCarthy, €3. J., and Holland, J. J. (1965). Pror. Natl. A~.ud.Sci. U.S. 54, 880-886. McClatrhy, J. K., and Rickenberg. H . V. (1967). J . Barti.riu1. 93, 115-121. Maggio, R., Vittorelli, M. L.. Rinaldi, A. M., and Monroy, A . (196.1). Biorhem. Biophy.r. Res. Commun. 15, 4 3 W 4 1 . Mangiarotti, G., and Schlessinger. D . (1967). 1. M o l . Biol. 29, 395-418. Manner, G., Gould, B. S., and Slayter, H. S. (1965). Blorhim. Biophys. Acta 108, 659676.

Mano, Y . , and Nagano, H. (1966). Biochem. Biophys. Res. Commun. 25, 210-215 Manor, H.. and Haselkorn, R. (1967). J . Mol. Biol. 24. 269-288.

2 30

KONRAD KECK

Marbaix, G., Burny, A., Huez, G.. and Chantrenne, H. (1966). Biorhim. Biophys. Arta 114, 4 0 4 4 0 6 . Marchis-Mouren, G., and Cozzone, A. ( 1966). Biochemistry 5, 3684-3690. Marks, P. A., Burka, E. R., and Schlessinger, D. (1962). Proc. Natl. A d . Sri. US.48, 2 163-2 171. Marks, P. A,, Rifkind, R. A., and Danon, D. (1963a). Pror. Natl. Acud. Sri. U S . 50, 33&342. Marks, P. A,, Burka, E. R., Rifkind, R. A,, and Danon, D . (1963b). Cold Spring Harbor Symp. Quant. B i d . 28, 223-226. Marks, P. A,, Burka, E. R., Coconi, F. M . , Perl, W . , and Rifkind, R. A . (1965). Pror. Nut!. Acad. St.;. U S . 53, 1437-1443. Martinez, R. J. (1966). J. Mol. Biol. 17, 10-17. Matsuda, K., and Siegel, A. (1967). Pror-. Natl. Arad. Sci. U.S. 58, 673-680. Merits, I. (1962). Biorhem. Biophys. Res. Commun. 10, 254-259. Meselson, M., Nomura, M., Brenner, S., Davern, C., and Schlessinger, D. (1964). J. Mol. Biol. 9, 696711. Miura, K. I., and Muto, A. (1966). Biorhim. Biophys. Acta 108, 707-709. Mizrahi, I. J. (1965). Biorhim. Biophys. Arta 108, 419-433. Monroy, A,, Maggio, R., and Rinaldi, A. M. (1965). Pro(.. Natl. Acud. Sci. U.S. 54, 107111.

Moore, P. B. (1966). J. Mol. Biol. 18, 38-47. Morris, A., Favelukes, S.. Arlinghaus, R., and Schweet. R. (1962). Biochem. Biophys. Res. Commuri. 7, 3 2 6 3 3 0 . Munro, A . J., and Korner, A. (1964). Nature 201, 1194-1197. Munro, A. J., Jackson, R. J., and Korner, A . (1964). Biorhem. J. 92. 289-299. Nakada, D., and Fan, D. P. (1964). 1. Mol. Biol. 8, 223-230. Nakada, D., and Magasanik, B. (1964). J. Mol. Biol. 8, 105-127. Naono, S., and Gros, F . (1960). Compt. Rend. 250, 3889-3891. Naono, S., Rouviere, J., and Gros, F. (1966). Biorhim. Biophyr. Arta 129, 271-287. Naora, H., Naora, H., and Brachet, J. (1960). J. Gen. Physiol. 43, 1083-1102. Nemer, M. (1962). Biochem. Biophys. Re.r. Commun. 8, 511-515. Nemer, M. (1967). Progr. Nurleic Acid Res. Mol. Biol. 7, 243-301. Nemer, M., and Bard, S. G. (1963). Science 140, 664-666. Nemer, M., and Infante, A. A. (1965). Federation Proc. 24, 283-283. Nirenberg, M. W., and Matthaei, J. H. (1961). Pror. Natl. Acud. Sri. U.S. 47, 1588-1602. Noll, H., Staehelin, T., and Wettstein, F. 0. (1963). Nature 198, 632-638. Nomura, M., Hall, B. D., and Spiegelman, S. (1960). J. Mol. Biol. 2, 306-326. Nygaard, A. P., and Hall, B. D. (1963). Biochrm. Biophys. Rej. Commun. 12, 98-104. Nygaard, A. P., and Hall, B. D. (1964). J. M o l . B i d . 9, 125-142. Papaconstantinou, J. ( 1967). Srierrre 156, 338-346. Papaconstantinou, J., Koehn, P. V., and Stewart, J. A. (1964). A m . Zoologist 4, 321-321. Papaconstantinou, J., Stewart, J. A., and Koehn, P.V. ( 1966). Biorhim. Biophys. Arta 114, 428-430. Pardee, A. B., and Prestidge, L. S. (1961). BioL-him. Biuphys. Arta 49, 77-88. Penman, S., Scherrer, K., Becker, Y.,and Darnell, J. E. (1963). Pror-. Natl. Acad. Sr-i. U.S. 49, 654-662. Peterkofsky, B., and Tomkins, G. M. (1967). J. Mol. B i d . 30, 49-61. Pitot, H. C. (1964). Perrpertives Biol. Med. 8, 50-70.

METABOLISM OF ENUCLEATED CELLS

23 1

Pitot, H. C . , Peraino, C.. Lamar, C . , Jr., and Kennan, A . L. ( 1 9 6 5 ) . Proc. Null. Ai-ad. Sci. U S . 54, 845-851. Plaut, W . ( 1 9 5 8 ) . Exptl. Cell Res. Suppl. 6, 69-77. Plaut, W., and Rustad, R. C. (1957). J. Biophys. Biorhem. Cytol. 3, 625-630. Plaut, W., and Rustad, R. C. (1959). Bioi-him. Biophys. Arta 33, 59-64. Pollock, M. R. (1963). Biorhim. Biophys. Aria 76, 80-93. Prescott, D. M. ( 1 9 5 9 ) . J. Bi0phy.r. Biorhem. Cytol. 6, 203-206. Prescott, D . M. (1960a). Ann. Rev. Physinl. 22, 17-44. Prescott, D. M. (1960b). Exptl. Cell Res. 19, 29-34. Prescott, D . M. (1962). J. Historhem. Cytorhem. 10, 145-153. Rabinowitz, M., Zak, R., Beller, B., Rampersad, O., and Wool, 1. G. ( 1 9 6 4 ) . Pror. Nutl. Acud. Sci. US.52, 1353-1360. Rake, A. V., and Graham, A. F. ( 1 9 6 2 ) . J. Celluluv Comp. Physid. 60, 139-147. Rechcigl, M., and Heston, W . E. (1967). Biorhem. Biophys. Res. Commun. 27, 119-124. Reeder, R., and Bell, E. ( 1 9 6 5 ) . Science 150, 71-72. Reich, E. (1964). Srienre 143, 684-689. Reich, E., and Goldbeg. I. H. ( 1 9 6 4 ) . P r o p . Nuclei‘. Arid Res. Mol. Biol. 3, 183-234. Reich, E., Franklin, R. M., Shatkin. A. J., and Tatum, E. L. (1961). Science 134, 556-557. Revel, M.. and Gros, F. (1967). Biochem. Biophys. Res. Commun. 27, 12-19. Revel, M., and Hiatt, H . H. (1964). Pvor. Nutl. Arud. Sci. U S . 51, 810-818. Revel, M.. Hiatt, H. H., and Revel, J. (1964). Srieni-e 146, 1311-1313. Roodyn, D. B., Suttie, J. W., and Work, T. S. (1962). Bioi-hem. J. 83, 29-40. Rosas del Valle, M. R., and Aronson, A . I. (1962). Biorhem. Biopbys. Res. C‘onzmun. 9, 4 2 1-42 5 . Rowley, P. T., and Morris. J. A. (1967). J. Biol. Chem. 242, 1533-1510. Salas, M., Smith, M. A,, Stanley, W . M., Jr., Whabba, A . J., and Ochoa, S. ( 1 9 6 5 ) . I. Biol. Chem. 240, 3988-3995. Salser, W., Janin, J., and Levinthal, C. (1968). J. Mol. Biol. 31, 237-266. Schaechter, K., Previc. E. P.. and Gillespie, M. E. (1965). J. Mol. Biol. 12, 119-129. Schaechter, M., and McQuillen, K. ( 1 9 6 6 ) . J. Mol. Biol. 22. 223-233. Scherrer, K., Latham, H., and Darnell, J. E. (1963). Pror. Nutl. Acad. Sci. U.S. 49, 240248. Schimke, R. T. (1964). J. Biol. Chem. 239, 3808-3817. Schimke, R. T., Sweeney, E. W., and Berlin, C. M. (1965). J. Biol. Chem. 240, 322-331. Schulman, H . M. (1968). Biorhim. Biophys. Arta 155, 253-261. Schwartz, H . S., Sodergren, J. E., Garofalo, M., and Sternberg, S. S. ( 1 9 6 5 ) . Canrer Res. 25, 307-317. Schweiger, E., and Berger, S. ( 1 9 6 4 ) . Bioi-him. Biophys. Artu 87. 533-535. Schweiger, H. G., Master, R. W. P.. and Werz. G. (1967). Nature 216, 55.4-557. Scott, R. B., and Bell, E. (1965). Science 147, 405-407. Seed, R. W., and Goldberg, I. H . (1963). Proc. Nutl. Arad. Sri. U S . 50, 275-282. Sekiguchi, M., and Cohen, S. S. (1963). J. Biol. Chem. 238, 349-356. Sidransky, H., Staehelin, T., and Verney, E. (1964). Science 146, 7 6 6 7 6 8 . Singer, M. F., and Tolbert, G. (1965). Biochemistry 4, 1319-1330. Slater, D. W., and Spiegelman, S. (1966a). Pror. Null. Acad. Sci. U.S.56, 164-170. Slater, D. W., and Spiegelman, S. (1966b). Biophys. J. 6, 385-404. Soeiro, R., and Amos, H. ( 1 9 6 6 ) . Biochim. Biophys. Actu 129, 4 0 6 4 0 9 . Spahr, P. F. (1964). J. Erol. G e m . 239, 3 7 1 6 3 7 2 6 .

232

KONRAD KECK

Spahr, P. F., and Schlessinger, D. (1963). J. B i d . Cbem. 238, 2251-2253. Spector, A., and Kinoshita, J. A. (1965). Biochim. Biophys. Arta 95, 561-568. Spencer, T., and Harris, H. (1964). Biochem. J. 91, 282-286. Spirin, A. S. (1966). Current Topics Develop. Bid. 1, 1-38. Spirin, A. S., and Nemer, M. (1965). Scierrre 150, 214-217. Spirin, A. S., Belitsina, N . V., and Ajtkhozhin, M . A . (1964). Zh. Obrhrh. Bioi. 25, 321 [English Transl. in Federation Proc. 24, T907-915 (1965) .I Spirin, A. S., Belitsima, N. V., and Lerman, M. I. (1966). J. Mol. B i d . 14, 611-615. Staehelin, T., Wettstein, F. O., and Noll, A. (1963a). Science 140, 180-183. Staehelin, T., Brinton, C. C., Wettstein, F. O., and Noll, H. (1963b). Nature 199, 865870.

Staehelin, T., Wettstein, F. O., Oura, H., and Noll, H. (1964). Nature 201, 264-270. Stewart, G . A., and Farber, E. (1967). Srienre 157, 67-69. Stewart, J. A,, and Papaconstantinou, J. (1967). Pro(. Stutz, E., and Noll, H . (1967). Pror. N d . Arad. Sr Sussman, M. (1965). Biorhem. Bioi1hy.r. Res. Comrrn. 18, 763-767. Sussman, M., and Sussman, R. R. (1965). Biochim. Biophys. Acts 108, 463-673. Svensmark, 0. (1961 ) . Acta Physiol. Scni7d. 52, 267-275. Sypherd, P. S. (1967). J. Mol. B i d . 24, 329-332. Tamaoki, T.. and Mueller, G. C. (1962). Biochem. Bi0phy.r. Rer. Commun. 9, 451-45.i. Terzaghi, B., Okada, Y., Streisinger, G., Emrich, J., Inouye, M., and Tsugita, A . (1966). Pror. Natl. Acud. Sri. US. 56, 500-507. Thach, R. E., Cecere, M. A,, Sundararajan, T. A,, and Doty, P. (1965). Pror. Natl. A c d . Sri. U S . 54. 1167-1173. Tissikres, A., and Hopkins, J. W . (1961). Pror. Ncrtl. Acad. Sri. U.S. 47, 2015-2023. Trakatellis, A. C., Axelrod, A. E., and Montjar, M. (1964). J. Biol. Chem. 239, 42374244.

Trakatellis, A. C., Heinie, E., Montjar, M., Axelrod, A. E.. and Jensen. W . N. (1965a). Arch. Biorhem. Biophys. 112, 89-97. Trakatellis, A. C . , Montjar, M., and Axelrod. A. E. (1965b). Biorbemirtvy 4, 1678-1686. Triplett, E. L., Steens-Lievens, A,, and Baltus, E. (1965). Exgtl. Cell Res. 38, 366-378. Tyler, A. (1962). In "Proceedings of a Conference o n Immuno Reproduction" ( A . Tyler. ed.), pp. 13-15. T h e Population Council, New York. Tyler, A. (1963). A m . Zoologirl 3, 109-126. Tyler, A. (1967). Develop. Biol. Strppl. 1, 170-226. Vanderhaeghe, F. (1954). Biochim. Biophys. Arta 15, 281-287. Villa-Trevino, S., Farber, E., Staehelin, T., Wettstein, F. O., and NOH, H. (1964). J . Biol. Chem. 239, 3826-3833. Volkin, E., and Astrachan, L. (1956). Virology 2, 149-161. Von der Decken, A. (1967). J . Cell Biol. 33, 657-663. Warner, J. R., Knopf, P. M., and Rich, A. (1963). Proc. Nail. Al-ad. Sc-i. U.S. 49, 1 2 2 129.

Watson, J. D. (1963). Scienre 140, 17-26. Webb, T. E., BlobeI. G . , and Potter, V. R. (1966). Cancer R ~ J 26, . 253-257. Wettstein, F. O., Staehelin, T., and Noll, H. (1963). Nature 197, 430-435. Whiteley, A . H., McCarthy, B. J., and Whiteley, H . R. (1966). Pror. Ncrtl. Arad. Sci. US.55, 519-525. Williamson, A. R., and Schwert. R . (1964). Nature 202, 435-437. Wilson, S. H., and Hoagland, M . B. (1967). Biorhem. J. 103, 556-566

METABOLISM OF ENUCLEATED CELLS

233

Wilson, S. H., Hill, H. Z., and Hoagland, M. B. (1967). Biorhem. J. 103, 567-572. Wilt, F. H . (1965). J . Mol. B i d . 12, 331-341. Wilt, F. H., and Hultin, T. (1962). Biorhern. Biophys. Re.r. Commun. 9, 313-317. Woese, C., Naono, S., Soffer, R., and Gros, F. (1963). Biochem. Biophys. Re.r. Comun. 11,435-440. Wolstenholme, D. R., and Plaut, W . (1964). J. Cell Biol. 22, 505-513. Woodward, D.0.. and Munkres, K. D. (1966). Pvor. Natl. Acad. Sri. U.S. 55, 872-880. Yankofsky. S. A., and Spiegelman, S. (1962). P i o r . Natl. Arad. Sci. US. 48, 14661472. Yankofsky, S. A,, and Spiegelman, S. (1963). Pro(.. Nad. Acad. Sri. U S . 49, 538-544. Yoshikawa, M., Fukada, T., and Kawade, Y. (1964). Biorhem. Biophys. Res. Commnun. 15, 22-26. Yoshikawa-Fukada, M., Fukada, T., and Kawade, Y. (1965). Biorhim. Biophys. Aria 103, 383-398. Yudkin, M. D. (1965). Biorhim. Biophys. Arta 103, 705-707. Yudkin, M.D. (1966). Biorhem. 1. 100, 501-506. Zimmerman. E. F. (1963). Biorhem. Biophys. Res. Commun. 11, 301-306. Zimmerman, R. A,, and Levinthal, C . (1967). J . MoI. Biol. 30, 349-370. Zucker, W. V., and Schulman, H. M. ( 1 9 6 8 ) . Ploc. Natl. A(-&. Sci. U.S. 59, 582-589.

This Page Intentionally Left Blank

Stereological Principles for Morphometry in Electron Microscopic Cytology' EWALD R. WEIBEL Department of Anatomy, Unii'ersity of Bern, Bern, Switzerland

I. Introduction . . . . . . . . , . . , . . , . . , , , . . , , . , . . . . . . , , . , , . . .

11.

111.

IV.

V. VI.

VII.

235

A. Purpose and Aims of Morphometric Cytology . . . . . . . . 235 B. The Problem of Measuring Structures on Sections . . . . 236 C. Classification of Structures . . . . . , . . , . . . . . . . . , . 237 Fundamental Stereological Principles . , . . , , . . . . . . . . . . . . 238 A. Basic Parameters Characterizing Structures and Their Correlates on Sections . . . . , , . . , , . . , . . . . . . . . . . . . . . . 238 B. Terminology and Symbolism , . . , . . . . . . . . , . . . 2 40 C. Stereological Assessment of Aggregate Structures . . . . . . 242 D . Characterization of the Size Distribution of Discrete ...................... Particles . . . . . . 257 in Thickness of Sheets . . 261 Application of Stereological Methods to Electron Microscopic Cytology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261 A. Specimen Preparation . . . . . , , . . . . . . . . . . . . . . , . . . . . . 261 B. Sampling of Tissue . . . . . . . , . . , , . . . . . . . . . . . . . . . . . . 263 C. Stereological Analysis of Electron Micrographs . . . . . . . . 273 An Example of Morphometric Characterization of Organelles: The Liver Cell . . . . . . . , . . . . . . . . . . . . . . . . . . . . . 286 A. General Concept of the Study and Sampling Procedures 286 8. Specific Methods for Estimating Morphometric Properties of Cells and Subcellular Components . . . . . . . . . . . . . . . . 287 C. Correlation of Biochemical with Morphometric Data . . . 293 Cytomorphometric Methods in Experimental Pathology . . . . 293 Problems Arising in Applying Stereological Methods to Anisotropic Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 294 A. Sampling from Anisotropic Tissues . . . . . . . . . . . . . . . . 294 B. Effect of Anisotropy on Stereological Measurements; Influence of Test Systems . , . . . . . . . . . . . . . . . . . . . . . . 295 C. Assessment of Structural Anisotropy . . . . . . . . . 291 Appreciation of Present State and Outlook o e Possibilities . . . . , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 298 . _ _ . . . . . . . . . . . . 299 References . . . . . , . . . . . . . . . . . . . . .

I. Introduction

A. PURPOSE AND AIMSOF MORPHOMETRIC CYTOLOGY Advances in electron microscopic cytology have shown the cell to be composed of a limited and well-definable spectrum of organelles, most of which can be 1 The personal work of the author was supported by Grant No. 3952 from the Schweizerischer Nationalfonds zur Forderung der Wissenschaftlichen Forschung, and by Grant No. 78 from BIGA.

235

236

EWALD R. WEIBEL

isolated for biochemical and physiological study, or can be functionally characterized h silu by cytochemical methods and radioautography. Changes in function are frequently associated with an augmentation or reduction of organelles rather than with unambiguous qualitative alterations. This is particularly true if the changes occur within the physiological range. Morphometry of the cell serves the purpose of furnishing quantitative information on cellular fine structure with the aim of allowing quantitative correlation of biochemical or physiological data with morphological data obtained on structurally intact cells. It is thus evident that the morphometric approach to cytology is not a purposc in itself, although admirable dimensional equilibria can be revealed which satisfy our esthetic needs. It is a means that serves the aim of structure-function correlation and derives its justification from the recognition that all orderly function must have an organized structural basis of a size that is adequate but not excessive. In this review, particular emphasis will therefore be placed on the possibilities of applying morphometric methods to correlative cell biology.

B. THEPROBLEM OF MEASURING STRUCTURES ON SECTIONS The cell is a compact array of structures, i.e., three-dimensional objects, which we can resolve only if we have some means of penetrating into the system. If we wish, at least partially, to preserve the relationship between structures we cut the fixed tissue into thin slices; and we know that the best resolution is provided by the thinnest slice or section. An ultrathin section, however, randomly cuts through the solid organelres and presents us with essentially flat profiles. From our experience we are usually able to subjectively interpret these profiles in terms of three-dimensional structures, but we are also aware that this interpretation may be erroneous: A circular profile may, for example, be derived from a spherical, ellipsoidal, conical, or cylindrical structure. A single profile does not, therefore, allow any conclusions about the three-dimensional shape of the organelle unless additional information is available. Likewise, the size of a profile is not representative of the size of the structure from which it arose. Quantitative relations exist, however, between the average dimensions of a large number of organelles and those of their profiles on sections. In this sense, the aggregate of profiles on the unit area of a section is quantitatively representative of the aggregate of organelles contained in the unit volume, so that measurements obtained on sections can be interpreted in terms of structural dimensions by means of stereological relations. This makes stereology2 a very 2 The two terms “morphometry” and “stereolopy” are closely related but are not synonymous (Weibel and Elias, 1967). Morphomrtty implies the use of quantitative data in the description of structural features. Morphornetric data citn be obtained by a variety of measuring procedures performed on any type of specimen, but they can also be derived from stereological analysis of tissue sections. Strreology implies a geometric analysis of

M O R P H O M E T R I C CYTOLOGY

237

attractive tool for cytologists since its methods allow a quantitative analysis of internal cell structure on electron micrographs of tissue sections. It is the purpose of this article to review the stereological methods that can be useful for morphometric cytology and to indicate a few general possibilities for their appl ication.

C. CLASSIFICATION O F STRUCTURES Stereological analysis of tissue sections is based on the assumption that the individual tissue components to be studied are ( I ) present in adequate number; ( 2 ) unambiguously identifiable on sections; and ( 3 ) similar in size and shape from one part of the tissue to the other. The last condition is not essential, however, for a great number of stereological principles. Fortunately, these conditions are met in most organs and cells, and if differences occur they are often of functional significance so that the structures can be grouped into two or more subclasses which are again homogeneous. Nevertheless, it must be borne in mind that all morphometric studies must be preceded by a thorough qualitative evaluation of structures, leading to an unambiguous definition of the structures to be measured. Distinct differences in structure, size, and shape of mitochondria in different parts of the liver lobule have, for example, been demonstrated by Loud ( 1968). In studying liver mitochondria it must therefore be decided whether or not these differences are relevant to the aim of the investigation; if so, the tissue must be sampled accordingly. In recent studies (Weibel et al., 1969; Staubli et al., 1969) this was not considered relevant since the morphomettic data were to be correlated with biochemical information obtained on subcellular fractions of liver homogenates that represented random samples of all cells. A further necessary condition for stereological analysis of tissue sections is random orientation of structures with respect to the section plane. This appears to make stereological analysis of biological tissues impossible since polarity of structures is one of the fundamental principles of biological organization. However, packing of higher-order structural units-such as glandular acini or liver lobules-to form a composite organ results in almost unlimited variation in the orientation of the polarity axis of units with respect to a fixed plane of reference, such as the random section plane. The condition of random orientation of structures with respect to the section plane is thus met. In certain classes of structures inherent anisotropy is not naturally eliminated. Skeletal muscle, peripheral nerve fibers, epidermis, and medulla of kidney are examples. To make stereological analysis applicable, such anisotropic material must be specially treated, as will be outlined in Section VI. structures and textures; it includes methods that allow direct derivation of metric properties uf structures from two-dimensional sections on the basis of geometrico-statistical reasoning.

238

EWALD R. WEIBEL

A further property of the material to be studied stereoIogicaIIy must be defined beforehand since it will determine the measuring procedure to be employed: its size with respect to the field of observation. W e must distinguish objects of finite expanse, i.e., of size smaller than the field of observation, from infinitely large objects. Liver parenchyma can always be regarded to be infinitely large as compared to an electron micrograph, while cytoplasm of individual hepatocytes must be considered a finite structure. Aortic endothelium, on the other hand, is an intermediate case: It is finite in thickness but infinite in width along the vessel wall. These properties must be given careful consideration when setting up the measuring procedure. The classification of structures and its implications for stereological analysis have been extensively treated by Sitte (1967).

11. Fundamental Stereological Principles A. BASICPARAMETERS CHARACTERIZING STRUCTURES AND THEIR CORRELATES ON SECTIONS Every real structure has three-, two-, one-, and zero-dimensional features; these can be described, generally speaking, as volume, surface area, length or diameter, and number. Recently, the generalization of stereological principles by means of integral geometry has led to the introduction of “mean curvature” and “Eulerian characteristic” as very general one- and zero-dimensional parameters (Giger, 1967). In a convex body, the mean curvature is directly related to its average width or mean tangent diameter; it is therefore a one-dimensional parameter which characterizes the size of the body. The Eulerian characteristic is a parameter characterizing the topological properties of the structure; most of the real tissue structures, being simply connected bodies, have an Eulerian characteristic of 1. The Eulerian characteristic is therefore related to the number of structures. For the time being, these unfamiliar parameters are chiefly of theoretical value and will only be mentioned when necessary for the sake of completeness. 1.

Parameters Characterizing Individual 0rganelle.i

Average volumes, surfaces, or diameters of organelles can often be derived from parameters characterizing organelle aggregates. Usually, however, it is necessary to make some assumptions as to their geometric shape. A better characterization is obtained by deriving size distributions; here again the shape of the organelles must be known. 2. Parameters Characterizing Organelle Aggregates

For the majority of propositions, though, it will be sufficient, if not preferable, to determine directly parameters that are related to a well-defined aggregate of

MORPHOMETRIC CYTOLOGY

2 39

organelles. As an example, information on the total volume or total membrane surface of all mitochondria in I u n . 3 of tissue may be more relevant to a study on tissue metabolism than a meticulous derivation of mitochondrial size distribution. Stereological methods allowing estimation of aggregate parameters are exceedingly simple, while determination of size distributions is more involved. Aggregate parameters are always defined with respect to a given containing volume; hence they are concentrations or densitie5. We would thus define the volume of all mitochondria in the unit tissue volume as their voltrme density (V,), their total surface in the unit tissue volume as their snrfclce deitsity (Sv), and so on. It is evident that the total mitochondria1 volume of a liver, for example, can be directly calculated from V , if the liver volume is known. 3 . Parameters of Section Profiles

If we examine more closely the effect of sectioning upon the structures, we observe that the dimensions of the profiles bear a direct relationship to those of the structures (Fig. 1 ) : The volume V of the structure is related to the area A V

+

A

s

-

0

M

-

Q

N

FIG. 1. Appearance of three-dimensional structures on two-dimensional section (from Weibel, 1967a).

of the profile, and the surface S of the structure is related to the profile boundary length B . A linear tissue element of length M will appear as a point Q on the section. W e can derive as a fundamental rule that sectioning reduces the dimensions do of structures by 1, thus generating profiles or traces of dimension d, = do - 1. From this we must further conclude that “point objects” are lost, i.e., that they can not be represented on a true two-dimensional section since their dimension is do = 0.

240

EWALD R. WEIBEL

If we consider the section as a planar probe that intersects the tissue, we can generalize these observations and derive rules which are of fundamental importance for the theory of stereology introduced in the following discussion (Weibel, 1967a). W e will then see that linear probes placed into the tissue intersect the structures and bear traces of dimension do - 2. The surface of the structure intersects the probe at a point I while its volume is represented by the length L of the line included in the structure. In generalized form we can state that probes of dimension d,, will generate profiles of dimension

+

d, = do d,, - 3 (1) There are two consequences to be derived from this law (Fig. 2a) : (1) The dimension of the trace decreases with the dimension of the probe and becomes 0 when d, = 3 - do (2) If the dimension of the probe is smaller there are no traces, i.e., the object is lost. ( 2 ) There is always one probe where d, = 0 and this is convenient because P R O B E ?

VOLUHE

w

LENGTH T R A C E S

3

U

NUnBER v)

a

b

FIG. 2. a, Dimensional relation between structures, probes, and traces of structures on probes, together with systematic symbolism; b, System of basic stereological parameters in relation to dimensions.

in this case “measurement” of the traces is reduced to simple counting. For all higher dimensions of dt actual measurement, linear, planimetric, or volumetric, is necessary. W e shall make ample use of this consequence in the following discussion. B. TERMINOLOGY AND SYMBOLISM In recent years there have been attempts to design a basic set of symbols for stereology which is at once simple, self-explanatory, and unambiguous (Underwood, 1964; Weibel et ul., 1966; Weibel and Elias, 1967; Elias, 1967). General agreement has been reached on the use of a double symbol for relative dimen-

241

MORPHOMETRIC CYTOLOGY

sions: The first capital letter defines the pawmeter and the second capital letter, mostly written as subscript, the reference system. For example, numerical density of a given structure in the unit volume is written as

N N,,=-

(3)

V

The original list of symbols was purposely restricted to a few self-explanatory letters (Underwood, 1964). Practical application, however, has shown that this restriction could cause confusion. W e have therefore found it necessary to enlarge the set of symbols to exclude ambiguities. Table I lists the symbols that LIST

OF

TABLE I BASIC SYMBOLS FOR STEREOLOGY~

~

Symbol

Definition

Dimension

V

Volume of structure or test volume

cm.3

S

Surface of structure Area of section (planar)

cm.2

A

M

Mean curvature or length of structure Boundary length of profile

Clll.:!

cm. cm. cni.

B L

Length of test line

N

Number of structures

cm.0

Q

tm.0

I

Number of transsection points on section Number of intersection points

P

Number of (test) points

cni.0

D

cm.

T

Characteristic linear dimension of structure (diameter, thickness, etc.) Section thickness

V"

Volume density of structures in tissue

cm.O

Surface density of structures in tissue Density of mean curvature or length in tissue volume

cm.--l

SV

MV

N" B.4 I, a

Numerical density of structures in tissue Density of boundary length on section area Density of intersections on test line length

cm.0

cm.

cm-2 cm.-:3 cin.-l

an.--'

Compare Fig. 2 . (Modified from Weibel and Elias, 1967.)

will be used throughout this review; their relation to the dimensions of objects, probes, and traces is shown in Fig. 2. In practical application, it is further necessary to identify the component to which the parameter refers. This has been done by indexing the stereological

242

EWALD R. W E I B E L

symbol with numbers, Greek letters, or lowercase roman letters. We prefer the latter since it again allows self-explanatory indexing. The use of computers in data analysis sets some restrictions on symbolism since it utilizes only roman capitals and arabic numerals. W e have found it no inconvenience to use a simple sequence of three or more capital letters for unambiguous identification of parameters : The first letter identifies the parameter, the second the reference system, and the rest the component. NVMI would thus mean: “numerical density of mitchondria in tissue volume.” As an alternative, the component index can be enclosed in parentheses: NV(M1). C. STEREOLOGICAL ASSESSMENT OF AGGREGATE STRUCTURES I. Relationship between Stractural Parameters and Profiles o n Sections

Let us assume a model tissue that contains a large number of similar bodies. These can be characterized by their volume and surface, or rather by their volume and surface density in the tissue, I/, and Sv, if we consider them as an aggregate. The question now arises how these parameters are represented on a random section through the tissue. In 1847, the French geologist Delesse developed the fundamental relation, often called the Delesse principle, that the areal density of profiles on the section A , is, on the average, equal to Vr7.This equality A, = V v is now well known; it has been derived empirically by Delesse but its proof has been given explicitly many times since (Smith and Guttman, 1953; Weibel, 1963; Underwood, 1967a; and others) so that it need not be repeated here. The surface of the structures appears on the section as a contour or border of profiles. It had first been shown by Saltykov (1958) that the density of profile border length on the section area L 3 , is directly proportional to Sir, namely B.4

JL

-

4

S,

(4 1

It is noted that both BA and Srrhave the dimension cm.-’. We have noted that linear features of the structures appear on the section as intersection points. Density of mean curvature is a linear parameter characterizing a vast class of structures, but for general structures its meaning is quite abstract. W e must therefore treat a special case: “linear” elements curving through space. These can be thin filamentous structures, the edges of polyhedral bodies, or the axes of tubular structures, and so on. Their total length in the unit volume MIis directly proportional to the density of intersection points on the unit section area Q A (Smith and Guttman, 1953; Saltykov, 1958) : 1

MORPHOMETRIC CYTOLOGY

243

Aggregates of discrete objects are further characterized by the number of structures per unit volume, which is called their numerical density N V . Being related to the Eulerian characteristic, as outlined in Section II,A, the number of structures is a zero-dimensional property. Each structure therefore has to be represented by one characteristic point such as, for example, its center of gravity. In Section I,B we have seen that “point objects” of space are not represented on ideal two-dimensional random sections. As a consequence, there exist no direct and simple relations between N V and the number of profiles N, observed on the unit area of an ideal section; N A in fact depends on Nv,on the shape of the structures, and on their size. Ideally, N, can therefore only be derived from three-dimensional tissue samples, i.e., from slices of known thickness. On nearly two-dimensional sections, a relationship between Nr and the profile density can only be established if some assumptions as to shape and size of the structures can be introduced. The same restriction applies to attempts to characterize size distribution of structures in which shape and number are implicitly involved. 2. MetbodJ f o r EJtinzatiug Voliimej

The volume density V Fof structures can be estimated directly from the Delesse relation V v = A , by planimetry of the combined area A , of all profiles, which is then divided by the enclosing test area A , . Delesse traced the profiles on heavy paper, cut them out and weighed them; this method is still practiced occasionally. Alternatively, A , can be measured by means of a polar planimeter, but this is rather cumbersome and may be affected by considerable error depending on the shape and size of the profiles. Rosiwal (1898) showed that A, could be obtained by so-called linear integration: If test lines of known length L are randomly placed on the section, it suffices to measure the fraction L, of these lines enclosed in the profiles by simple linear measurement to arrive directly at the desired estimate

This method is much more efficient than planimetry and yields very reliable results if sufficiently dense lines are used. It is advocated by Loud (1962, 1968; Loud et al., 1965) for volumetric analysis of cellular composition on electron micrographs. Automatic scanning devices, such as those developed by Lazarow and Carpenter (1962 ; Clawson ef al., 1958 ; Carpenter and Latarow, 1966) or the Quantimet (see Section III,C,6), make use of this method. A disadvantage of this method is the necessity for measuring intercepts within profiles which is still arduous if no automatic recording devices are available. It was therefore a great advance when the Russian geologist Glagoleff demonstrated in 1933 that A, could be estimated by superimposing a regular point

244

EWALD R. WEIBEL

lattice on the section and determining, by simple count, the fraction PI, of all points enclosed in profiles; this yielded a third relation

V" = P,.

(7)

The theoretical basis of this method had bten worked out by Blichfeldt (1914) ; it is actually quite plausible: Planimetry can be performed by dividing the entire section plane into small squares of unit edge length d (Fig. 3) and determining the number of squares enclosed in the profile. The squares cut by the profile border have to be counted as fractions in proportion to the fraction covered by the profile. Any degree of precision can be obtained by making the squares small enough. The problem of rounding off can be made easier if the center point of O

O

O

FIG. 3.

O

O

O

O

D

O

O

Point-counting planimetry.

each square is marked: If the center point is inside the profile the square is counted full, if i t is outside it is rejected. This, however, leads directly to pointcount planimetry if the squares are replaced by the lattice of their center points. It is obvious that the point marked need not be the center point but may well be one of the corners of the squares. Point-counting volumetry can thus be performed with the cross-points of a square grid as markers (Fig. 4 ) . As early as 1943, Chalkley, not knowing of Glagoleff's work, proposed an analogous method for volumetric analysis of tissue sections. He used random points, irregularly arranged, to arrive at the same result. Indeed, the regular array of test points is of no theoretical importance although it is convenient in practice, and may often yield smaller statistical errors (Hilliard and Cahn, 1961; Hally, 1964; Hennig, 1967). Use of these methods in light microscopic histology was also advocated by Attardi (1953), Eranko (1955), and Hennig (1956).

MORPHOMETRIC CYTOLOGY

245

In summary, we note that three related methods are available for estimation of

vJ7

v,,

A , = L[, = P ,

(8)

The question arises as to which is the best. It immediately appears that precision in estimating the size of the individual profile decreases from areal to pointcount analysis. However, Hilliard and Cahn (1961 ) have established the apparent paradox that volume densities obtained by point counting are affected by smaller overall error than those obtained by areal analysis. With the lattice of test points properly set up (cf. Section III,B,3), only a part of the profiles will contain a test point. Hence, the number of profiles lying under the entire test system is larger than the number of occupied lattice points. For a given number

FIG.4 . Differential point ( P ) and intersection ( I ) counting for estimation of fractional area (volume) and boundaries (interfaces) of composite tissue section. of observations the point count thus encompasses a larger fraction of section area than areal analysis. The resulting improvement in sampling variance (cf. Section II1,B) more than compensates for the apparent inaccuracy of the individual measurement. In addition, point counting is many times more eficient than both areal and Iinear analysis. Real tissues are made up of a compxct array of different components or phases which we may label a, b, c, etc. (Fig. 4). The sum of the partial volume densities VVa,VVb,Vve, and so forth, is by definition equal to the unit volume. To determine the relative volumes of these component phases it suffices to overlay a lattice of test points on a section and to differentially count and classify the points lying on profiles o f a, b, c, and so forth. In a three-phase system we obtain from the primary point counts Pa, PI,, and P,, the respective volume densities as

246

EWALD R. WEIBEL

3. Methods for Estimating Surface Areas

On sections, the surface density Sv of tissue structures is quantitatively represented by the contour length density B A of the profile borders as shown in Eq. (4).It thus suffices to measure BA by means of a map-measuring device or a thread. This is inefficient, however, and not very accurate either. Smith and Guttman (1953) and Saltykov (1958) have shown that B A can be estimated by placing a grid of test lines on the section and counting the intersections formed by these lines with the profile border, ZL being the number of intersections per unit length of test lines. The relationship is

It derives directly from the well-known Buffon needle problem (Buffon, 1777). Since the derivation of this formula is typical of stereological problems involving geometrical probability and is, nevertheless, easy to follow, it may be appropriate to present it here. A curve of length B drawn on a plane surface can be deliberately divided into n short straight segments of equal length I (Fig. 5). If we superimpose a grid of parallel lines of distance d > I on the curve, a number Z < n of the segments will intersect the grid lines. The problem is to derive the probability p = Z/n that a segment will intersect a grid line as a function of I and d. This problem was put before the French Academy by Buffon in 1777. The solution is as follows:

247

MORPHOMETRIC CYTOLOGY

If all segments are perpendicular to the grid lines, parallel, p 10. If the segments are oriented at an angle then x 1 p ( 0 ) = - = - x sin e d d

p = l / d ; if e to the grid

they are (Fig. 5 )

If we allow all angles 0 to occur with equal probability then we must allow the segments to rotate through one quadrant of a circle and must integrate p ( e ) over the range 0 < 6 < n/2 and divide by the upper limit to obtain an average probability:

The integral of Eq. ( 1 2 ) being 1 we obtain the desired result 1

P = - = 12= -

2x1

7lXd

Going back to the original situation we note that the length of our curve was B = n x 1 and obtain by rearrangement of Eq. (1 3 ) 51:

B=-XIXd 2

If both the curve and the test grid are enclosed in an area A (Fig. 5), the length of the test lines is L = A/d. Substituting this for d in Eq. (14) yields the relationship between the curve length density B, and the number of intersections per unit test line length of Eq. (10)

The method thus derived provides an efficient means for determining the length of profile borders. However, we are usually more interested in obtaining directly the surface density of the tissue structures. It is immediately seen that substitution of Eq. (10) into Eq. (4) yields a direct relationship between Sv and I L : sv = 2 11. (15)

x

This simple formula is one of the fundamental stereological principles. It has been derived independently at least six times by different authors (Tomkeieff, 1945;Smith and Guttman, 1953;Duffin et a/., 1953;Horrikawa, 1953;Hennig,

248

BWALD R. WEIBEL

1956; Saltykov, 1958). The decade encompassed by these articles also marks the period during which awareness of the necessity for refined methods of morphological measurement on sections became acute, particularly in the material sciences. Within this period, a related principle by Chalkley et al. (1949), which will be presented later, was developed in the context of biological work. One wonders why it took more than a decade for the potential of these simple methods to be recognized, while 2 0 years later they are still not extensively used in biology. We should dwell a little more on the meaning of S., It is a two-dimensional feature and actually measures the interface area between two adjacent component phases. When we talk about surface of mitochondria we actually mean “contact surface between mitochondria and cytoplasmic matrix.” This notion of interface area is of functional significance, since interfaces are at once boundaries and exchange regions between compartments. In the three-phase system a, b, c, discussed earlier (Fig. 4 ) we can observe different phase contacts: a-h, a-c, b-c, a-a, b-b, c-c. The relative extent of these interfaces can be easily assessed if the intersection points with random test lines are differentially counted : I,,.,,, I,,.,., I,,,., etc. To simplify notation we will, in general, use only one subscript when we determine the surface of a given organelle without regard for the adjoining component phase. In cytology, the interfaces between cell components are usually marked by membranes so that we can estimate their “membrane surface density” by counting intersections between membrane traces and test lines. However, we should bear in mind that membranes are sheets of finite thickness rather than true surfaces and have, consequently, two (active) surfaces. This may sometimes be of importance.

4. Methods

foY

Assessiug Litiear Parameters

We have seen earlier that curved lines in space or filamentous tissue structures intersect a random section plane in proportion to their line density M , . From Eq. ( 5 ) we directly derive the relation M V

=2

x Q.4

which allows estimation of Mv by counting the number of intersections Q, of the filamentous structures with the unit section area. W e note correspondence of this formula with Eq. ( 1 5 ) for deriving surface density from intersections with test lines. In f x t , these relations are identical since these basic stereological principles are independent of shape and arrangement of structure and test probe as long as structure and probe remain stochastically independent. This method was recently used by Haug ( 1 9 6 7 ~ )to estimate the length of nerve fiber segments from electron micrographs.

MORPHOMETRIC CYTOLOGY

249

A further parameter which can often be useful in describing one-dimensional properties of solid bodies is their average volume-to-surface ratio (v/sj as introduced by Chalkley et al. ( 1949) and Cornfield and Chalkley (1951 ). The ratio ( v / s ) can be determined by a simple point-counting method, which is in essence a combination of point-counting volumetry and surface estimation by intersection count. To facilitate this measurement, a group of short straight test lines of equal length is placed over the section (Fig. 6); each line is characterized by its length z and by two end points. It is evident that the end points can be used as markers for point-counting volumetry and the straight line segment

FIG. 6 . Estimation of volume-to-surface ratio according to Eq. (17) with test lines of length z.

between them for surface intersection counts. According to Chalkley and Cornfield the volume-to-surface ratio is obtained by

where P, is the number of end points enclosed in profiles of structure i, and Ii the number of intersections of test lines with the boundary of these profiles. It should be noted that Eq. (17) can be easily derived from Eqs. ( 7 and 1 5 ) ; if there are R test lines of length z deposited on the section, then the total number of test points in Eq. ( 7 j is PTv= 2 ~ and , the total test line length is LT r g z . Substitution of these two values and division of Eq. (7) by Eq. (15) yields the result of Eq. ( 1 7 ) . This also shows that P / S can be calculated from values of I/,, and Sc- which have been determined independently on the same sections. The short-line test system of Chalkley (Fig. 6) is thus not essential for estimating v / s although it proves to be convenient in practice,

250

The geometric meaning of spheres it is found that

EWA1.D R. WEIBEL

Z~/J

depends on the shape of the structures. For U/J

= 2r/6

(18)

= d/6

(19)

and for cubes, interestingly enough, V/J

It must, however, be noted that the diameter 2r of spheres and the edge length d of cubes do not have the same geometric meaning. Nevertheless, it is evident that U/J can give information on specific average linear dimensions of structures if some assumptions about their shape can be made. Two further useful examples: for long cylindrical ;tructures of radius r V/J

= V/2

V/J

= d/2

and for broad sheets

-

where 2 is the arithmetic mean thickness of the sheet (Weibel and Knight, 1964). The last relation is useful in estimating the mean thickness of cellular or cytoplasmic layers (Weibel and Knight, 1964) or of endoplasmic reticulum cisternae (Weibel et ul., 1969). It should, however, be borne in mind that 5 may not be a meaningful parameter if transport phenomena across tissue sheets are studied; for this purpose the harmonic mean thickness, described in Section II,E is more appropriate (Weibel and Knight, 1964). Another useful one-dimensional parameter of structures is their "mean linear intercept length" G, as defined by Underwood (1967a).3 It is defined as the mean length of segments of random lines traversing the structures (Fig. 7 ) , is independent of their shape, and can be estimated by stereological methods on random sections by the following principle. In Section II,C,2, it has been shown that a fraction L L = V v of a random test line passes through profiles. LI, is the sum of all intercept lengths on the unit test line. Consequently, their mean is easily defined as

L, = LL/NL

(22)

where N , is the number of profiles intercepted by the test line. On the other hand, we have defined lLas the number of intersections of the profile boundary with the test line of unit length. It is readily seen that I L = 2 N L since there are always two boundary intersections per intercept. Hence we obtain

G =2(LL/IL)

(23)

3 The subscript 3 specifies that the test line intercept with respect to the three-dimensional structure is considered.

25 1

MORPHOMETRIC CYTOLOGY

I

I( 1 1-

FIG. 7.

I

I

J!

Relation between linear intercept L, and surface intersections I (Eq. 2 3 ) .

From Eq. (8) it follows that L , = P,. Substituting PI, into Eq. ( 2 3 ) , and using a short-line test system as in Fig. 6, we arrive at the simple relation

The mean linear intercept length is thus directly related to the volume-to-surface ratio. 5 . MethodJ fov EJtimatiiig the Number o f StmrtzireJ

On the basis of theoretical arguments it was concluded in Section I,C,1 that the numerical density of structures N , cmnot be directly assessed from true section analysis through a simple relationship. The average number of profiles per unit section area N , depends not only on N,., but also on the shape and size of the structures. For randomly oriented particulate structures, De Hoff and Rhines (1961) have shown that N , zN A / B (25) where is defined as the average tangent (or caliper) diameter. If the shape of all particles is the same, depends on their size, or rather on their size distribution. Methods for determining size distributions of spherical and spheroid particles are discussed in Section II,D,2. If it is justified to assume approximately equal size and shape for all particles, a theoretically calculated value of 6 can be introduced into Eq. ( 2 5 ) . Hilliard (1967b) gives formulas for calculation of D for various geometric shapes, and De Hoff (1964) has worked out a plot from which D can be read off for spheroids and cylinders of various axial ratios. It should furthermore be pointed out that a direct relationship exists between D and V/J for various shapes. For spherical particles

a

252

EWALD R. WEIBEI. -

D, z ~ ( v / J )

and for cubic particles

0,= 9 ( V / J ) The parameter (v/J) can be estimated by a simple counting procedure through Eq. (1 7 ) , but this is not an unbiased estimate if the size and shape of the structures vary. However, for many practical purposes it may still be adequate. Aherne (1967) has developed a simple counting procedure. Introducing a shape-dependent coefficient k =V')/:~/J. he finds the number of structures as

N =v x

kn/(V/J)"

(28)

A similar method for determining the number of cylindrical structures N,, has been proposed by Loud et ul. (1965). Both methods are interesting since they derive N Y without an actual counting of profiles of the structures. An alternative method for determining N, has been developed by Weibel and Gomez (1962) and Knight et ul. (1963)

where the coefficient f3 relates to shape and K to the size distribution. Figure 8 presents a plot of as a function of axial ratios for spheroids and cylinders; for spheres (3 = V G . The coefficient K is defined as

FIG. 8 . Shape coefficient

for ellipsoids and cylinders as function of axial ratio h

MORPHOMCTRIC CYTOLOGY

253

where D, and D , are the first and third moment of the size distribution; it is thus always larger than 1, except when all structures are of equal size. For a normal distribution with a coefficient of variation of + 2 5 % of the mean, K = 1.07. It has been shown previously (Knight et a/., 1963; Weibel et a/., 1966) that for biological objects K rarely exceeds 1.1 but will most often be in the range of 1.01-1.1. For many practical purposes, ‘I roughly estimated value of K can be introduced, particularly when comparisons between control and experimental data are sought. A new method of particle counting has been recently proposed by Hilliard (1967a). The section is scanned along a test line length L,. A very short segment of length AL is mxked and displaced along the scanning line (Fig. 9 ) .

L,

FIG. 9. Counting of spherical particles on sections from number of linear intercepts AL with profiles according to Eq. ( 3 1 ) .

<

The number AN of intercept lengths of the scanning line with particle profiles, which are smaller than AL, is counted. For spherical particles we find N V by

The practical application of this method is not easy but it certainly has advantages, particularly when used in connection with automatic or semiautomatic scanning devices. All the counting methods outlined so far presuppose true sections of no thickness. A number of alternative methods are applicable to tissue slices of known thickness T . These have recently been reviewed by Haug (1967a,b). T h e nurnbers of profiles counted per test area A are here to be regarded as “profile” numbers enclosed in a spice of volume I/ = A T . An alternative new method for counting in sections of finite thickness has been proposed by Aherne (1967). In summary, we observe that a great variety of methods has been proposed for estimating the number of structures per unit volume of tissue. All depend on assumptions with regard to shape and size distribution of the structures; and

x

2 54

EWALD R. WEIBEL

all, consequently, will be inexact in practical application. However, if one remains aware that these methods yield approximations, one or another of them will prove useful. The choice of the method depends on the actual conditions and is, moreover, largely a matter of taste. 6 . A Coherent S p t e m of Stereological Form~~lus; Choice of the Method

From the preceding discussion a coherent system of basic stereological formulas has evolved which can be summarized as follows:

Vv T

A , = L , = Pp

(32)

It is noted that this system of formulas is intimately related to Fig. 2 in which the dimension of “traces” of structures on probes of dimension d, has been analyzed. Accordingly, the List term on each line has dimension 0 and can thus be determined by simple counting with respect to an appropriate system of reference. In the first line this system of reference is the total number of test points used, in the second line the length of a test line, in the third the size of the test area, and in the last a test volume. The intermediate formulas require a linear or planimetric measurement to be made. This conclusion influences the selection of appropriate methods for practical stereological work and, hence, the design of useful test systems. Whenever possible, we shall use counting procedures since these are much more efficient than measurements. This system of formulas is correct for the class of aggregates of structures that can be considered very large with respect to the size of the test section. Recently, an equally coherent system of stereological equations, which is valid for a class of very general structures in three-dimensional space with no restrictions necessary, has been developed by Giger (1967) through methods of integral geometry. This system of formulas contains the one presented here as a special case. 7. T h e E f e c t of FAiite Section ThickiZeJJ

The aforementioned stereulogical princples have been based on the assumption that the tissue sections investigated are true two-dimensional sections and hence have no thickness. This condition is actually fulfilled only when polished surfaces of transections through opaque materials, such as rocks, are examined with in-

255

MORPHOMETRIC CYTOLOGY

cident light. In contrast, so-called “ultrathin” sections of plastic embedded tissue examined by electron transmission microscopy are always slices that have a real but finite thickness T > 0, and we can only demand that T + 0 , or in words, that T is reduced as far as technically possible (see Section 111,A). What then is the effect of the residual section thickness on stereological parameters measured by the aforementioned methods ? In essence, opaque structures will be overestimated at the expense of translucent components. The slice of tissue examined is made up of slices of the individual structures. The light or electron beam projects these slices onto the plane of observation (Fig. l o ) . It is evident that the contour of the projection image

< .... , . . .: .p . . ... .. i :,..... ; ........ -

T

Dri<

FIG. 10. Effect of section thickness T and particle diamrter D on extent of projection image, the so-called Holmes effect (from Weibel and Elias, 1967).

of an opaque structure will represent the widest cross-section of its slice contained in the section. A certain fractiun of adjoining translucent material will therefore be covered up. Consequently, the “profiles” of opaque structures will appear too large, and those of translucent structures too small. The degree of overestimation o f opaque structures evidently depends on their curvature, measured by their ratfius of curvature R = ‘/ZD, and on the section thickness. Figure 10 illustrates that spherical opaque structures occupy a much D . It follows larger fraction of a projected section image if T = D than if T therefore that the apparent fraction of a section area covered by profiles of opaque structures is

<

VV” x K,(D,T)

(33)

where the correction coefficient K O is larger than 1 and depends on T and on a characteristic diameter D of the opaque structures. This effect of section thickness

256

EWALD R. WEIBEL

was first recognized by Holmes (1927) and is therefore often referred to as the Holmes effect. T h e value of K , also depends on the shape of the structures. If these can be approximated by spheres of average diameter D, it was found by Holmes (1927) and Hennig (1957) that

<

It is observed that K,, + 1 and becomes negligible when T D. If T / D = 0.1, the coefficient K,, x 1.15, i.e., the uncorrected volumetric density of opaque structures, as obtained by point counting or planimetry, overestimates the true value by 15:h. This error falls to 5%) when T / D = 0.03. I t is a matter of judgment at what point the systematic error resulting from the Holmes effect can be disregarded. It should be noted that, in practice, part of this effect is compensated for by a related but inverse effect; namely, by the failure to recognize thin polar sections of the structures or the thin margin of profiles if the “translucent’’ phase has a certain opacity itself, so that contrast becomes insufficient for distinction of the structures. This is particularly relevant for electron microscopy in which recognition of structures essentially depends on contrast. T h e loss of profiles as a result of contrast deficiency tends to increase with section thickness, at least as long as T < D. I n practice, it may be appropriate to introduce a Holmes correction when T D / l O . The volume density of opaque structures is then

>

and that of the translucent phase

to make

v1.0 + Vvr = 1

If the opaque structures are somewhat granular, the correction coefficient for spheres given in Eq. (34) can be used in first approximation. This is good enough in most cases, since often only crude estimates of T and D are available. T h e general cdse of projection image malysis relating to the study of thick tissue slices has been extensively treated by Underwood (1967b). A detailed review of this work, which provides valuable information for numerous special cases, would extend beyond the scope of this article. Some practical applications will be discussed below.

257

M O R P H O M E T R I C CYTOLOGY

D.

CHARACTERIZATION O F T H E SJZE D I S T R I B U T I O N OF

DJSCRETE PARTICLES

It may sometimes be necessary to estimate the variation in size of some particulate structures. For example, the size distribution of cell nuclei may give some indication of the frequency of diploidy, tetraploidy, and so forth and may be of relevance in studies on cell kinetics (Heiniger et ul., 1967; and others). Or, as another example, the size distribution of lipid droplets, ph'igolysosomes, and so forth, may give some insights into pathological cellular changes. The basic problem related to the characterization of a size distribution of psrticles from a study on thin sections mdy be exposed by first studying a model situation (Fig. 11) : If s population of spheres of equal diameter 2R is randomly P

0.4

0.2

r

a

b

R

FIG. 11. Profile size distribution ( r ) resulting from sectioning spheres of equal radius R.

sectioned, circular profiles of varying diameter 2~ appear on the section. The profile radius r depends on the distance of the sphere center from the plane of section. The largest circles will have Y = R; as the section approaches the tangent plane to the sphere r - 0 . Figure 1 l b shows the relative frequencies of section radii to be expected in this model. Elias and Pauly (1966) have pointed out that 86.4% of profile diameters will be larger and 13.6% smaller than onehalf the sphere diameter. If the model contains three size classes of spheres, each class will give rise to a size distribution of profiles similar to that of Fig. 1 Ib, but these will be superimposed to form a compound size distribution (Fig. 12). If we have determined the size distribution of profiles on sections, we ate hence faced with the difficult problem of unraveling the particle size distribution from it. This problem has been given extensive theoretical treatment (Wicksell, 1925; Lenz, 1955; Bach, 1963, 1967; Saltykov, 1967; Hilliard, 1967a; Bockstiegel, 1967; Giger and Riedwyl, 1969). Examples of practical application of these methods to electron microscopic cytology are given by Baudhuin and Berthet (lc)67), Heiniger rt aL. (1967), and Coupland (1968). A review of all these papers would greatly

2 58

EWALD R. WEIBEL

exceed the space available but should be treated separately. We shall restrict ourselves to indicating some practical methods. In all these methods three basic conditions must be met, unless they are not applicable: (1) The particles must all be of the same shape and may vary only in size (similarity condition); ( 2 ) their shape must be known; ( 3 ) their shape

a

R

0.20

0.10

+

b

r

FIG. 12. Profile size distributions (b) generated by three size classes of spheres ( a ) . Heavy line with open circles marks compound distribution of profile radii resulting from section of the mixture of sphere sizes.

must be such that a random plane can intersect each particle only once, thus forming only one profile. Furthermore, the usual conditions for the application of stereological methods must be met: i.e., the particles must be randomly oriented with respect to the section plane, and they must be evenly dispersed to such an extent that representative samples can be defined. There are basically two ways to proceed: (1) The complete size distribution

MORPHOMETRIC CYTOLOGY

2 59

of particles can be directly deduced from the measured distribution of profile sizes. ( 2 ) Some parameters characterizing the particle size distribution can be computed on the basis of theoretically derived formulas from parameters of the profile distribution. I . Direct Derivation o f Size Distribiitions

When given a measured distribution of profile radii from spherical structures it is evident that the class of largest profiles represents equatorial sections through the largest particles. According to Fig. 1I b we can estimate the number of profiles contributed to all smaller classes by these spheres (Fig. 1 2 ) ; these can be subtracted from the second largest class, and so on, until the profile distribution is “used up.’’ The first method of this type was proposed by Wicksell (1925) ; it has recently been applied by Baudhuin and Berthet (1967) in characterizing the size distribution of mitochondria in subcellular fractions of liver. It should be noted that small profiles are usually missed to an increasing degree the smaller they are, mainly because of the lack in contrast of polar sections (see Section II,C,6). Very often no profiles are recorded in the smallest size classes. The profile size distribution determined is therefore incomplete in the small size classes and must be corrected by extrapolation toward zero. Methods similar to that of Wicksell have been proposed by Schwartz (1934), Scheil (1935), Elias and Hennig (1967), and others. They differ mainly in the procedures used for correction of small profile loss and in the computational approach. The use of appropriate computer programs (Baudhuin and Berthet, 1967) greatly simplifies the application of these methods. A new method by Saltykov (1967) should be pointed out. The profile diameters are classified on a logarithmic scale and expressed as profile area divided by the area of the largest profile (a/n,,,,). Twelve size classes are usually sufficient. The computational procedure of this method is very simple. If the particles are not spherical, much greater problems arise. Saltykov (1967) has pointed out that the frequency distribution of profile areas of polyhedra differs considerably from that of spheres, as exemplified by that for cubes in Fig. 13, in which the largest class has the lowest frequency. Wicksell (1926) has given a method that applies for ellipsoids. 2. Derivution of Parameterr of Size Di.rt~ibi/tio~

It may often be sufficient to characterize the size distribution of particles by determining its moments. It is impossible to review, in this article, all the methods developed for this purpose; we shall rather restrict ourselves to a general indication of the possibilities. Giger and Riedwyl (1969) have developed a semigraphical method for deriving from the distribution of profile diameters the mean diameter D and the

260

EWALD R. WElBEL

standard deviation 0 of a normally distributed population of (nearly) spherical particles if their size can be assumed to be approximately normally distributed. D can be calculated from the mean profile diameter 2 through the basic relationship

if all particles have the same chance of being cut by a random section. In a randomly dispersed population of spheres of varying size the chance of being cut, however, is proportional to size. The mean diameter calculated by Eq. (37)

A’AIn,,

FIG. 13. Profile density per unit section area ( N , ) as function of relative profile area (,4/,4,,,ax) resulting from random wctioning of cubes (from Saltykov, 1967).

will hence be an overestimate of true mean diameter, the degree of overestimation depending on ci. Giger and Reidwyl (1969) have worked out an easy graphical method for estimating both ci and the correction factor to be applied to the mean diameter calculated according to E l . (37). They are estimated from the area under the profile size distribution curve to the right of D. Bach (1963, 1967) has proposed a rather involved method which allows derivation of various moments of size distribution of particles from measurements of profiles on sections of finite thickness. Practical application of this method, which has fewer restrictions imposed and yields more information than the above-mentioned, requires use of a computer. Instead of using measurements of profile diameters, Hilliard (1967a) and Bockstiegel ( t 967) derived particle size distributions from the distribution of intercept lengths of profiles with random linear probes. These methods have the advantage of being directly applicable to automatic linear scanning analysis if the specimen allows this (cf. Section TII,C,6). Compared to direct measurement of profile diameter these methods forgo a considerable amount of information; this

MORPHOMETRIC CYTOLOGY

261

is only justified if the structures occur in great number, as may be the case in studying mitochondria.

E. CHARACTERIZATION O F VARIATION I N THICKNESS O F SHEETS A special case of size distribution is given in the variation in thickness of a tissue sheet; this may be relevant in the study of transfer of material across a cellular barrier, and so forth. If passive transfer by diffusion can be assumed, the flow of material will be inversely proportional to the local barrier thickness. The average diffusion resistance is hence related to the mean reciprocal thickness or harmonic mean thickness D,, of the barrier. It has been shown (Weibel and Knight, 1964) that D,, can be estimated from the distribution of mean linear intercepts L, of random lines with the barrier (cf. Section I I , C , 4 ) through

=t(+) 1

D,,

where

is the reciprocal of the harmonic mean of L,. The harmonic mean D,, is the (-1) moment of the size distribution while the arithmetic mean thickness D is the first moment and follows from

Equation (40) follows immediately from Eqs. (21) and ( 2 4 ) ; can thus be easily estimated by a point-counting procedure as proposed in Eq. ( 2 4 ) without need to obtain 1ine.u measurements (Weibel and Knight, 1964). 111. Application of Stereological Methods to Electron Microscopic Cytology

A. SPECIMEN PREPARATION

A number of reasons demand that utmost care be exercised in preparing biological tissues for morphometric analysis: ( 1) Dimensions obtained on processed tissue are only meaningful if they are representative of vital conditions; ( 2 ) preservation should be equally adequate for all parts of the tissue since rigorous random sampling does not permit selection of “well-preserved” areas. The fixation and embedding techniques used today for electron microscopy

262

EWALD R. WBIHEL

satisfy these conditions to a large extent (for references, cf. Wischnitzer, 1967; Sjostrand, 1967). Particular care should, however, be taken in adjusting osmolarity of the fixation medium to physiological ranges, which for mammalian tissues is 330 milliosmols. It should be noted that the customary solutions of 6.25% glutaraldehyde in 0.075 M phosphate buffer have an osmolarity of over 1000 milliosmols. Approximately isotonic solutions are obtained by using 1.5% glutaraldehyde in 0.114 M s-collidine buffer,, for example (Gil and Weibel, 1968). Because of considerable variations in stock solutions it is advisable to test osmolarity in an osmometer. It is of further importance to start dehydration with 70% alcohol since lower concentrations cause swelling of fixed cells (Weibe1 and Knight, 1964). Embedding in epoxy resins appears to yield good results with apparent overall shrinkage of tissue of around 3-5cjo. Loud et ul. (1965) have studied the effect of different preparation procedures on liver cells. They observed some buffer-dependent variation in the volume density of mitchondria; however, they comment that the specimen fixed by phosphate-buffered OsO, and embedded in Epon was probably faulty, since this method usually gives good results. In our laboratory, no significant quantitative difference could be found in fixing liver blocks with OsO, or by double fixation with glutaraldehyde followed by OsO,, as long as osmolarity was properly adjusted (Hess, 1967). The way of application of fixative will essentially depend on the organ and on the proposition. For many tissues, such as liver, immersion fixation of small dices is adequate. Hess et al. (1968) have shown in a study on the dog that needle biopsies of liver yield excellent specimens with the majority of morphometric measurements identical to those obtained on matched block biopsies; this opens the possibility of repeated tissue sampling in experimental animals, as well as in man. Fixation by vascular perfusion, e.g., by the method of Forssmann et al. (1967), may be of advantage under certain circumstances. However, particular care is indicated to avoid too drastic changes of internal conditions of the tissue because of perfusion pressure, flow rate, or inadequate osmolarity. In fixing lungs in sitz~,instillation of fixative through the airways under appropriate pressure gives good results (Kistler et ul., 1967). Baudhuin and Berthet (1967) have advocated the use of subcellular fractions for morphometric study of organelles. This is excellent if the data are compared with biochemical information obtained on the same fractions; however, it appears hardly adequate for a characterization of cellular or tissue composition since the preparation procedures involved must lead to considerable artifactual changes and to a loss of organelles (Weibel et d., 1969). 4 This is obtained by the following formula: 0.2 M s-collidine solution ( 5 4 0 rnl.); 25% glutaraldehyde (60 m l . ) ; distilled water ad 1000 ml. The osmotic pressure should always be checked in an osmometer, since considerable variations are possible.

MORPHOMETRIC CYTOLOGY

263

Ultrathin sectioning presents the greatest problems, since compression of the section is inversely proportional to section thickness. Hence, the basic requirement that section thickness be reduced as far as possible to avoid the Holmes effect (see Section II,C,7) is limited by the adverse requirement that the section be distorted or compressed as little as possible. A compromise will have to be sought to satisfy both conditions optimally. Section compression, if it is uniform for all components, is of no importance for volumetric analysis, but it affects the data when linear or planar test systems of calibrated dimension are used, such as in surface determinations. Loud et al. (1965) have, however, estimated that “ordinary” minimal section compression is, at least partly, compensated for by the optical distortions of the electron microscope. Freeze-etching methods (Moor, 1964) cannot be used for stereological studies since the “section plane” does not randomly cut through the tissue but rather follows given structures, such as membranes. The presently available methods for specimen preparation appear to be adequate although not fully satisfactory. Refinement in cytomorphometric approach, however, requires that systematic studies on optimal preparation procedures be undertaken.

B. SAMPLING O F TISSUE 1. Random Sampling

Stereological measurements, being based on geometrico-statistical principles, are derived from the probability with which section profiles of structures coincide with an appropriate test system. Consequently, it is essential that the tissue sample be confronted with the test system through a bias-free random process. This demands rigorous random sampling procedures at all stages, from choice of the animals, through selection of tissue blocks, to recording and analysis of electron micrographs. On the other hand, the sample investigated should be representative of the material studied. Hence, the electron micrographs should be well dispersed throughout the entire domain under consideration, e.g., through an entire organ. To ensure adequate dispersion of the sample, simple random sampling, when selection of each subsample depends on chance, can be replaced by systematic random sampling. Here the dispersion of the subsamples is assured by their regular spacing within a sampling lattice (Fig. 1 4 ) . Sampling is still random if (1) the lattice is randomly applied to the material, and (2) the material has no inherent periodicity which could interfere with that of the sampling lattice. Good dispersion is also assured by stratified random sampling (Fig. 1 5 ) : A number of equally spaced slices is resected from the organ; each is diced and the resulting pool of tissue blocks is processed separately; one or more blocks are picked at random from each pool.

264

EWALD R. WEIREL

Ebbesson and Tang ( 1967) have experimentally compared simple random, and systematic random sampling in counting nuceoli in the superior cervical sympathetic ganglion, They found that systematic sampling gave the smallest standard error, with stratified sampling being rather similar. Simple random

FIG. 14. Comparison of sample distribution in simple ( a ) versus systematic ( b ) random sampling of pituitary.

FIG. 1 5 . Stratified random sampling of rat liver; separate sampl from regions (strata) A, B, C, and D.

are drawn at random

265

MORPHOMETRIC CYTOLOGY

sampling proved to be the worst procedure (Fig. 16). Hennig (1967) has reached the same conclusion through theoretical reasoning. Normally, each block will provide a single section for further analysis. If more sections are to be derived from one block these should be spaced wide apart by 1000 r

6 -

\

\,-Syatemalic \

4 -

\.

2 -

I

I

- random FIG. 16. Comparison of the precision of simplc, stratified. and systematicsampling for estimating the average number of nucleoli per section N = 60; X = 510.3; S = 329.6. (from Ebbesson and Tang, 1967).

intermediate trimming down of the block to avoid duplicate measurement of the same structures, for this could introduce bias. It is much more difficult to avoid bias in recording electron micrographs since prejudice may guide the microscopist to center his micrographs on “interesting”

266

EWALD

R. WEIBEL

features. Bias can be excluded only if the position of the micrographs is fixed with respect to a reference system that is independent of the structures. One easy procedure positions the fluorescent screen tangentially in one specified corner of the squares of the supporting copper grid as shown in Fig. 17 (Weibel et al., 1966). By using grids of appropriate mesh, any desired spacing of micrographs can be achieved. It should be noted that this yields a systematic random sample. If this procedure is for some reason not adequate, latex particles can be sprayed on the section; the micrograph is recorded with the latex particle in fixed

FIG. 17. Practical procedure for systematic sampling of electron microscope fields (from Weibel et al., 1966).

MORPHOMETRIC CYTOLOGY

267

relation to a reference point on the screen (Fig. 18). This yields a simple random sample; adequate dispersion can be ensured by recording a fixed number of micrographs within each square of the supporting copper grid, which amounts to stratification.

FIG. 18. IJnbiased simple random sampling procedure; position of viewing screen is brought into predetermined relation to latex particle ( L ) deposited on section. Lines mark cross-hair engraved on screen (white) and position of photographic frame (black).

268

EWALD B. WEIBEL

For light microscopy, an automatic motor-driven sampling stage has been designed which allows systematic sampling of fields on histological sections (Freere and Weibel, 1967; Gander, 1967) ; a similar drive with preselectable step distances could be imagined for electron microscope stages. This may be useful, for example, in sampling polarized structures such as vascular endothelium when it is essential to obtain a sample of equidistant micrographs along the vessel wall (Burri et al., 1968). In his stereological studies on liver cells, Loud (1962; Loud e f al., 1965) centered his micrographs on equatorial sections of hepatocytes. His reference system thus was the cell nucleus and can therefore not be considered independent of the material under study. This method may well be adequate for study of some cytoplasmic organelles but will fail to yield a sample representative of the cell or even of the tissue. If certain organelles show a preferential orientation toward the nucleus, an unbiased sample will not be furnished. In his recent paper, Loud (1968) has modified his sampling procedure but still has not used an independent reference system for positioning of micrographs. The last sampling step involves random confrontation of the micrographs with the stereological test system. Here again, the test points or lines are best grouped in a regular lattice which is then randomly superimposed on the micrograph; details will be discussed in Section 111,CJ.The simplest method is to draw the test system on a transparent celluloid sheet which is placed over paper prints. Loud et al. (1965) have fitted a grid of fine wires to the frame of their enlarging easel ; this test system appears as a square grid of white lines on the prints. Often it is possible to work directly with the negative by projecting it on a white cardboard with the test system drawn out in ink. For reasons of efficiency, we prefer tQ record the micrographs on 35-mm. films; positives, contact printed on film, are viewed and analyzed in a compact table projector having an exchangeable screen that carries the test system (Weibel et al., 1966). Details are discussed in Section III,C,l. 2. Reasons aizd M e a m for Selecting Specific

Regiom

Random sampling from an entire organ may not always be appropiate. For example, glomeruli are concentrated in the labyrinthal region of renal cortex; in a study of glomeruli, it would thus appear wasteful to study in detail medullary regions. This is avoided by separately sampling from cortical and medullar regions after determining the relative volumes of the two regions. This leads to multiple stage sampling as outlined in Section IV,A. Similarly, in cytological studies it may be sufficient to determine the dimensions of a certain organelle with respect to one specific cell type or even with respect to its cytoplasm only. The tissue sample should again be obtained by a random process; however, those parts of the micrographs that do not contain the

MORPHOMETRIC CYTOLOGY

2 69

specified cell under study can be disregarded. One example is the study of the volume density of a specific organelle in the cytoplasm of endothelial cells (Fuchs and Weibel, 1966; Burri and Weibel, 1968) ; it is evident that intimal or smooth muscle cells, although present on the micrograph, can be excluded from analysis. As will be outlined in Section III,C,2., the test system must then be defined with respect to the containing volume, in this case endotheha1 cytoplasm. 3 . Sample Size

The larger the sample investigated the more reliable the information and the larger the effort required. Definition of sample size will hence depend on the accuracy required and on the time available for study, but it will also depend on some characteristics of the structures themselves. Unfortunately, no generally applicable rules are yet available for determining sample size beforehand. Evidently, determination of sample size is a statistical problem. It has been shown, e.g., by Giger (1967), that through the basic stereological formulas of Eqs. ( 3 2 ) Pp, I , , and Q, provide unbiased estimates of the structural parameters Vv,Sr,and Mv,respectively. Giger and Erkan (1968) have investigated the relation between structural parameters, designated by X , and the corresponding unbiased stereological estimate x, and have shown that X can be estimated by the mean gn of j z observations and that the variance of the mean a2zn is estimated by (Xi

- .,)X

This result is in complete agreement with general statistical theory (Cochran, 1953), Eq. (41) representing the square of the standard error of the mean. It holds, irrespective of geometrical properties of the structures, as long as the system is isotropic. If the samples x are normally distributed, 2 2 0 ; defines the “95%) confidence interval” (cf. Cochran, 1953). On this assumption, D e Hoff (1967) has proposed to estimate the number n of observations necessary to yield a result with a 95% confidence interval of +y percent of the mean E from

where 2 and S, can be estimated from a comparatively small sample. In many cases, however, normal distribution can not be assumed. Giger and Erkan (1968) have proposed an alternative method, which is independent of the type of distribution, for estimating the variance for point-counting volumetry,

270

BWALD R. WEIBBL

or rather, point-counting planimetry as the true observation on the section. If a point net is laid out on the section, an average of 7 points will be enclosed in individual profiles of structures. With d as the area of the fundamental parallelogram of the point grid (Fig. 21), the profile area A is estimated by

A=aXF

(43)

and the variance of P as

S’, 5 F (P*

-F )

(44)

where P* is the maximum number of points enclosed in profiles. A rough preliminary estimate of 7 and P* provides an estimate of the variance to be expected; from this, the point density needed to obtain a satisfactory accuracy can be derived. Giger and Erkan (1968) have also discussed a similar method for choosing the proper test line density for intersection counting for estimating Sv with a specified accuracy. of the mean is In biological work, a 95% confidence interval (y) of usually adequate. If application of Eq. ( 4 2 ) would necessitate the study of an excessively large sample-judged on requirement and availability of time-it may often be acceptable to reduce the accuracy requirement. It should be borne in mind that increasing the confidence interval to *15% will cut the necessary number of observations in half. The variation among individual observations critically depends on whether the field of observation encompasses a representative fraction of the tissue. “Representance,” that is, the degree by which a sample is representative of the population, must again be defined in terms of accepted confidence interval. Giger (1969) has developed a general method for deriving representative size of tissue samples for stereological analysis. By this method it can be shown that the representative section area A, is inversely proportional to the volume density V,, of the structures and depends furthermore on some, as yet undefined, measure of dispersion. It was found in a study of liver that A, for hepatocyte nuclei is about 10 times larger than that for mitochondria although mitochondria1 volume density exceeds that of nuclei only by a factor of about 2.5; this is because of the higher degree of dispersion of mitochondria as compared to the coarse and widely spaced nuclei (Weibel and Gnagi, 1968). This problem is also discussed by Chayes (1965). It is thus evident that the representative section area stands in relation to a “fundamental domain” of the material, i.e., to the minimal domain or volume that contains a representative amount of all structures. In cytology, the fundamental domain appears to coincide with the individual cell. Definition of representative section area in terms of fundamental domain is, however, not easy since a random section through one fundamental domain is not necessarily representa-

MORPHOMETRIC CYTOLOGY

271

tive, and in fact in all probability is not. The theoretical basis for a sound practical solution of this problem is still lacking.

4. Multiple

Stdge

Samplirig

It has become evident from the foregoing that the wide range in size and in degree of dispersion of cellular constituents introduces considerable problems in defining representative, and yet reasonable, sample size in applying stereological techniques to electron microscopic cytology. Is it, for example, reasonable to measure mitochondria on a sample of micrographs t o times larger than necessary, only to satisfy accuracy requirements for nuclei? Or should the accuracy requirements for nuclei be reduced to a level at which the data become meaningless, just to keep sample size in a reasonabIe range with respect to mitochondrial measurements ? The solution to this dilemma is multiple stage sampling which provides the means for satisfying identical accuracy requirements for all constituents while keeping the analytic effort in a tolerable range. The particulars of multiple stage sampling are essentially dictated (1) by the minimal magnification necessary to unambiguously recognize a given constituent, and to identify its intersections with test lines, and so forth and ( 2 ) by the basic proposition and the aim of the study. Figure 19 shows a sequence of increasing magnifications of rat liver cells in relation to test lines. The differentiation of test points falling on mitochondria and on microbodies is easy at l0,OOox magnification, but at least 40,OOOX magnification is needed to identify line intersections with membranes of rough and smooth endoplasmic reticulum or with mitochondrial cristae. In practical application of multiple stage sampling, a hierarchic sequence of reference systems is defined by which detailed but still efficient measurements can be obtained even on sparse subcellular constituents, while still allowing the establishment of their relationship to the entire organ, or even to the organism. Such a sequence will be defined for the liver as an example, the sequential reference systems being body weight ( W ) , liver weight and volume (VL), volume of hepatic parenchyma (V,), and volume of hepatocyte cytopIasm (V,), with the following relationship:

VL==fXW V , = vv, x V , = vv,

VL

= vv, x f

xw

x v,= vvc x vv,,x VIA= V v c x VvII x f x w

The coefficient f indicates the fraction of body volume occupied by the liver. It is evident that determination of the corresponding volume fractions V v will suffice to allow transformation to any other reference system of all measurements obtained with reference to a given subspace. It goes also without saying that

272

EWALD R. WEIBEL

MORPHOMETRIC CYTOLOGY

273

similar sequences of reference spaces can be defined for any other organ or tissue.

C. STEREOLOGICAL ANALYSIS O F ELECTRONMICROGRAPHS 1.

Recording of Elertron Mirropzphs

iiz

Stiitnble Form

Stereological work requires a comparatively large number of micrographs to be recorded and analyzed. The customary recording on large-sized film or plates followed by printing on 8 x 10-inch or 18 x 24-cm. paper thus appears cumbersome and expensive although it is certainly acceptable. Paper printing has the drawback, however, of bad dimensional stability of photographic papers which, upon drying, can shrink by up to 10% in one direction, thus leading to appreciable distortion. Loud (1968; Loud et al., 1965) has eliminated this difficulty by simultaneously printing test system and micrograph. If the paper shrinks, both micrograph and test system are distorted to the same degree. Care must be taken in calibration of the test system, however. W e have found recording on 35-mm. film most efficient since 40 to 50 micrographs can be recorded in one operation of the microscope. The quality of micrographs is adequate for this type of work. Unfortunately, a number of microscopes that would be excellent for efficient screening, such as the Zeiss EM 9, do not as yet have facilities for 35-mm. recording, but it can be hoped that this will be corrected. Aside from efficiency and low cost, 3j-mm. micrographs have the ,Idvantage of being analyzable in compact projection systems without need of paper printing. In the device shown in Fig. 2 0 , the micrograph is projected via twu mirrors onto a ground glass screen which contains the test system (Weibel et nl., 1966) ; the film transport is in easy reach of the operator. Although it is possible to project the negatives, we prefer to contact print them in a long light box on a strip of the same film. With large-sized negatives a similar setup can be made with a photographic enlarger by projecting the negatives, or contact prints thereof, on a white board carrying the test screen. W e have recently modified the device shown in Fig. 20 to accommodate both 35-mm. and 70-mm. film; it can also be made to accept lantern-slide plates. 2. Stereological Test S ~fems J f o r the Stiidj, of Aggregntes of Stt.iictureJ

It follows from Section II,C that efficient application of stereological methods requires the following basic test probes: (1) a set of points for volume estimation; ( 2 ) test linej of known length for estimation of surface or boundary FIG. 19. Increasing magnification reduces amount of material encompassed by field but improves relationship between size of traces and thickness of test lines. Magnifications: a, 10,000~; b. 2 0 , 0 0 0 ~ ;( , 4o,(looX; d, 8u.(~(~ox.

274

EW'A1.D R. WElBEL

areas; ( 3 ) a test areit of known size for estimation of the length of curvilinear features and for particle (profile) counting. These probes are simultaneously realized in a simple quadratic lattice of lines as illustrated in Fig. 21: T h e cross-

FIG.20. Stereological laboratory unit for analysis of electron micrographs. A projector ( P ) with carriage for 35-mm film ( P ) projects the micrograph via two mirrors onto a screen (S) fitted with an appropriate test grid. The data-compiling unit consists of a keyboard ( K ) which feeds the counts into a data accumulator ( D ) with 10 counters. These data are automatically transferred onto tables (T) and, optionally, into a card puncher not shown on the picture.

points of the lines serve as markers fur point-counting vtrlumetry, the lines for intersection counting, and the area between the outermost lines and the dotted lines for counting profiles and intersections of linear features with the section plane. This test system is coherent in the sense that there exists an exact relationship between the number of points, the length of test lines, and the size of the test area. Its unit is a square of area n = &, of which two sides and one point are available as probes. With IZ unit squares making u p the test lattice the system is defined as follows: test points: P, = 12 test line: L, 12 n x d 1P, z 12 x dz = P, test area:

x 2d x dz

The use of coherent test systems is advantageous since it allows the study of structures that are contained in relatively small compartments. As stated above, it may often be preferable to estimate the surface density of endoplasmic reticulum in cytoplasm of a given cell type rather than in whole tissue. However, the profile of cytoplasm may not always fill the entire screen (Fig. 2 2 ) . T h e length

MORPHOMli’IRI(.

275

(.YTOLOGY

of test lines in cytoplasm L, can then be easily estimated by counting the number of test points P, included in cytoplasm: L, = P, x 2d; and the test area for counting organelIes In cytoplasm is A , = P, x da. It is also evident that the

volume density of mitochondria in cytoplasm, for example, is Yp:i = Pnki/Pe. The optimal density o f test points depends on the size of the unit constituent.

’ d

a

= d2

I

f

FIG. 21.

--____.I

Coherent test system with square unit

In point-counting volumetry, the test points should be so spaced that no more than one point will be included in the individual profile if V v 0.5; under these conditions the sampling error is smallest for a given total number of test points (Hilliard and Cahn, 1961). In a quadratic lattice the spacing d of test points thus depends on the maximal profile area din:

<

d2 > 4 1 ,

(45)

On the other hand, the total number of test points needed to achieve a prescribed accuracy is inversely proportional to the square root of V c - (Hennig, 1957; Weibel. 1963). A point net constructed according to Eq. (45) may therefore still be too dense in studying 3 frequent constituent since it is better, for reasons related to representative sample size, to distribute a smaller number of points on a larger number of micrographs than to test with many points on only few tissue smples.

It has become clear from the foregoing that in setting up a procedure for multiple stage sampling we have to take into account the following conditions: (1) the optimal magnification; ( 2 ) the minimal number M of micrographs at this magnification needed to form a representatitve sample of total area A,.; (3) the total number of test points P,,? necessary to yield a result of specified accuracy; ( 4 ) the optimal spacing d of test points according to Eq. ( 4 5 ) . With these

FIG. 2 2 . Application of coherent test system of Fig. 2 1 to a liver cell in order to exclude the nucleus from measurement (data related to cytoplasmic volume),

MORPHOMETRIC CYTOLOGY

277

conditions established, the number of test points P , to be applied to each micrograph of area A IA,/rM is fixed between the two following limits (PT//+f)

< p-1 < ( A / d 2 )

(46)

whereby, for reasons of efficiency, P, should be as close to P,/M as possible. W e would like to deviate from this rule, however, if the organelle studied is or less of the volume. In this very rare, i.e., i f it occupies only about 2-5 case, we would like to make sure that practically every profile present on the section is apprehended by the test system and hence contains about one lattice point. In such cases, we would choose the lattice period so that d 2 is about equal to the average profile area. T h e foregoing argument was related to point-counting volumetry. Identical prescriptions can be given fur setting up the test system for surface density measurements. According tc Hilliard ( 1965) the t o t d number I, of intersection counts required to achieve a given accuracy is

v,

IT

7==3

0.4

[S,./O(S,.)]’

(47)

From this we can estimate the total test line length L,, which is to be distributed over M micrographs, as

It will thus suffice to take a limited number of counts with an arbitrary test line to roughly estimate i,,and a(lr,) in order to fix L,. T h e test line density L,4 to be applied to each micrograph is then found by

In using a simple grid of parallel equidistant test lines, their distance is chosen as d = l / L , , ; with a square grid d = 2/L.,,. T h e foregoing argument has shown that the spacing of test points should conform to the size and to the frequency of the structures studied. With the wide disparity in characteristics of cytoplasmic organelles, one test system is certainly not appropriate for all structures. Coherent multiple lattice test systems, however, allow simultaneous estimation of widely disparate structures (Weibel et ul., 1 9 6 6 ) . Such test systems (Fig. 23) consist of a square lattice of spacing d in which every gth line is heavier. With P, cross points of the heavy lines, the total number of points in the test system is P’, 1g2 x P,. T h e advantages of multiple lattice test systems-which in some way represent a type of multiple stage sampling-are obvious: in estimating the volume density of a sparse organelle, o, in cytoplasm, c, it is only necessary to count all the

278

EWALD R. WEIHEI.

points Po included in profiles of o, and the number of heavy points P, included in cytoplasm. It then follows that V y , = P,,//jj' x P,. A double lattice test system with g = 5 was successfully used in studying a scarce organelle of V.ISCUlar endothelia (Fuchs and Weibel, 1966; Burri and Weibcl, 1968). In a morphometric study on the liver cell, a double lattice with g = 3 allowed

FIG.23. Coherent double lattice test systems of lattice point ratio 1 : 4 ( a ) and 1 : 9 ( b ) .

simultaneous and efficient estimation of the volume density of nuclei, cytoplasm, and mitochondria with the coarse grid, and of microbodies and lysosomes with the fine grid (Weibel et a/., 1969). It should be noted that coherent double lattice test systems obviate the necessity for scanning all points of the fine grid and are thus very efficient. The major drawback of the square grids discussed so far is the excessive density of test lines with respect to test points. It was explained above that, for

MORPHOMETRIC CYTOLOGY

279

statistical reasons, the test system shouId have no more than one intersection with the features (Hilliard and Cahn, 1961). In determining surfaces and volumes simultaneously, we would thus set up the test system in such a fashion that the profile contains no more than one test point and that its boundary forms one or two intersections. (It must be noted that test lines form at least two intersections with closed curves if their end points lie outside the curve). With the quadratic line grid this cannot be achieved, since the boundary of any profile that contains a point will have d t least four intersections with test lines. Furthermore, for all profiles completely within the test area there will always be even numbers of intersections, in minimum two, since for every “entry” of the line there will be an associated “exit” point. The number of intersections counted is hence at least twice as large as would be required. To eliminate this difficulty, a test system composed of short test lines of equal length z was developed (Weibel et al., 1966); it had its origin in a proposition by Chalkley et ul., (1949) to use short “needles” for the estimation of volumeto-surface ratios. The test systems illustrated in Fig. 24 are all coherent in the above-mentioned sense. By using the end points of the n test lines as markers for volumetry, the test system is defined as follows

PT = 271

For n = 21 or 84, the test area is roughly a square; if it is desired to have PT = 100, the test area is rectangular (Fig. 24b). The arrangement of lines in a triangular lattice results in homogeneous dispersion of points and test lines; it further provides easy working conditions in that the rows of lines can he conveniently scanned. An unlimited number of alternate test systems can be designed, particularly in view of special applications. In studying anisotropic structures, the curvilinear test system of Merz (1968) can be useful (Fig. 2 5 ) ; it is essentially composed of semicircles that are so arranged as to allow easy scanning. The use of circular test lines eliminates directional bias resulting from structural anisotropy. Sitte (1967) proposes the use of triangular test lines for this purpose. I t should be noted that a quadratic lattice also eliminates the effects of structural anisotropy, at least partially. The problem of anisotropy will be further discussed in Section VI. In practical application to electron micrographs we have found a screen size of the order of 30 x 30 cm. to be most convenient, be it for the table projector

280

EWALD R. WEIBEL

unit of Fig. 21 or for projection of larger negatives onto a white board. The thickness of lines critically affects error; they should be thick enough to be easily recognized but as thin as possible for unambiguous identification of intersections. Optimal thickness lies between 0.2 and 0 . 3 mm.

.............. .............. .............. ............... .............. ............... .............. ............... .............. ,;;;;;;a;;;;;:'

I

FIG. 24. Coherent multipurpose test systeius with 42 test points ( a ) points (b) .

and 100 test

3 . Krrovditig of P r i n i q Duta

Primary stereological data are classified counts; they can thus be recorded in any tallying device such as mechanical or electrical hematological counters. These counters should have a capacity of 5 to 10 classes, each with at least two to three digits. The counts are written out on tables atfer reading each micro-

MORPHOMETRIC CYTOLOGY

28 1

graph. This is cumbersome and furthermore prone to error; if stereological work becomes a serious research tool a more efficient recording system will be required. For this purpose we have recently developed a 10-counter electrical tallying device with selectable totalizers (Fig. 2 0 ) , which prints out tables and punch cards automatically (Weibel, 1967c, cf. Section III,C,6).

FIG.2 5 . Coherent semicircular test grid for eliminating anisotropy of test lines Lfrom Merz, W. A., Mikioskopie 22, 1 3 2 (1968)i.

4 . Calculation

of

Morphonzetric Puvameterj of Aggregutes

The compiled primary data are introduced into the appropriate formulas to yield the desired morphometric information. 170r obtaining volume densities the differential point counts are introduced into Eq. ( 9 ) . Surface densities of membranes are derived from Eq. (1 5 ) by introducing the counted number of intersections and the total test line length. For estimating numerical densities on the basis of Eq. ( 2 9 ) , the number of profiles must be divided by the test area (N,)and set in relation to the volume density estimated by point counting; the shape and size distribution coefficients must be independently determined. It is evident that combination of the basic formulas of Section II,C, can provide information on an unlimited number of specified parameters characterizing any cell type. This must be left to the ingenuity of the investigator; a specific example related to the liver cell is explicitly discussed elsewhere (Weibel et a]., 1969). It also goes without saying that computation of all these data is easily performed with the aid of a computer (see Section IIl,C,6). It should be noted that the primary data must be related to a rigorously de-

282

EWALD R. WEIBEL

fined reference system; it is, for example, essential to decide beforehand, i.e., before beginning the study, whether the volume of a given organelle is to be related to the volume of total tissue or to that of cytoplasm of a specific cell. The morphometric parameters can be expressed as relative or as absolute dimensions, whereby the latter must likewise be defined with respect to a reference system, e.g., body weight. For practical purposes we have found the following types of parameters useful: (1) Concentrations or demities in tissue ( X , ) ; these follow directly from stereological formulas. (2) Absolzrte dimemiom per organ or animal, obtained by multiplying X , by the organ volume V,. (3) lfSpecific” d i m e m i o m relating the amount of structure to the unit body weight W ; this is obtained by multiplying X , by the “specific” organ volume V,/W and is particularly useful for elimination of variations in body size of animals in experimental studies (Weibel et a/., 1969; Staubli et al., 1969). (4) Relative d i m e m i o m , in the strict sense, expressing one parameter with respect to the other, e.g., surface of mitochondria1 cristae to matrix volume. 5 . Statistical Treatment o f Data

In discussing sample size (Section III,B,3) we have pointed out some of the statistical problems inherent in stereological analysis. In short, well-founded statistical procedures for testing the validity of data are not yet established. Hennig (1957) and Hally (1964) have used the notion that the expected error in volume estimations by point counting must be inversely proportional to the square root of the number of test points applied to the specimen and must furthermore depend on the volume density of the component investigated. For the present, we can avail ourselves of the customary statistical methods and can determine standard errors of means or 95% confidence intervals; or, in experimental studies, group averages can be compared by means of Student’s t-test, for example (Staubli et al., 1969), on the assumption that the parameters estimated for individual animals are normaliy distributed. More appropriate methods should become avv’l I a1>1e soon. One of the still unsolved problems is how to define the appropriate sample unit for statistical analysis. Should the morphometric parameters be calculated for every microscopic field with these estimates averaged and subjected to statistical analysis ? O r should statistical tests be performed on primary data with the morphometric parameters calculated from mean primary data? This may yield different results mainly when the size of the reference system varies from field to field as is the case in determining relative dimension. W e are, for the present, more confident in relative parameters derived from average primary data; there is, however, no method yet available to express the statistical validity of these parameters.

283

MORPHOMETRIC CYTOLOGY

G. Po.uibilifiej

of

Antomation

Recent years have intruduced a number of devices for automation of stereological analysis; these have been reviewed by Iischmeister (1967). Probably the most advanced instrument is the Quantimet Image Analysing Computer of Metals Research, Ltd., Cambridge, England (Fisher, 1967). In this device (Fig. 26) the image is broken up into dense lines by means of a television camera.

MONITOR

4~

CAMERA

EPIDIASCOG

OR

1

I?;-I METER

c ?

MICROSCOPE

SPECIMEN

OUTPUT

@

FIG. 26. Functional diagram of Quantimet automatic stereological image analyzer. Epidiascope feeds electron micrographs into camera which can also be directly attached to electron microscope. (From Instrument Documentation of Metals Research, Ltd., Cambridge, England).

Changes in beam intensity resulting from variation in contrast of the micrograph are analyzed in a small computer which transforms these primary signals into stereological information. For example, the fraction of the linear sweep LL passing through regions of a given level of contrast is used to estimate the volume fraction V v of the corresponding component [Eq. (6) 1; or the number of steps or abrupt changes in the contrast level per unit length of sweep I , serves for estimation of surface or interface density Sv according to Eq. ( 1 5 ) . The device can also yield profile size distributions and can be programmed to calculate directly particle size distributions. As a most recent development, the Quantimet can be directly connected to the AEI EM 6 electron microscope thus obviating the necessity for recording micrographs. Although this excellent instrument can distinguish between five contrast levels, it still demands as essential conditions for its applicability that the components of interest are unambiguously characterized by well-defined differences in contrast and that within each profile contrast does not vary significantly. In biological electron microscopy these conditions are evidently not met. Many organelles

2 84

IiU’A 1 .I) R. W I d Ii li I .

are recognized only on the basis of a characteristic configuration of membranes or of associated structures. It appears, at present, hopeless to have a computer automatically discrimate mitochondria or rough endoplasmic reticulum cisternae out of the maze of the cytoplasmic membrane system. It is thus unfortunate that most biologists will not be able to profit from automatic image analysis; decision on the classification of subcellular structures must be left to the experienced investigator. Even the extremely advanced computer SADIE (Scanning Analogue to Digital Input Equipment) of Moore (1967) is of no help in our case. Semiautomation, however, is of great help in improving efficiency of stereological analysis. Lazarow and Carpenter ( 1962) have developed a semiautomatic scanning instrument for electron micrographs. A point marker traverses the micrograph while the investigator presses a key assigned to the component traversed; the length of traverse is recorded automatically. This device has many interesting features which in further development may make it very versatile, while still leaving identification of structures to the investigator. It should be noted that recording of individual traverses or intercept lengths in such a device may be used for derivation of particle size distributions according to the method of Hilliard (1967a) quoted above. In our laboratory, we use point-counting procedures whenever possible. T h e reilsons are many. Above all it is the most efficient and the least arduous of all the methods. Hilliard and Cahn (1961) have also demonstrated that point counting is statistically superior to linear analysis for volumetry in spite of great inaccuracy in estimating the contribution of an individual organelle. Aside from this, point counting leaves fewer uncertainties than linear analysis: Ambiguities always occur at transition points from one component phase to the other; in linear analysis such ambiguities are bound to occur constantly, as well as in automatic systems, while in point counting they are restricted to those comparatively rare points that lie “exactly” on the boundary between phases. To clearly establish the location of such points the investigator can pause, while in scanning devices he is pressured to reach a quick decision. Point counting does not require any automation of the analytic part since no measurements have to be taken. However, the necessity for tabulating the tallies after short intervals has made it desirable to automate the entire procedure of transferring primary data onto tables and punch cards. This is successfully accomplished in the device illustrated in Fig. 2 0 (Weibel, 1967b,c). The 10 counters are fed from a keyboard, the assignment of keys to counters being preselected by means of a pegboard. Each counter can be coupled with one of two totalizers or with none; this provides, for example, the possibility of checking on the total number of test points counted. Upon operation of a switch the content of the counters is automatically typed out onto a table by an IBM 72 typewriter. Simultaneously, the data can be transferred to an IBM card puncher.

MORPHOMETRIC CYTOLOGY

285

The last step in automation relates to data analysis. A complete computer program for automatic calculation of a great variety of morphometric parameters from primary data obtained by point counting has been worked out (Gnagi and Weibel, 1968); it is available in FORTRANlayout from the authors. The general program is composed of a number of basic and special subprograms which are listed in a catalog; they can be combined by means of entry cards to suit any specific purpose. The first stage of the program performs computation of the requested basic morphometric parameters for all micrographs forming the primary sample, representing, for example, one experimental animal, and determines the standard error of the mean for each parameter. A second stage calculates group averages, and a third performs their statistical comparison by means of Student’s t-test (Weibel e/ nl., 1969; Staubli et d.,1969). The introduction of this limited degree of automation in data collection and computation has many times increased efficiency and output of our stereological laboratory. 7. Determinatioii avd Recordiiig of Profile Size

Di.\ tvibzrtion

The derivation of size distributions of particles or of their parameters from sections requires size distributions of profiles to be determined. This involves direct measurement of the dimensions of individual profiles, such as the diameter (mean) of circular profiles derived from roughly spherical particles. It is most convenient to estimate the profile diameter by fitting circles to the profile (Fig. 2 7 ) . This is easily performed by means of a transparent plastic stencil with graded circles as used for graphic work, or with a set of concentric circles drawn on a clear plastic sheet. Proper choice of intervals allows direct classification into a number of size classes whereby it is found that 1 2 to 1 5 classes are

iz FIG. 27. Plastic stencil with graded circles for estimation of average diameter of profiles.

286

EWALD R. WEIBEL

usually sufficient to characterize the profile size distribution (Saltykov, 1967; and others). The Zeiss Particle Size Analyzer (Endter and Gebauer, 1956) is a refined tool for analysis of electron micrographs printed on paper: The size of a circular light disc, projected onto the micrograph, is approximated to the profile; the size of the disc is automatically read off and, on operating a switch, a tally is added to the appropriate class recorded in one of 50 counters. The intervals between the classes can be selected on either a linear or a logarithmic scale. This device has been used, for example, by Heiniger et ul. (1967) and by Staubli et al. (1969) for sizing nuclear profiles of lymphocytes and hepatocytes on electron micrographs, and by Haug ( 1 9 6 7 ~ ) for deriving size distribution of nerve fibers.

IV. A n Example of Morphometric Characterization of Organelles: The Liver Cell A. GENERAL CONCEPT OF

THE

STUDYAND SAMPLING PROCEDURES

To illustrate application of stereological methods to morphometric cytology the essential steps in a recent systematic study of the subcellular composition of rat hepatocytes will be brieflly reviewed (Weibel et al., 1969). This study was undertaken with the aim of obtaining quantitative correlation between morphological and biochemical changes after treatment with phenobarbital (Staubli et ul., 1969). The morphometric data therefore had to be obtained in a way thdt conformed with the biocheniicnl studies. These were done on subcelluldr fractions of homogenized livers, which essentially represent random samples o f components from all parts of the organ ; hence, morphometric sampling had likewise to be random on the assumption that hepatocytes form an entity. This contrasts with a recent study by Loud (1968), who studied regional variations in the morphometric parameters of liver cells in different zones of the lobule and consequently had to study a sample of cells selected from three different zones of liver lobules. In accordance with the biochemical practice of relating the data to the unit weight of liver tissue, the basic morphometric data were expressed in relation to 1 ml. of tissue,5 i.e., as structural densities. Because of the wide range of structures-from lobules about 1 mm. in diameter to smooth endoplasmic reticulum tubules 300 A. wide-the following method of multiple stage sampling was adopted : ( I ) Goldner-stained paraffin sections were evaluated at 200x magnification using an automatic sampling stage microscope of WILD for systematic sampling 5

T h e specific gravity of liver tissue was determined to be 1 0 6 7 on the average.

MORPHOMETRIC CYTOLOGY

287

of fields (Freere and Weibel, 1967; Gander, 1967). By means of a 100-point square lattice test system on the screen of the projection head, the volume density V,, of lobular parenchyma in the whole liver was estimated. (11) From a pool of osmium-fixed and Epon-embedded tissue cubes a random sample of five blocks per animal was collected. One-micron-thick sections served for light microscopic evaluation of the number and size of nuclear profiles at looox magnification; this allowed estimation of the number of cells in the unit tissue volume.

( H I ) From the sdme blocks, ultrathin sections (600-900 A , ) were cut and mounted on 200-mesh copper grids for electron microscopy. Six electron micrographs, systematically sampled by positioning the viewing screen into specified corners of the supporting grid (Fig. 17) (Weibel et al., 1966), were recorded at 2500X magnification. Fields that contained no lobular parenchyma were discarded; the data obtained on this sample were thus related to lobular parenchyma and had to be multiplied by Vv,,to establish their relationship to whole liver tissue. The final magnification of the micrograph projected onto the test screen of the projector unit was 2 2 , 5 0 0 ~ By . using a 9 : l double lattice test screen (Fig. 2 8 ) , the volume density o f hepatocyte cytoplasm V v pof nuclei and of larger cytoplasmic organelles (mitochondria, microbodies, lysosomes) in lobular parenchyma was estimated, the fine mesh being only used for microbodies and lysosomes. (IV) A second sample of six micrographs from each of these five sections per animal was recorded at a primary magnification of ~ 0 , 0 0 0 The ~ . rules for positioning the screen were as in level 111; however, fields that contained less than 50%) cytoplasm were discarded, since at this level all measurements were related to cytoplasmic volume. Stereological analysis was performed at 90,000 final magnification by using a multipurpose test screen with 18 lines (Fig. 29). It included an estimation of volume and membrane surface of rough and smooth endoplasmic reticulum, as well as of mitochondria1 envelope and cristae. To establish the relationship to whole liver tissue these data had to be multiplied by (V,,, x Vvc.), obtained in levels I and 111.

x

B. SPECIFICMETHODS FOR ESTIMATING MORPHOMETRK PROPERTIES O F CELLS AND SUBCELLULAR COMPONENTS In indicating specific methods for morphometric characterization of cellular structure from electron micrographs, we shall refer essentially to the liver cell as a model (Weibel et d.,1969). It is evident that part of these methods may have to be silghtly modified when applied to other cells. Furthermore, we shall consistently propose point-counting methods for the reasons presented in Section II,C,2.

288

BWALD R. W E I R E L

FIG. 2 8 . Double lattice (9: 1) test system superimposed on section of liver (sampling level 111) with sinusoid ( S ) , biliary capillary ( B ) , and hepatocyte nucleus ( N ) . Note that coarse points are adequate for estimating volume of mitochondria ( M I ) , while sparser organelles (lysosomes, 1.Y) must be assessed with fine grid (from Wribel CI al., 1969). Magnification I2.500X.

MORPHOMETRIC CYTOLOGY

289

1 . Cell Size am! i\.l~ilorCell Compcirtmeizts

The number of cells per unit volume was estimated from the number of hepdtocyte nuclei N,-,, counted in the sample of stage 11; binucleated cells were not considered separately. At level I1 the volume density of cytoplasm V,.,. and of nuclei Vvn was estimated by using a square lattice of 99 points; the point

FIG. 29. Electron micrograph of liver cell (sampling level IV) with test system i n actual size relationship (from Weibel et al.. 1969). Magnification 37,500X.

distance was 2.5 cm. on the screen. T h e average volume of a mononuclented hepatocyte was found from TIl

= (V,.,. + VL~ll)jNl~lL

(50)

The cell surface was measured in level-111 micrographs with a square-line lattice of 2-cm. spacing, whereby intersections with sinusoidal, biliary, and juxtacellular surfaces were separately recorded to appreciate cell polarity. An estimate

290

EWALD R. WEIBEL

of the average cell diameter 2 was obtained from the volume-to-surface-ratio (v /.r ), since a= 6 ( 2 ~ / 5 ) = 6 ( V v / S v ) . 2 . Mitochondricl

Mitochondria are quantitatively characterized by their volume density VVrmi, their numerical density Nvmi,and by the surface density of outer (Svmo) and inner membrane with cristae (Sr,nc.).The ratio of cristae membrane surface to matrix volume may, of course, be derived from these basic parameters. VVllliand Nvmiare best derived at a lower power such as level 111 ( 2 2 , 500x); the coarse 99-point grid was found to be adequate for determining Vvllli. Calculation of Nvmi through Eq. (29) presupposes a knowledge of mitochondrial shape; on the basis of an analysis of axial ratios of the roughly elliptic profiles, it was judged to be ellipsoidal with an average axial ratio of 4:1 which gave a value of p = 2.35. This can evidently yield only coarse estimates since considerable variation in mitochondrial shape must be expected. For counting mitochondrial profiles, Ndmi,a frame of 103 p2 was used; profiles completely within the frame and those intersecting its left and upper side were counted, while those intersecting the right and lower side were disregarded; this is analogous to customary erythrocyte counting. Counting intersections of test lines with mitochondrial membranes requires higher magnifications for better resolution of intersection points; 9 0 , 0 0 0 x was found convenient although 50,000X would still be adequate (Fig. 19). For convenience, intersections with cristae (ILmc) and with outer-plus-inner inembrane (ILmo) were counted as one point although two membranes were involved (Fig. 30). While the surface density of envelope membrane could be directly calculated through Eq. (15), that of inner membrane with cristae was obtained by

where an asterisk indicates that surf'ice density and test line length refer to cytoplasm and not to whole tissue (cf. Section III,C,2). It should he noted that hnite section thickness causes the extent of cristae to be underestimated by 20-400/0, while the envelope membrane area is not greatly affected by this error. 3. EiidoplaJmic Keticiiltim

Measurement of this finely textured component demands high magnification of 50,0O0-9O,OOOx (Figs. 19 and 30). An 18-line multipurpose test screen is again appropriate (Fig. 2 9 ) . In determining volume and surface densities of endoplasmic reticulum, it is often of prime importance to distinguish between its rough and smooth form. To avoid uncertainties and arbitrary judgment at

MORPHOMETRIC CYTOLOGY

291

FIG. 30. Actual magnification of portion of electron micrograph as used for stereological analysis at level IV. Intersections of test lines with mitochondria1 membranes are encircled, one circle representing one count. Note that the double intersections of cristae and “marginal” membranes are counted as one point for simplicity’s sake (cf. Eq. 5 0 ) . Magnification 80,000X.

points of transition, it was found useful to define endopIasmic reticulum (ER) as including all membrane-bounded cytoplasmic cisternae, tubules, and vesicles (Weibel et af., 1969). In particular, the perinuclear cisterna was considered a part of rough ER space, its external ribosome-studded membrane a portion of rough ER membrane. Similarly, all Golgi elements were attributed to smooth

EK. It is evident that this simplification will not be appropriate for every proposition. T h e number of membrane-bound ribosomes can be determined by an indirect procedure: O n high-power electron micrographs ( - 9 0 , 0 0 0 ~ ) , profiles of rough ER with clearly discernible membranes are randomly selected (Fig. 30). T h e contour length b of the membrane trace is measured with a thread or a map-measuring device and the number of ribosomes attached to this profile is counted. It must be assumed that these ribosomes are attached to a strip o f membrane extending through the entire section thickness which, in our study, was 600-900 A. O n the basis of the mode of selection (“clearly discernible trace”), the width of the strip can be estimated to be about 1.2 times the section thickness, or roughly 850 A. T h e number of ribosomes can thus be related to a strip of membrane of area ( b x 850 A ) which allows calculation of the nuniber of p~rticlesper 1 pz of membrane. T h e number of ribosomes per unit volume of cell or tissue is obtained by multiplying this figure with SVrer. As in the case of mitochondrial cristae, section thickness causes an underestimation of Vv,,rm d Svrrby 20-30%, because profiles formed by grazing sections cannot be recognized. Loud ( 1967) estimates this underestimate to be even higher, but this may also be because of the lower powers used in his studies.

4. Cytop1u.r mir Gruiz/~les Cytoplasmic granules, or paraplasmatic organelles, are usually scarce components; estimation of their volume density therefore requires a relatively dense point grid applied to low power, i.e., large field, electron micrographs. I n liver cells, microbodies (peroxisomes) and lysosonies are easily identified on micrographs of 2 2 , 5 0 0 x magnification. Their volume density was determined with the 891 (99 x 9 ) fine points of the 9:1 double lattice test system of level 111 (Fig. 2 8 ) ; the coarse point grid was found to be inadequate (Weibel et dl., 1960). To estimate the numerical density of peroxisomes by Eq. ( 2 9 ) , the profiles were counted within the frame used for mitochondria1 counts; since peroxisomes are short ellipsoids, the shape factor 13 was assumed to be 1.45. In studies on rod-shaped granules of endothelial cells (Fuchs and Weibel, 1966; Burri and Weibel, 1968), higher powers were necessary since the specific organelle under investigation could only be identified on the basis of a fine internal structure. For the purpose of relating their volume density to cytoplasmic volume, a 25: t double-lattice test system was used (Weibel et al., 1 9 6 6 ) ; the fine points were employed to estimate the volume of organelles, the coarse points to assess the extent o f cytoplasm. It should be pointed out that this was a difficult case in terms of sampling since endothelial cells are highly polarized thin cells and the organelles are not evenly dispersed throughout cytoplasm but occur in clusters (Burri et a/., 1968).

MORPHOMETRIC CYTOLOGY

c.

CORRELATION O F

BIOCHEMI(.AI. WITH

293

MORPHOMETRIC D A TA

As one of its main potentials, morphometry offers the opportunity o f correlating quantative data from biochemical and morphological studies of the same material. So far, limited use has been made of this possibility. In their morphometric study on mitochondria, Baudhuin and Berthet ( 1967) have determined cytochronie oxidase activity but have used this information chiefly to establish the size of their sample with respect to the whole liver since they worked with subcellular fractions. Kimberg et d.(1968) have studied mitochondria1 respiration and oxidative phosphorylation of mitochondria after cortisone treatment in conjunction with a cytomorphometric study (‘Wiener et B., 1968). The mitochondria were found to be larger and less numerous than in controls, but the surface density of cristae membranes per cell remained normal. These morphological findings did not allow a straightforward interpretation of the observed decrease in oxygen consumption and uncoupling of oxidative phosphorylation on the basis of structural alterations. Phenobarbital treatment has been known to induce proliferation of smooth ER in liver cells and an increase in the activity of various microsomal enzymes. In a recent study, Staubli ef a/. (1969) have attempted a quantitative correlation of morphological with biochemical alterations. Rats were given phenobarbital for up to 5 days. Starting at 16 hours after the first dose the animals were sacrified at different time points; each liver was divided into two samples: one for morphometric electron microscopy, the other for fractionation and biochemical study of the microsomal fraction. This study established a linear relationship between the proliferation in membrane surface of the smooth ER and the increase in activity of three microsomal enzymes involved in drug metabolism. I n addition, it was shown that rough ER increased initially with a pronounced augmentation of ribosomal number, followed by regression to contrul dimensions after 2 days; this was interpreted as possibly being related to the synthesis of enzymes and structural membrane proteins in the rough ER, which were secondarily shifted to the smooth ER. It can be anticipated that this correlative approach to quantitative cell biology may play an increasing role when the concepts of molecular biology are carried from the microbiological level to that of the animal cell with its complex organization.

V. Cytomorphometric Methods in Experimental Pathology Pathological histology is increasingly availing itself of electron microscopy for precise localization of tissue damages. Here again, cytomorphometric methods can be most useful. Some studies on fine structural changes of the lung under the effect of pure oxygen breathing may serve as examples. Using electron micro-

294

EWALD R. WEIBEL

scopic morphometric methods, Kistler et ul. ( 1967) have demonstrated that pulmonary alveolar tissue of rats reacts to oxygen breathing with the initial formation of edema which causes a doubling of the air-blood barrier thickness. Kapanci e l ul. (1969) have then shown that similar events occur in monkey lungs although there are distinct species differences in the type of cellular reaction. In both studies, the quantitative changes in the tissue could be correlated with the functional impairment observed in the animals. Kapanci et al. (1969) have also quantitatively defined the restitution of tissue structure during recovery of oxygen-poisoned monkeys. It is evident that this type of study bears great potential for the establishment of quantitative structure-function correlations in human pathology also. Two further examples should be mentioned. Hollmann ( 1968) has compared cytomorphometric data of mammary cancer cells in the mouse with those from normal lactating tissue. H e found a significantly smaller density of organelles in the cancer cells. Poche et ul. (1968) have stereologically studied the volume ratio between mitochondria and myofibrils in rat myocardium in experimental hypertension. Myocardial hypertrophy led to a reduction of this ratio to less than 50% of the control value. These studies show that stereological methods can produce significant results in many as yet uninvestigated projects.

VI. Problems Arising in Applying Stereological Methods to Anisotropic Systems

By their nature, cells are anisotropic, for their function requires orientation with respect to related functional spaces: Gland cells extend between interstitial and luminal space, muscle cells between origin and insertion. Nevertheless, the aggregate of large numbers of anisotropic cells is very often isotropic in the stereological sense because preferential orientations of individual elements toward the section plane cancel out. This is, in fact, an essential prerequisite for the applicability of stereological methods to biological tissues. If anisotropy is not naturally eliminated-as is the case in surface epithelia, muscle, etc.-it has to be given explicit consideration in setting up stereological procedures. These relate to sampling and to choice and application of test systems.

A. SAMPLINGFROM ANISOTROPICTISSUES 1.

Distribiition of Sumple

It is plausible that all parts of an anisotropic system should be represented in the sample in proportion to their frequency, in other words, in proportion to their fractional volume. The cells of surface epithelia, for example, show a functional and structural polarity whose axis is perpendicular to the surface. In

MORPHOMETRIC CYTOLOGY

295

a first stage of stereological work the density gradient of structures along the axis of polarity has to be assessed. It should be noted that a line coinciding with the polarity axis will traverse the various layers in proportion to their relative volumes; this follows from Eq. ( 6 ) . T o this end the sample should extend through the entire depth of the cell layer; if a high magnification is required, such that one field cannot encompass the entire depth, a sequence of adjacent frames must be studied. Care should be taken that the frames do not overlap. If possible, point-counting volumetry should be used to establish the structural gradient since this method is not affected by any additional internal anisotropy of the cells resulting from preferential orientation of membranes or organelles with respect to the surface. The test points can be conveniently arranged in rows which are oriented parallel to the polarity axis. Such “columnar” sample units should be repeatedly taken along the surface, preferably at equal intervals in the sense of systematic sampling, unless an inherent periodicity demands random sampling to avoid bias. If tissue anisotropy results from sequential arrangement of clearly differentiated units of specialized structure and function, multiple stage sampling may be efficient: In a first low power stage the fractional volume of the different structural units is estimated, while subsequent higher power stages may center on specific units to obtain more precision in assessing their composition. The procedure of weighing the detailed results in terms of the total tissue will essentially depend on the specific conditions of the study. 2.

Orientatioii of Sectioits

The plane sections used for microscopic study are themselves anisotropic; this demands careful orientation of the section with respect to the axis of tissue anisotropy to assure an unbiased sample. In general, the plane of section should contain the axis of anisotropy; epithelia should therefore be sectioned perpendicular to the surface. Deviations from this rule may, of course, be indicated under special circumstances. If the tissue has several symmetry axes, as is the case in skeletal muscle, sections should be oriented parallel to each of them; in skeletal muscle longitudinal and transverse sections will suffice.

B. EFFECTOF ANISOTROPY ON STEREOLOGICAL MEASUREMENTS; OF TESTSYSTEMS INFLUENCE It has already been pointed out that anisotropy will only create problems if it somehow interferes with an anisotropic test system. Volumetry by differential point counting will hence be unproblematic if the section is an unbiased tissue sample. It may be appropriate though to use random point nets to avoid interference with periodic tissue structures. Random point nets with even dispersion can be obtained by marking one random point, for example, within each of

296

EWALD R. WEIBEL

25 squares of a square grid AS shown in Fig. 31. This also facilitates scanning of the field. The ordinary grids of parallel straight lines used for estimation of surface areas by intersection counts are strongly anisotropic. Particular care in their application is hence indicated if preferential orientation of the structural surfaces must be expected. There are several remedies. One of them is the use of an isotropic test line represented by a circle. Merz (1968) has proposed an isotropic test system composed of a sequence of semicircles which is easy to scan (Fig. 2 5 ) . Triangular test lines (Sitte, 1967) or short test lines arranged in equal number in three directions at 60' to each other (Weibel and Knight, 1964) have a comparatively low degree of anisotropy, as do square lattices of lines. Sitte ( 1967) has pointed out that satisfactory compensation for aniso-

FIG. 31. Grid with 25 random points. In each square the location of one point was fixed on the basis of two-digit random numbers read from a table.

tropic effects in oriented structures can be obtained if the lines of triangular or square lattices are oriented at fixed angles to the axis of structural anisotropy. In case of doubt, several readings in different orientation can be taken. It should be noted that estimation of Sv on sections cut parallel to the axis of orientation, as recommended above for sampling reasons, will yield a biased estimate. In general, it is necessary to take measurements on several sections of different orientation. Hilliard ( 1967c) has extensively discussed this problem; he also has shown that the use of a special elliptic test figure (Fig. 32) yields an unbiased estimate of Sv through Eq. ( 1 5 ) by counting intersection points with the surface trace on a single section parallel to the orientation axis, irrespective of the degree of anisotropy. If the structures are cylindrical or prismatic, Sv can also be estimated from exact cross sections. Counting the intersections I,, of the surface trace per unit length of an appropriate line grid we obtain

MORPHOMETRIC CYTOLOGY

297

where h is the length or height of the cylinders. It is evident that here the surface density is directly proportional to the profile contour length density B,4 as defined in Eq. ( l o ) . Similar restrictions and rules apply to the assessment of curve length density A4, and of numerical density N,. o f oriented structures. Hilliard ( 1 9 6 7 ~ )has also discussed the problem of ill,.. C. ASSESSMENT O F STRUCTURAL ANISOTROPY

It may often be of functional significance to assess the degree of anisotropy of cell structure. The most sensitive measures will be either Sv or M v . Any surface

FIG. 3 2 . Test figure for estimating 5,. of anisotropic structures. Constructed according to Hilliard ( 1 9 6 7 ~ ) .

can be split up into very small plane elements; the orientation of these elements is defined by two angles. In isotropic systems all angles are equally frequent, but in anisotropic systems there is a distribution function which depends on the degree of anisotropy (Hilliard, 1 9 6 7 ~ ) Since . the probability of intersection of a straight test line with a plane depends on the angle between the test line and the normal to the plane, anisotropy will also result in a distribution function of the number of intersections depending on the angle of orientation between test lines and axis of anisotropy. The number of intersections will be small (0 in perfectly parallel structures) when this angle is O o , and it will be largest when the angle is 90". This can be used to assess the degree of anisotropy within a structure. A rough estimate of the degree of anisotropy can be obtained by counting intersection densities I, on square grids by separately recording the number of intersections on the lines that are parallel to the axis of anisotropy ( I L o ) and on those that are perpendicularly oriented (/L9,,); the smaller the ratio ~ L O / I L Q O the higher the degree of anisotropy.

298

EWALD R. WEIBEL

Polmty of cells or subcellul~trorganelles is often defined by the proximity of parts of their surface to other structures. In the example of the liver cell discussed above, polarity was assessed by the fraction of cell surface abutting on sinusoid, biliary capillary, and neighbor cell; it was estimated from the ratio of respective intersection points with a square lattice (Weibel et al., 1967). Sitte (1767) estimated the proportions of basal, lateral, and apical surface of renal tubular cells by the same method. Dorfler (1967) has introduced the term “proximity parameter” to describe the degree of contact of one structure with another; this parameter is estimated by the relative size of the surface of immediate contxt.

VII. Appreciation of Present State and Outlook on Future Possibilities Methodical progress of recent years has made stereological techniques applicable to electron microscopic cytology without undue effort. In fact, there now exists a body of methods which permits quantitative investigation of almost any structural property of cells and tissues at any level of magnification required. The sources of error are defined and, to a large extent, practical methods for their correction or ‘ippteciation are available. The specific methods emphasized in this review are also quite efficient if they are properly applied. In our experience, estimation of up to 10 parameters on a sample of 30 to 40 electron micrographs can be performed in about 2 hours by using point-counting methods. This is a small effort when compared to the information produced. It should therefore be demanded that the pseudoquantitative descriptions still customary among morphologists be replaced by true morphometric data which can be statistically tested. This is of particular importance when morphological findings are to be correldted with biochemical or physiological information, or when pathological changes at the cellular level are to be quantitatively interpreted in terms of functional impairment of the organism. Stereological methods can also be applied to histo- or cytochemical studies. As shown by the group of Carpenter and Lazarow in various papers (1966, 1967; Lazarow and Carpenter, 1962) and by Leibnitz (1964) the amount of histochemical reaction product can be quantitated very easily by point-counting or linear analysis. Ross and Benditt (1965) have successfully used point-counting volumetry in conjunction with autoradiography to derive an index of specific labeling of cell components. A related method has recently been worked out by Williams (1768). In this field, use of stereological methods is only at its beginning. The coming y e m will probably bring new and extended approaches. I t may appear unfortunate that stereological methods still require active participation of the investigator in interpreting the micrographs and in deciding

MORPHOMETRIC CYTOLOGY

299

on the classification of points or measurements. The fault lies not with stereology, but rather with the characteristics of the electron image of sectioned biological material, in which organelles are recognized by their configuration and context only. It is not possible yet, however, to unambiguously differentiate cell components on the basis of electrun contrast. If cytochemical methods become available that can specifically and quantitatively enhance contrast of certain organelles so that contrast discrimination will suffice for their recognition on electron micrographs, then automatic scanning techniques can be applied to biological electron microscopy just as they are presently used in the materials sciences. It is quite likely that such methods will become available for some organelles in the not-too-distant future. For the present, however, efficient pointcounting methods and semiautomation of the analytic procedure must suffice. However, these methods are valuable tools which deserve to be introduced more widely into electron microscopic cytology.

ACKNOWLEDGMENTS In closing this review I wish to gratefully acknowledge the help and encouragement received over the last years i n developing the concepts presented. Without the mathematical collaboration of Drs. D. M. Gomez, €3. W. Knight, and H. Giger much would have remained rudimentary; and to Dr. G. E. Palade I owe thanks for a challenging introduction into cell biology. Much stimulation has also come from interdisciplinary discussions in the framework of the International Society for Stereology. I would like to thank Dr. Hans Giger for critical review of this article.

REFERENCES Aherne, W. (1967). J. Roy. Microsrvp. Sol-. 87, ‘193. Attardi, G. (1953). A r f a Anat. 18, 177. Bach, G. (1963). Z . 1ViJs. Mikroskopie 65, 285. Bach, G. (1967). Z n “Quantitative Methods in Morphology” ( E . R. Weibel and H. Elias, eds.), pp, 23-45, Springer, Berlin. Baudhuin, P., and Berthet, J. (1967). J. Cell Biol. 35, 611. Blichfeldt, H. F. (1914). Trans. Am. Math. Sor. 15, 227. Bockstiegel, G. (1967). I n “Stereology” (H. Was, ed.), pp. 193-194. Springer, New York. Buffon, G . (1777). Suppl. i 1’Histoire Naturelle, Vol. 4. Burri, P., and Weibel, E. R. (1968). Z . Zellforsrh. Mikroskop. ATrat. 88, 426. Burri, P., Giger, H., Gnggi, H. R., and Weibel, E. K. (1968). Pror. European Regional Conf. Electron Microscopy Rome, Vol. I, p. 593. Tipograha Poliglotta Vaticana, Rome. Carpenter, A. M., and Lazarow, A. (1966). J. Hisforhem. Cylochem. 14, 834. Carpenter, A . M., and Lazarow, A . (1967). In “Stereology”’ (H. Was, ed.). pp. 189190. Springer, New York. Chalkley, H. W. (1943). J. Natl. Cancer Inst. 4, 47. Chalkley, H. W., Cornfield, J., and Park, H. (1949). Srience 110, 295. Chayes, F. (1965). Lab. Invest. 14, 987. Clawson, C., Carpenter, A. M., Vernier, R., Hartman, J. F., and Latarow, A. (1958). J . Historhem. Cytorher. 6, 393.

300

EWALD R. WEIBEL

Cochran, W . G . (1953). “Sampling Techniques.” Wiley, New York. Cornfield, J., and Chalkley, H. W. (1951). 1. 1Y‘a.rh. A d . Sri. 41, 226. Coupland, R. E. (1968). Nature 217, 384. De Hoff, R. T. (1964). Trans. AIME 230, 764. De Hoff, R. T. (1967). In “Stereology” ( H . Elias, ed.), pp. 119-130. Springer, New York. De Hoff, R. T., and Rhines, F. N . (1961). Trans. AIME 221, 975. Delesse, M. A. (18,17). Compf. Rend. 25, 544. Dorfler, G. (1967). In “Stereology” ( H . Elias, ed.), pp. 211-212. Springer, New York. Duffin, R. J., Meussner, R. A,, and Rhines, F. N. (1953). Carnegjr I u s ~ . Tri-hnul., Repi. NO. 32, CIT-AF 8 A-1 R 32. Ebbesson, S. O., and Tang, D. B. (1967). I n “Stereology” ( H . Elias, ed.), pp. 131-132. Springer, New York. Elias, H., ed. (1967). “Stereology.” Springer, New York. Elias, H., and Hennig, A. (1967). In “Quantitative Methods in Morphology” ( E . R. Weibel and H. Elias, eds.), pp, 130-166. Springer, Berlin. Elias, H., and Pauly, J. E. (1966). ”Human Microanatomy.” 3rd Ed. Davis, Philadelphia, Pennsylvania. Endter, F., and Gebauer, H. (1956). Optik 13, 97. Eranko, 0. ( I955 ) . “Quantitative Methods in Histology and Microscopic Histachernistry.” Karger, Basel. Fischmeister, H . F. (1967). In “Quantitative Methods in Morphology” (E. R. Weibel and H. Elks, eds.), pp. 221-249. Springer, Berlin. Fisher, C. (1967). In “Stereology” ( H . Elias, ed.), pp. 285-286. Springer, New York. Forssmann, W . G., Siegrist. G., Orci, L., Girardier, L., Pictet, R., and Rouiller, C. (1967). J . Microci-upie 6, 279. Freere, R. H.. and Weibel, E. R. (1967). 1. Roy. Microscop. Sor.. 87, 25. Fuchs, A,, and Weibel, E. R. (1966). Z . Zellforsch. Mikroskop. Anat. 73, 1. Gander, R. H . (1967). I n “Stereology” ( H . Elias, ed.), pp. 289-290. Springer, New York. Giger, H . (1967). I n “Stereology” ( H . Elias. ed.), pp. 257-258. Springer, New York. Giger, H. (1969). To be published. Giger, H., and Erkan, Y . (1068). 2. 1Virs. M i k w s k o p i e 69, 36. Giger, H., and Riedwyl, H . (1969). To be published. Gil, J., and Weibrl, E. R. (1968). J. Ulwa.rtru1.l. Res. 25 (in press). Glagoleff, A. A . (1933). 7’ran.r. 1n.rt. Er-on. Minrral. ( U S S R ) 59. Gnagi. H. R., and Weibel, E. R. (1968). To be published. Hally, A. D . (1964). Qmarl. 1. Mirrosr-op. Sci. 105, 5 0 3 . Haug, H. (1967a). I n “Quantitative Methods in Morphology” (E. R. Weibel and H . Elias, eds.), pp. 58-78. Springer, Berlin. Haug, H. (1967b). In “Stereolopy” ( H . Elias, ed.), pp. 199-210. Springer, New York. Haug, H. ( 1 9 6 7 ~ ) Z. . Zellforsch. Mikroskop. Anat. 83, 265. Heiniger, H. J., Riedwyl, H., Giger, G., Sordat, B., and Cattier, H . (1967). Blood 30, 288. Hennig, A . (1956). Mikroskopie 11, 1. Hennig, A. (1957). Mikroskopie 12, 7. Hennig, A. (1967). In “Quantitative Methods in Morphology” (E. R. Weibel and H . Elias, eds.). pp. 99-129. Springer, Berlin. Hess, F. A. (1967). Unpublished results.

MORPHOMETRIC CYTOLOGY

30 1

Hess, F. A., Preisig, R.. Gnagi, H. R.. and Weibel, E. R. ( 1968). Proc. Europear2 Regional Conf. Electron Micvorropg. Rome. Vol. 11, p. 475. Tipografia Poliglotta Vaticana, Rome. Hilliard, J. 1:. (1965). It1 “Recrystallization, Grain Growth and Textures” pp, 267-286. Am. Soc. Metals, Metals Park, Ohio. Hilliard. J. E. (I967a). In “Stereology” ( H . Elias, ed.), pp. 195-196. Springer, New York. Hilliard, J. E. (1967b). 771 “Stereology” ( H . Elias, ed.), pp. 211-215. Springer, New York. Hilliard, J. E. ( 1 9 6 7 ~ ) .f u ”Stereology” ( H . iilias, ed.), pp. 219-230. Springer, New York. Hilliard, J. E., and Cahn. J. W . (1961). Trans. A f M E 221, 34.1. Hollmann. K. H. (1968). Z . ZellfotTCh. Mikioskop. Anat. 87, 266. Holmes. A. H . ( 1 9 2 7 ) . “Petrographic Methods and Calculations.” Murby. London. Horrikawa, E. ( 1 9 5 3 ) . T e / r r r T o Hagane 40, 991. Kapanci, Y., Weibel, E. R., Kaplan, H. P., and Robinson, F. R. (1969). Lab. fnt,rst. 20 (in press). Kimberg, D. V., Loud. A. V., and Wiener, J. (1968). J . Cell B i d . 37, 63. Kistler, G . S.. Cdldwell. P. R., and Weihel, E. R. (1967). J. Cell Biol. 32, 6 0 5 . Knight, B. W.. Weibel, 1:. R., and Gomez, D. M. (1963). f’ror. 1st f n t e ~ u .Corr‘qv. Steveol., I’ietitza, p. 18. Medical Academy, Vienna. Lazarow, A,, and Carpenter. A. M. (1962). J . Hhtol-hem. Cytoi-hem. 10, 324. Leibnitz, L. (1964). Hir/uc.henrie 4, 123. Lenz. F. ( 1 9 5 5 ) . Z . 1Vi.r.r. Mihvoskopie 63, 5 0 . Loud, A. V. (1962). J . Cell B i d . 15, 481. Loud, A. V. (1967). Pror. 2sth Meeling Elei-trot/ Mii-rori-o,Py Sol.. A m . p. 1.M.(C.J. Arceneaux, ed.) Claitor’s Book Store, Baton Rouge, Louisiana. Loud, A. V. (1968). J . Cell Biol. 37, 27. Loud, A. V., Barany, W . C., and Pack, B. A. (1965). Lab. fnve.rL. 14, 996. M e n , W. A. (1968). Mikroskopie 22, 132. Moor. H. (1964). Z . Zellforsch. Mikroskop. A m / . 62, 581. Moore, G. A . (1967). I N “Stereology” ( H . Elias. ed.), pp. 281-284. Springer. New York. Poche. R., de Mello Mattos, C. M.. Rembarz, H. W., and Stoepel, K. (1968). Arch. Pathol. Arrar. Physiol. Abt. A, 344, 100. Rosiwal, A. (1898). Verhandl. K.K. Grol. Reii-hjat/.rtali, W’ie72 p. 143. Ross, R., and Benditt, E. P. (1965). 1. Cell B i d . 27. 83. Saltykov. S. A. (1958). “Stereometric Metallography,” 2nd Ed. State Publishing House for Metals Sciences, Moscow. Saltykov, S. A. (1967). 7 ~ 2“Stereology” ( H . Elias, ed.). pp. 63-61, Springer. New York. Scheil. E. (19.35). Z . M e t a l l k . 27. 199. J Schwartz, H. A. ( 1 9 3 4 ) . Metals A ~ ~ J5,J 139. Sitte, H . (1967). 172 “Quantitative Methods in Morphology” (11. R. Weibel and H . Elias, eds.), pp. 167-198. Springer, Berlin. Sjostrand, F. S. (1967). “Electron Microscopy of Cells and Tissues.” Academic Press, New York. Smith, C. S., and Guttman, L. ( 1 9 5 3 ) . Tra72.r. Am. f m t . Minifzg Met. Petrol. Engrs. 197, 81.

Staubli. W., Weibel, E. R.. and Hess, R. (1969). J , Cell B i d . (in press). Tomkeieff, S. I. (1945). Nature 155, 24.

302

BWALD R. WElBEL

Underwood, E. E. (196,1). Slereologia 3, 5. Underwood, E. I!. (1967a). [n “Stereology” ( H . Elias, ed.), pp. 49-60. Springer. New York. IJnderwood, E. E. (1967b). Z n “Stereology” (H. Elias, ed.), pp. 79-80. Springer. New York. Weibel, E. R. (1963). “Morphometry of the Human Lung.” Springer, Berlin and Academic Press, New York. Weibel, E. R. (1967a). I n “Stereology” (H. Elias, ed.), pp. 15-26, Springer, New York. Weibel, E. R. (I967b). In “Stereology” (H. Elias, e d . ) , pp. 275-276. Springer, New York. Weibel, E. R. ( 1 9 6 7 ~ ) .I n “Stereology” (H. Elias, ed.). pp. 330-331. Springer, New York. Weibel, E. R., and Elias, H. (1967). “Quantitative Methods in Morphology.“ Springer, Berlin. Weibel, E. R., and Gnagi, H. R. (1968). Puoc. European Rrgioual Cowgr. Elel-tron Micro.rropy, Rome, Vol. I , p . 601. Weibel, E. R., and Gomez, D. M. (1962). J. Appl. Physiol. 17, 343. Weibel. E. R., and Knight. €3. W. (1964). J. Cell B i d . 21, 367. Weibel, E. R., Kistler, G. S.. and Scherle, W . F. (1966). J. Cell B i d . 30, 23. Weibel, E. R., Stzubli, W., Gnagi, H. R., and Hess, F. (1969). J. Cell B i d . ( i n press). Wicksell, S. D. (1925). Biomelriku 17, 84. Wicksell. S. D. (1926). Bioviet~iRa18, 151. Wiener, J., Loud, A. V., Kimberg, D. V . , and Spiro, D. (1968). 1. Cell B i d . 37, 47. Williams, M. A. (1968). Proc. Roy. Microscop. Sac. 3, Pt. 2. Wischnitzer, S. ( 1 9 6 7 ) . Intern. Rerj. Cytol. 22, 1.

Some Possible Roles for Isotymic Substitutions during Cold Hardening in Plants D. W. A ROBERTS Rerear: h Station. Caiiada Depattmeti~of Agri: ultuie Lethbiid,te, Alheitn, Canada I. Introduction . . . ... ..................... 103 Effects and Prev of I ation . . . . . . . . . . . . . . . . 304 A. Extracellular Ice . . . . . . . . . . . . 304 B. lntracellular Ice . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 111. The Effect of Low I’eniper;iture on Proteins . . . . . . . . . . . . . . 309 A . Description of Effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109 B. Protection of Proteins against Low-Temperature Injury 311 IV. Metabolic Imbalance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313 A . Effects of Temperature Changes on the Regulatory 113 Machinery of the Crll . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Effects of Temperature on the Rate of Enzymic Reactions 314 C. Prevention of Metabolic Imbalance . . . . . . . . . . . . . . . . . . 31 6 318 V. T h e Hypothesis of Isozymic Substitution . . . . . . . . . . . . . 318 A. Nature of the Hypothesis . . . . . . . . . . . . . . . . . . . . . . . . B. lsozymic Substitutions and Cold Hardening . . . . . . . . . . 319 C . Problems Associated with Testing the Hypothesis of lsozymic Substitution . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121 D. Implications of the Hypothesis of Isozymic Substitution 121 VI. Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122 References . . . . . . . . . . . . . . . . 323 11.

I. Introduction As pointed out earlier (Levitt, 1962), there is a pressing need for new approaches to the problem of cold hardening and cold hardiness in higher plants. This article reviews the subject of cold hardiness to indicate where and how the substitution at low growth temperatures of ‘I modified form of a protein for the form normally present at higher growth temperatures would be advantageous for the plant. A plausible mechanism for making such substitutions will be presented. Some of the few known possible examples of such substitutions will be discussed critically to indicate the experiments that must be performed to prove that the proposed hypothesis is operative. Since very Iittle work along these lines has been done with higher plants, the plausibility of many of the proposals will be supported with evidence from microorganisms and poikilotherms. It is assumed that the same basic mechanisms of cold resistance and adaptation operate in plants as in microorganisms and poikilotherms. It is to be expected that the relative importance and details of these mechanisms will vary from one group of organisms to another. 301

304

D. W . A. ROBERTS

Plants face a number of difficulties as the temperature drops. Ice forms if the temperature falls low enough. Proteins undergo reversible or irreversible conformational changes at low temperatures. Metabolic imbalance may arise and may either be detrimental or lead to a remodeling of the biochemical system. It is suggested that cold hardening is such a remodeling, which enables the plant to counteract partially some of the adverse effects of low temperatures. This remodeling may, in part, consist of the substitution of a modified protein for that form of the protein that carries out the same function at higher growth temperatures. These substitutions may result from masking and unmasking different portions of the genome. 11. Effects and Prevention of Ice Formation

Ice formation may occur in plant tissues in two types of locations, extracellular or intracellular. Extracellular ice formation is not necessarily lethal whereas intracellular ice formation usually is lethal (Asahina, 1956; Levitt, 1958). Although intracellular ice formation has rarely been observed in nature (Levitt, 1956), it may occur in nonhardy cells frozen very slowly (Asahina, 1956) and even in hardy cells at low temperatures following extracellular freezing (Tumanov and Krasavtsev, 1959). Since ice, which forms inside the cell wall but outside the protoplast and vacuole, probably produces effects partly similar to extracellular ice and partly similar to intracellular ice, it will not be discussed. Ice formation in living organisms has been recently reviewed (Levitt, 1956, 1966; Mazur, 1966). A. EXTRA<:ELI.IILAR ICE Extracellular ice forms in the intercellular air spaces in higher plants (Levitt, L956). When the temperature falls sufficiently below the freezing point, the extracellular water freezes but the intraprotoplasmic water usually does not freeze. As long as the vapor pressure of the water in the protoplasts is greater than that of the ice, water will leave the protoplasts if possible and subsequently freeze in the intercellular spaces thus initiating the growth of ice there. This ice growth will have two effects, desiccation of the protoplasts and, if the crystals grow large enough, mechanical deformation of the cells (Meryman, 1957). The desiccating effects of the formation of extracellular ice may in part explain the correlation of drought resistance and cold resistance which is observed in many but not all plants (Levitt, 1956). The desiccation of the protoplast will greatly increase the concentration of the substances dissolved in the aqueous phase within the plant cell. Such changes may induce great increases in the rates of chemical reactions much as they may do in partially frozen solutions (Pincock and Kiovsky, 1966). Some of these reactions may be detrimental to the plant.

ISOZYMES A N D COLD H A R D E N I N G

305

The sulfhydryl-disulfide hypothesis (Levitt, 1962) may exemplify a special case of the more generalized situation considered here. The concentration changes themselves will increase salt concentration and may result in the precipitation of some substances and possibly in changes in pH. Proteins may be denatured by exposure to high salt concentrations or to pH changes if these are sufficiently large. Certain types of mechanical deformation of plant tissues are known to increase respiration rates (Audus, 1935, 1939; Roberts, 1951). The mechanism of this effect is now under investigation (Bagi and Farkas, 1967). Perhaps mechanical deformation damages the protoplast. Such damage could contribute to intracellular ice formation. Plants might protect themselves from damage due to extracellular ice formation by blocking the loss of water. Such an action would increase the chance of intracellular ice formation and be detrimental. Furthermore, there are many suggestions in the literature that increased rather than decreased membrane permeability accompanies cold hardening (Levitt, 1956). Plants might protect themselves from the effects of high salt concentrations produced by desiccation by replacing the normal proteins with slightly modified proteins having greater resistance to this type of denaturation. Such modified forms of proteins have been discovered in animals (Warren and Peterson, 1966). Protection from the other deleterious effects of desiccation such as increased concentrations of other metabolites and soluble proteins may also be possible by making appropriate protein replacements.

B. INTRACELLULAR ICE The formation of intracellular ice is usually considered lethal although the reasons for this are unknown. Among the possible causes are (1) protein denaturation (see Section 111); ( 2 ) increased rates of chemical reactions such as those occurring in frozen systems iiz vitvo; ( 3 ) mechanical damage resulting from disruption of the very elaborate structure within the cells by expansion of water on freezing. Since little is known about mechanical damage it will not be discussed further. Quite large increases in the rates of cheinical reactions have been observed in frozen solutions. Thorough kinetic investigations have shown that some of these rate increases are the result of increasing the concentrations of reactants in tiny liquid pockets in the ice (Wang, 1961; Kiovsky and Pincock, 1966; Pincock and Kiovsky, 1966). The possibility remains that some rate changes may arise from other factors (Grant et ul., 1966). This possibility needs very thorough kinetic study since it is important to know whether or not factors other than concentration of reactants in liquid pockets and the formation of dry surface films (Wang, 1961) are involved in the increase in reaction rate in frozen

306

D. W. A. ROBERTS

solutions. The increases observed are great enough that reactions occurring at negligible rates in the liquid before freezing may occur in the frozen liquid at appreciable rates. So, it is likely that damage in frozen tissue may result from chemical reactions that are too slow at higher temperatures to produce deleterious effects. This is one reasonable explanation for the damage known to occur when frozen tissues are kept under conditions of freezing equilibrium (Levitt, 1956, 19%).

Intracellular freezing could be prevented if the water in the cell could ( 1 ) be maintained in a supercooled condition, ( 2 ) be prevented from freezing at the temperatures that are encountered, or ( 3 ) freely leave the cell. Maintaining water indefinitely in the supercooled condition at subfreezing temperatures requires the prevention of nucleation. Nucleation may originate outside or inside the protoplast. Only extraprotoplasmic ice will be formed by extraprotoplasmic nucleation if there is a barrier that prevents extraprotoplasmic ice from nucleating the intraprotoplasmic water. The plasma membrane may provide such a barrier (Chambers and Hale, 1932). There is probably also a barrier that prevents ice formation in one protoplast from nucleating the water in adjoining protoplasts. The maintenance of these barriers at low temperatures will be discussed later together with modifications in their permeability. The chance of nucleation within the protoplast could be reduced if the membranes of proteins and lipids within the cell were modified to reduce the efficiency of the nucleating sites. One method of modification would be to substitute at low temperatures a new form of a protein for the form normally present at higher temperatures. Another method would be to change the fatty acid composition of the membranes with changes in growth temperature (Howell and Collins, 1957; Marr and Ingraham, 1962). Either or both of these methods would change the properties of the membranes by altering, perhaps only slightly, the conformation of the membrane surface and consequently its efficiency as a nucleating agent. That surface films of some long-chain compounds, including some proteins, are relatively inactive as nucleating agents (Evans, 1966a) may partially explain the ability of protoplasts to supercool. Unfortunately, the identity of those nucleating agents that may exist in cells is unknown (Salt, 1958). The freezing point of water can be lowered by increasing the concentrations of the substances dissolved in it or by converting it into bound water. Plants often take advantage of the first of these phenomena by increasing the concentrations of protein protectants (see Section III,B) and possibly other compounds of low molecular weight inside their cells during hardening. A consideration of the role of bound water in cold hardiness requires an understanding of the structure and role of water in the living cell. Unfortunately, the structure of liquid water and the mechanism involved in its

ISOZYMES A N D COLD HARDENING

307

nucleation and freezing in vitvo, much less in vivo, are still being debated (Frank and Wen, 1957; Bernal, 1965; Frank, 1965; Evans, 1966b; Falk and Ford, 1966; Wicke, 1966). Important advances in understanding the behavior of water within cells must await the solution of these problems. Water in the cell probably exists as free water, a thin layer spread over the macromolecular surface, and in fine pores. Water may be involved in holding together some layers of the membranes within a cell (Hechter, 1965a). Some of this water present as thin films or capillary columns may have drastically modified properties, including a lowered freezing point (Hori, 1956; Mazur, 1960, 1966; Derjaguin, 1966; Meryman, 1966). Some of this water may be bound and may exist as water in tissues down to -60" or -70°C. (Wood and Rosenberg, 1957; Sussman and Chin, 1966). If so, its structure and properties are still uncertain as are the environmental factors responsible at the molecular level for its existence. In addition to the dubious older evidence for bound water in tissues (Levitt, 1956), a number of independent modern techniques suggest the existence of bound water both inside and outside living cells (Klotz, 1958; Vasil'eva ef ul., 1964; Ling, 1965; Schwan, 1965). Some of these techniques may prove useful in comparing the water-binding ability of proteins and membranes from cold-hardened plants with those of similar components from coldsensitive plants. A few attempts have been made to use some of these methods to study the water within living cells (Hopkins, 1960; Cerbon, 1964; Verzhbinskaya and Sidorova, 1964; Koga et nl., 1966). Difficulties both with interpretation and technique (Kowlasky and Cohn, 1964) are great and as yet there are no simple thoroughly reliable methods for the study of bound water in living cells. It has been pustulated that the ability of tissues to bind water is correlated with their cold hardiness (Levitt, 1956, 1966). Proteins and conjugated proteins may be responsible for binding some of this water. There have been suggestions that proteins of widely differing structure have quite different amounts of bound water associated with them (Berendsen and Migchelsen, 1965). If they do, small changes in structure of a protein might cause changes in its ability to bind water. Substitution of a modified protein with high water-binding capacity at low growth temperatures for one with lower water-binding capacity normally present at higher temperatures might reduce the amount of water that freezes. If such a development were carried to the extreme, then all the water required for viability might be in the bound form. This may be the case in dry viable seeds. So, it might be advantageous for a plant to lose all the free water from its cells, retaining only sufficient bound water to maintain viability. This loss may occur during the formation of extracellular ice if the plasma membrane is sufficiently permeable to water. This may explain why many cells show a correlation between increased permeability and cold hardiness (Levitt, 1956). If increased permeability of the plasma membrane to water accompanies cold

308

D. W. A. ROBERTS

hardening and serves to protect the cell from intracellular ice formation, and if the plasma membrane prevents extracellular ice from nucleating the water in the protoplast, then factors influencing the structure and stability of the plasma membrane must be important in cold hardening. The plasma membrane apparently consists of proteins, lipoproteins, and lipids. Unfortunately, the details of the arrangement of these components within the membrane are still a matter for debate (Hurry, 1964; Hechter, 196513; Korn, 1966; Wallach and Zahler, 1966). It is not yet possible to relate the structure of the plasma membrane to its permeability to water or to its stability to cold. Evidence is accumulating that some of the proteins of the plasma membrane function in the transport of inorganic ions such as N a + and K + across the membrane (Skou, 1965; Baker, 1966), while others are involved in the transport of sugars such as galactose (Fox and Kennedy, 1965). The identification and isolation of such proteins opens up the possibility for the study of their low-temperature stability and the effects of concentrated solutions of other metabolites (including salts and protective substances) on their stability. In some cases, presumably intact membranes can be isolated and studied. Comparative studies of the properties of such proteins and membranes from cold-hardened and cold-susceptible organisms could prove informative. While it has been known for a long time that low-temperature injury and death result in the loss of the semipermeability of the membranes of living cells, recent studies have produced evidence that the plasma membranes of some organisms are not stable at low temperatures and that the protein complements of the plasma membranes of other organisms may be modified by the environment. Even partial loss of membrane semipermeability would permit partial mixing of metabolites with each other or with enzymes from which they are normally separated by membranes. Such mixing could produce additional damage to the cells involved. Changes in cell permeability that result from chilling have been observed (Strange, 1964; Ring, L965a,b). Cold shock in bacteria may be an example of this (Strange and Postgate, 1964). Sometimes cold shock may be reduced by the presence of a protein protectant such as sucrose in the external medium (Meynell, 1958). The protective action of some nonpenetrating compounds may serve to reduce this cold sensitivity. If rigorous proof of the failure of these compounds to enter the cells that they protect can be provided, then the plasma membrane is a site of injury. This need not be true generally. The properties of isolated erythrocyte membranes are modified by freezing (Scharff and Vestergaard-Bogind, 1966). Isolated preparations of the ATPase involved in the transport of N a + and K + across the membranes of rat brain microsomes are coldsensitive (Gruener and Avi-Dor, 1966). A comparable enzyme in beef heart mitochondria is protected from inhibition by cold in vivo by its association with another protein (Pullman and Monroy, 1963). Few data are available on the

ISOZYMES AND COLD HAKDENING

309

comparable enzyme o r enzymes in plant cells (Dodds and Ellis, 1966; Pitman and Saddler, 1967). Evidence for changes in the protein complements of the membranes of organisms is derived from studies on galactose induction in bacteria. These studies indicate the existence of a galactose permease in the plasma membrane of induced cells but not in the plasma membrane of noninduced cells. Evidence for the association of a protein with galactose permease in bacteria has recently been obtained (Fox and Kennedy, 1965). This result shows that the proteins present in the plasma membrane may chmge, depending on the medium used to support growth. If the protein complement of the plasma membrane can change with the medium used for growth, it is reasonable to postulate similar changes with changed growth temperatures and comparable changes in the other membranes of the cells. Such changes may occur in higher plants during cold hardening and provide a basis for the increased permeability and increased cold stability of the plasma membrane during cold hardening. Some of these changes may involve the substitution of one protein for another that serves the same function but has a slightly different structure. The changes in the properties of the ouabain-sensitive ATPase of the intestinal mucosa of goldfish during cold acclimation suggest that it undergoes such a substitution (Smith, 1967). 111. T h e Effect of Low Temperature o n Proteins

A.

I)ESCRIPTION

OF

EFFECTS

Although many proteins are apparently unaffected by low temperatures, some are denatured when they are frozen in solution while others are denatured when they are chilled in solutions even though no ice forms. Examples of proteins that are denatured when their solutions are frozen h z vitro are known (Nord, 1936). Some lipoproteins show this effect (Bornstein, 1953; Lovelock, 1957). This may be significant since lipoproteins are believed to be important constituents of cell membranes. Other types of proteins are also denatured when their solutions are frozen (Leibo and Jones, 1964). Chemical changes resulting from freezing red blood cells have been studied (Chanutin and Curnish, 1966). The products of denaturation of the proteins of red blood cells have been separated electrophoretically from the undenatured protein. These changes were not observed after freezing at -75°C. although they occurred after freezing at -13" to -20OC. Storage time increased the quantities of denatured protein in red cell hemolysates. Some of the proteins from winter wheat are coagulated by freezing temperatures (Heber, 1959). Sucrose (Ullrich and Heber, 1958) and other protein protectants may have a protective action against such phenomena. The mechanism of these denaturation phenomena is unknown. The phe-

310

D. W. A. ROBERTS

nomenon of increased reaction rates in frozen solutions may be involved (see Section 11,B). Contact of the protein with hexagonal ice may be all that is required to produce denaturation (Shikama, 1963) since the native configuration of a protein is greatly influenced by the structure and composition of the surrounding medium (Kauzmann, 1959; Reithel, 1963). O n the other hand, changes in labile bonds (e.g., hydrogen bonds) may be involved (Leibo and Jones, 1964) and result in changes in protein conformation. Some enzymes dissociate and lose activity reversibly at low temperatures even though ice is absent (Penefsky and Warner, 1965). Other enzyme proteins apparently undergo reversible or partially reversible changes in conformation when their solutions are chilled (Numa and Ringelmann, 1965; Jarabak et al., 1966; Massey et al., 1966). These changes probably explain the considerable increase in the energy of activation that occurs with several enzymes (Roberts, 1967b) at low temperatures. Changes in the energy of activation of enzymic reactions at low temperatures would modify the metabolic balance at low temperatures. Such changes might be detrimental or might be used advantageously by the plant for metabolic adjustment. Another result of conformational changes is the reversible or irreversible inactivation induced by low temperatures. In some cases reversible inactivation is followed by slow irreversible changes which result in loss of activity of enzymes on prolonged exposure to low temperatures (Penefsky and Warner, 1965; Jarabak et ul., 1966). In the case of the obligate psychrophile, Vibvio marinus, purified malic dehydrogenase undergoes reversible low- and high-temperature inactivation in vitvo although no evidence for low-temperature inactivation was found in vivo (Langridge and Morita, 1966). The presence of ammonium sulfate in vitvo stabilizes this enzyme. The environment of this protein greatly affects its stability. Above 2 O T . this enzyme is inactivated and the organism will not grow. Mutant and isozymic forms of enzymes and proteins differing in their ability to withstand low temperatures are known (Fincham, 1957; Zondag, 1963; Hultin et ul., 1966). In Neurosporu a cold-sensitive mutant form of glutamic acid dehydrogenase interferes with metabolic processes below 20°C. (Fincham and Pateman, 1957). While the examples of enzymes cited have reduced stability at low temperatures, enzymes with increased stability may exist. Such modified enzymes might be utilized by cold-hardy plants to assist them in withstanding low temperatures. One facet of cold hardening would then be the replacement of a cold-sensitive form of an enzyme with a cold-resistant one. A comparable theory involving heat-labile enzymes has been postulated to explain the behavior of races of AvubidopJis thulium at high temperatures (Langridge and Griffing, 1959). In this species the enzyme substitutions appear in different races of a species rather than in the same race of a given species in response to changed growing conditions. Support for this theory comes from

ISOZYMES AND COLD HARDENING

311

work on temperature-sensitive mutants of tobacco mosaic virus. Mutants with known differences in amino acid sequences in their coat proteins differ in heat stability both i ~ zuitro and in vjvo (Jockusch, 1966). O F PROTEINS AGAINST LOW-TEMPERATURE INJURY B. PROTECTION

I n v i f w studies suggest that enzymes may be protected from low-temperature inactivation by using suitable protectants such as glycerol or sucrose (Ullrich and Heber, 1958, 1961; Shikama and Yamazaki, 1961; Chanutin and Curnish, 1966; Jarabak et al., 1966). Glycerol, sucrose, and other nontoxic polyhydroxy compounds also protect proteins in vitrn from denaturation by heat (Beilinson, 1929; Kiermeier and Koberlein, 1957; Jarabak et d.,1962; Yasumatsu et ul., 1965) and by urea (Jarabak et al., 1966). Glycerol protects isolated lactic dehydrogenase from radiation damage (Lohmann et ul., 1964). Glycerol (Sumner and Somers, 1947; Meyerhof and Ohlmeyer, 1952; Langer and Engel, 1958) and sucrose (Potter, 1955) are often used in enzyme extraction and in preparing enzymically active particulate fractions (Axelrod, 1955; Gorham, 1955; Hogeboom, 1955). They are supposedly used to produce isotonicity although it is not certain that this is their only or even their chief means of protecting the enzymes and proteins of particulates from denaturation. Some of the cases cited suggest that glycerol and sucrose act as protein protectants regardless of their osmotic effect. Unfortunately, the mechanism of these effects is unknown (Jarabak et al., 1966). These same compounds in uiuo protect organisms from damage caused not only by low temperatures (Polge et nl., 1949; Luyet and Keane, 1952; Lovelock, 1954; Perkins and Andrews, 1960; Trunova, 1964) but also by high temperatures (Molotkovskii and Zhestkova, 1964) and radiation (Vos, 1965). The protective action of these compounds may well explain the advantage to insects of accumulating glycerol (Salt, 1961 ) and to plants of accumulating sugar (Levitt, 1956; Parker, 1963). Modifications of the metabolism of an organism at lower temperatures to increase the accumulation of a suitable protein protectant is no doubt another facet of cold hardening. Such modifications could be brought about by isozyniic substitutions which augment the natural effect of temperature. Accumulation of compounds of low molecular weight would also result in a lowered freezing point of the cellular liquids in which such accumulation occurred. I think that this lowering of the freezing point is of secondary importance to protein protection. This would explain why relatively low concentrations are more effective in some cases than would be predicted on the basis of the ability of such concentrations to reduce the freezing point. Zfz vivo, dimethyl sulfoxide apparently has a protective action similar to that of glycerol and sugars (Lovelock and Bishop, 1959; Sherman, 1964; Bouroncle, 1965). In the few cases that have been investigated, dimethyl sulfoxide has

312

D. W. A. ROBERTS

been found capable of protecting proteins from low-temperature inactivation (Chilson et al., 1965; Graves et al., 1965; Chanutin and Curnish, 1966) and radiation damage (Lohmann et al., 1966). It also protects animals against radiation damage (Vos, 1965). However, dimethyl sulfoxide is metabolized by animal tissues (Williams et al., 1966) and so may not be the compound active in protein protection iz vivo. Furthermore, it may increase membrane permeability in vivo (Kligman, 1965; Altland et al., 1966; Fowler and Zabin, 1966; Hellman et al., 1967) and this would help to prevent intraprotoplasmic freezing. There are examples in which glycerol, sucrose, or dimethyl sulfoxide do not protect against cold (Hollander and Nell, 1954; Taylor and Gerstner, 1955; Terumoto, 1965; Wang and Marquardt, 1966). The differences in protective action do not vitiate the concept that these compounds act as antifreezes where (1 ) it has been shown that the compounds that do not work are toxic whereas the effective compound or compounds are not toxic or ( 2 ) the effective compounds penetrate the cells but the ineffective ones do not. If such demonstrations fail, the antifreeze concept must be rejected for specific cases. On the basis of their protective effects at different rates of freezing, it has been suggested that the mechanism by which glycerol and dimethyl sulfoxide work differs from that of sugars (Rapatz and Luyet, 1965). Perhaps further studies on the protein-protectant effect of these compounds may explain these discrepancies. Different sugars differ in their protective effect on higher plants (Tumanov and Trunova, 1957; Perkins and Andrews, 1960; Trunova, 1964). Only those sugars that enter the plant cells and are subsequently metabolized are effective in increasing frost resistance (Trunova, 1964). If true, this result indicates that sugars do not protect higher plants by lowering the freezing point of the cell fluids or by serving as protein protectants. More detailed information is required on the ability of the different sugars to penetrate plant cells, to act as protein protectants, and to serve as metabolic regulators through catabolite repression (Magasanik, 1961; Maas and McFall, 1964) and other induction and repression phenomena (Glasziou et al., 1966; Marri. et al., 1965). A number of other compounds have protective action if, vivo. These compounds seem to be characterized by their high affinity for water (Nash, 1966). They form strong hydrogen bonds with both themselves and water (Mazur, 1966). This group of compounds, which includes dimethylformamide, dimethylacetamide, and N-methylpyrrolidinone do not appear to have been tested as protein protectants iiz v h o . They deserve testing on cold-labile enzymes. A second group of protectants that work at quite low concentrations includes maleic hydrazide (Sxkai, 1957; Gaskins, 1959; Stewart and Leonard, 1960; Hendershott, 1962), 2-chloroethyltrimethyl ammonium chloride (CCC or CYCOCEL) (Wiinsche, 1966), and N-dimethylaminosuccinamic acid (B9 or B995) (Marth, 1965) and dormin (Irving and Lanphear, 1968). The ability

313

ISOZYMES A N D COLD HARDENING

of these compounds to increase cold resistance has been little investigated and their mode of action is currently unknown. This group of compounds is effective at such low concentrations that they might behave ;is hormones triggering the production of proteins that aid in cold resistance or terminating the production of proteins inimical to cold hardiness. An additional possibility is that gibberellins are the hormones and that a reduction in their concentration or effectiveness is associated with cold hardening. This possibility is suggested by the natural dwarfing often associated with cold hardening (Levitt, 1956) and the known physiological effects of dormin and CCC (Anderson and Moore, 1967; Khan and Faust, 1967).

IV. Metabolic Imbalance Changes in temperature may affect the relative rates of turnover of different enzymic pathways by at least two mechanisms, namely, changes in allosteric inhibition and changes in the rate of individual enzymic reactions with temperature. A consideration of these disruptive effects suggests possible mechanisms for overcoming them. One mechanism involves substituting one form of a protein for another. The final section of this article considers this possibility in relation to cold hardening and cold hardiness. CHANGESO N A. EFFECTSO F TEMPERATURE REGLJLATORY MACHINERY O F THE CELL

THE

Allosteric inhibition (Monod and J x o b , 1961; Jacob and Monod, 1963; Monod et al., 1963) is believed to be an important factor in regulating metabolism in bacteria by feedback inhibition of metabolic pathways (Umbarger, 1961) and by induction-repression phenomena (Jacob and Monod, 1961). Both of these mechanisms probably operate in higher plants (Umbarger, 1963). Some enzymes involved in feedback inhibition show changes in sensitivity to inhibition with changes in temperature (Taketa and Pogell, 1965; Bailin and Lukton, 1966). Mutant forms of enzymes subject to feedback inhibition occur which differ not only in sensitivity to allosteric inhibition but also in the effect of temperature on this sensitivity. This phenomenon has been proposed as an explanation for the low-temperature requirements for histidine for growth by mutants of Escheril-hid r-di (O’Donovan and Ingraham, 1965). Such a situation could create a serious metabolic upset resulting in slow death in higher plants. This phenomenon suggests the possibility of switching from one isozymic form of an enzyme subject to feedback inhibition to another to adjust the metabolic balance with changes in temperature. Comparable phenomena may occur in the regulatory machinery of induction and repression, since temperature-sensitive alleles of regulator genes are known

3 14

D. W. A . ROBERTS

(Horiuchi and Novick, 1961; Gallant, 1962; Sussman and Jacob, 1962). Some of these are constitutive at high temperature and inducible at lower temperatures (Horiuchi et ul., 1961; Udaka and Horiuchi, 1965). Others are inducible or constitutive at low temperatures (Gartner and Riley, t965). Changes in apparent inducibility could also result from changes in quantity of repressor present at different temperatures (Halpern, 1961 ) . These phenomena may explain the disappearance of dextransucrase and its replacement by invertase in a strain of Luctobucillw when grown at temperatures above 37°C. (Dunican and Seeley, 1963). The changed structure of the polysaccharides that accumulate in O~cill'ztoriu grown at temperatures below 5°C. may be a result of changed relative rates of synthesis of phosphorylase and the branching enzyme (Fredrick, 1953). The failure of some strains of Newo.rpom to produce tyrosinase at 35OC. when they do produce it at 25°C. (Horowitz and Shen, 1952) may be a further example of the effect of temperature on the regulatory machinery of the cell. The genetics of this effect is complex and several regulatory phenomena may be involved (Horowitz and Fling, 1953). B. EFFECTSO F TEMPERATURE ON T H E RATE OF ENZYMICREACTIONS The rate of enzyrnically catalyzed reactions falls with declining temperature. The rate of fall differs for each enzyme (Sizer, 1 9 4 3 ) . Consequently, as the temperature drops the pool sizes of the metabolites that accumulate changes. Consequently, the drop in temperature might (1) so decrease the rate of a reaction that damage or death to the organism will result from a shortage of the products of the reaction, ( 2 ) induce accumulation of toxic amounts of substrates of some enzymes, ( 3 ) change the pool size of metabolites that serve as metabolic regulators or protein protectants. Unfortunately, cases of chilling injury have not been analyzed in sufficient detail to be placed in one or other of these categories. Chilling injury must, therefore, be treated in general terms. However, because changes in pool size have possible interesting theoretical consequences, this subject will be discussed separately. Since the continued life of a cell o r whole organism depends on the proper functioning of a very complex, delicately balanced, integrated network of chemical reactions, the further the temperature departs from the normal operating range of the system the more likely it is that this network of chemical reactions will be upset. Metabolic imbalance should, therefore, become more important with falling temperatures. There are a number of examples of chilling injury that occur above the freezing point but very few examples of such damage below the freezing point. Because this does not appear reasonable a search should be made for damage resulting from metabolic imbalance below the freezing point. It is quite likely that metabolic imbalance is one of the factors involved in death by cold in the field, even in plants in temperate climates.

ISOZYMES A N D COLD HARDENING

315

Specific responses at low temperatures of tomatoes to nicotinic acid, cosmos to B vitamins, and eggplant to mixed ribosides (Ketellapper, 1963) suggest that the rates of some reactions relative to others may be reduced sufficiently by low temperatures to interfere with growth. The lowest temperature used in these studies was 10°C. Experiments using lower temperatures should be undertaken. A number of examples of low-temperature injury occurring above the freezing point are known (Sellschop and Salmon, 1928; Pentzer and Heinze, 1954; Lieberman et al., 1958; Harrington and Kihara, 1960; Youngner, 1961; Kislyuk, 1964a,b). Although their mechanism has not been fully documented, these effects probably result from metabolic imbalance. For example, low-temperature breakdown in apples is associated with the accumulation of oxalacetic acid (Hulme et al., 1964). In their early stages, some of these processes can be reversed by raising the temperature. Death or injury occurring above the freezing point probably usually can be explained on the bases of this sort of metabolic imbalance, the effects of low-temperature conformational changes in proteins, and interference with the regulatory machinery of either feedback inhibition or the induction-repression mechanism (Ng et al., 1962). Some experiments on bacteria suggest that metabolic injury may occur at subzero temperatures (Straka and Stokes, 1959; Macleod et al., 1966; Moss, 1966). Among higher plants, tobacco callus furnishes another example (Das et d., 1966) of injury below the freezing point that is apparently not caused by ice formation. In this case, lethal injury required an appreciable time at -10°C. and did not manifest itself until after rewarming and partial reformation of the cytoplasmic strands, which had disappeared during the chilling. The observat'ions on cell devision in this tissue at above-freezing temperatures could easily be explained on the basis of metabolic imbalance occurring in the j0-12"C. temperature range. Death would be expected on long exposure to lower temperatures. Changing the pool size of compounds that art as metabolic regulators would cause feedback inhibition and mechanisms similar to induction and repression to come into play (Dennis and Coultate, 1966). Such changes might also permit the operation of other regulatory machinery (Monod and Jacob, 1961; Korner, 1966) such as one possibly involving histones (Banner and Huang, 1962; Bonner et d., 1963; Huang et d,,1964) or one operating at the ribosomal level (Kerr et al., 1966). Induction and repression appear to operate in higher plants (see Roberts, 1967b, for examples). The operation of these mechanisms would tend to have a homeostatic effect increasing the concentration of the enzymes of pathways that are producing relatively insufficient product and decreasing the concentration of the enzymes of pathways that are producing relative excesses of product. In spite of these mechanisms, changes do occur in the concentration of metabolites that accumulate. T h e temperature-dependent starch-sugar (Levitt,

316

D. W. A. ROBERTS

1956) and glycogen-glycerol (Asahina, 1966) interconversions in plants and insects, respectively, are pertinent examples. Ascorbic acid accumulates in plants at low temperatures (L’vov and Altukhova, 195 1; Areshidze and Podrazhanskaya, 1956; Franke, 1957; Andrews and Roberts, 1961). The increase in value of the ratio of unsaturated to saturated fatty acids in plants grown at low temperatures is a further example (Howell and Collins, 1957; Marr and Ingraham, 1962; Gerloff, 1966). There are many other examples (e.g., Schwemmle, 1953; Hasegawa et al., 1966). If the pool sizes of hormonal-type regulators are altered, then extensive changes in the proteins synthesized may be expected. For example, gibberellin induces the formation of amylase and other hydrolases in the endosperm of barley and wild oats (Simpson and Naylor, 1962; Briggs, 1963; Varner et al., 1965). Indoleacetic acid modifies the peroxidase isozyme pattern of pea stems (Ockerse et al., 1966).

C. PREVENTION O F METABOLIC IMBALANCE The deleterious effects of metabolic imbalance caused by differential thermally induced reduction in the rates of enzymic reactions could be overcome by substituting isozymic forms of enzymes that have lower energies of activation for (he form of the enzyme normally present at higher temperatures. This substitution appears to occur for invertase (Blagoveshchenskii and Gavrilova, 1954; Roberts, 1967b) and possibly catalase during the cold hardening of cold-hardy varieties of wheat. However, it has not been proved that the lower energy of activation for the enzymically catalyzed hydrolysis of sucrose in cold-hardened Kharkov wheat is really caused by isozymic substitution. There are other possible explanations (Roberts, 1967b). Several cases are known in which psychrophilic bacteria appear to contain forms of enzymes with lower energies of activation than those present in related forms of mesophilic and thermophilic bacteria (Brown, 1957; Sultzer, 1961; Langridge, 1963). These cases need thorough investigation to show whether this phenomenon is caused by differences between the enzymes in primary structure or conformation. Such conformational changes have been suggested as the basis for changes in wing venation with changes in temperature in Dro.rophila (Milkman, 1963; Milkman and Hille, 1966). Conformational changes and a monomer-dimer conversion appear to be involved in the loss of heat stability of glucose dehydrogenase in spores of Bacillus cevei4.r when they germinate (Sadoff el al., 1965). Conformational changes may explain changes in thermostability of aldolase from Bucjllus .rtearothermophilzis since treatment of this enzyme with sulfhydryl compounds causes loss of heat stability (Thompson et a/., 1958). A slow conformational change at low temperatures might explain the development of resistance to freezing and thawing reported for purified catalase (Shikama and Yamazaki, 1961 ) . An investigation of these

ISOZYMES A N D COLD HARDENING

317

two possibilities should be possible with suitable material. Both may actually operate in ,iv+o. A temperature-sensitive mutant of bacteriophage T4 possesses a deoxycytidylate hydroxymethylase which has a lower temperature coefficient than the wild type. This mutation is mapped in the region that is believed to control the structure of this enzyme (Wiberg and Buchanan, 1964). This observation supports the concept that enzymes with changed temperature coefficients differ from their normal counterparts because of small changes in their amino acid sequences (i.e., they are isozymes) . Examples of what also appears to be the substitution of one form of a protein for another during cold acclimation are known in cold-blooded animals (Prosser, 1963; Smith, 1966). The electrophoretic patterns produced by lactic dehydrogenases of goldfish liver indicate that the pattern changes with acclimation. The change suggests ;1 relative increase in production of one type of subunit at the lower temperatures (Hochachka, 1965). This type of change in a protein consisting of several subunits might allow an almost continuous adaptation to changes in temperature. Changes occur in the amino acid composition of protein synthesized by goldfish intestinal mucosa with changes in temperature (Morris and Smith, 1967). In the acclimation o f the goIdfish, the inhibitory effect of actinomycin D on production of a part of the new protein produced in response t o a rise in temperature suggests that it is produced after new mRNA has been synthesized (Smith and Morris, 1966). This observation is consistent with the hypothesis of the substitution of different isozymic forms of functionally similar proteins for each other when the environmental temperature changes. At the upper end of the temperature range there is evidence for similar isozymic substitutions. Both Bciri1lii.r coagr/lan.s and B. stearotherrnophilz/.s produce amylases with greater thermostability when grown at 55OC. than when grown at 35OC. (Campbell, 19541). These enzymes have been crystallized (Campbell, 1954b). Unfortunately, it is impossible to be sure that they differ in primary structure because the amino acid composition of only one of the amylases involved has been determined (Campbell and Manning, 1961 ) . Changes in the heat stability of pyrophosphatase with changes in growth temperature have been observed with 6. stearothermophilz~s (Brown et al., 1957). In Drosophilu larvae, two additional puffs on the salivary chromosomes (Clever, 1964) may be induced by exposing the larvae at 37'C. (Ritossa, 1963). This suggests that the production of at least two new proteins is induced by exposure of Dro.rophilu larvae to high temperatures. Some thermophilic bacteria grown at higher temperatures are known to produce more heat-stable enzymes and proteins than related mesophilic forms (Koffler et d.,1957; Purohit and Stokes, 1967). In one case (Loginova et ul., 1967), such differences in heat stability

318

D. W . A . ROBERTS

appear to be accompanied by changes in amino acid composition. None of these has been thoroughly investigated. Among the higher plants, heat-hardened cucumber leaves appear to contain a more thermostable urease than do nonhardened leaves (Fel'dman, 1966). The leaf protein fraction I of heat-hardened beans appears to be more heat-stable than that from unhardened beans (Sullivan and Kinbacher, 1967).

V. T h e Hypothesis of Isozymic Substitution A. NATURE OF

THE

HYPOTHESIS

Some of the deleterious effects of low temperatures on plants have been considered. It is proposed that these deleterious effects might be partially offset either by substituting one isozymic form of a protein for a different form or by changing the relative proportions o f the isozymes present. As the temperature falls, such substitutions or changes might result from changes in the composition of the plant which were caused by a drop in temperature. The present hypothesis proposes th'tt such substitutions or changes are an important part of the process ot cold hardening in plants. To substantiate such a hypothesis it must be shown not only that isozymic substitutions and changes occur but also that those which do occur have adaptive advantages for cold resistance. Experiments will be needed to distinguish between substitutions of proteins with different amino acid sequences and proteins with the same amino acid sequence but different conformations. While there is evidence that both types of substitution occur in nature, this review deals chiefly with the substitution of proteins with different amino acid sequences. A possible metabolic mechanism for producing isozymic substitutions and changes has already been considered together with a few relevant examples such as invertase in wheat and thermostable amylases in bacteria. Differences in the isozymic forms of lactic dehydrogenase in different tissues of higher animals (Markert, 1963) is another sort of example of isozymic change. This enzyme consists of four subunits which may be of either of two types. Five isozymic forms exist. Animal cells vary the relative proportions of these isozymic forms continuously by altering the proportions of the two types of subunits they produce. Differences between the temperature coefficients and optimum substrate concentrations for liver and heart lactic dehydrogenases probably arise from differences in isozynie composition of these two types of lactic dehydrogenase (Krieg el d.,1967). For enzymes consisting of more than one subunit a mechanism of this type would be very valuable to an organism in making gradual adjustments to changes in temperature. To an investigator this would appear as an enzyme with properties that varied continuously with changes in the growing conditions of the organism.

ISOZYMES A N D COLD H A R D E N I N G

319

Data on the duplicate alcohol dehydrogenase genes in corn (Schwartz, 1966) suggest that the postulate of isozymic substitution is reasonable for higher plants. One of these two genes specifies an enzyme with much reduced activity. This gene is normally repressed but it can be derepressed in the scutellum. In this tissue anaerobic conditions can alter the relative proportions of the two isozymes. It appears, therefore, that such substitutions and changes are possible when cells are subjected to changed environments. There are indications that such substitutions also occur in geographic races. They probably confer adaptive advantages on the organisms. Among ecological races of Typha latifolicl an example of an isozymic substitution is known. The heat stability of malic dehydrogenase from different races differed (McNaughton, 1966) but these races did not differ in the heat stabilities of their aldolases or glutamic-oxalacetic acid transaminases. Isozymic substitutions or comparable substitutions among the structural proteins of the photosynthetic machinery of plants could account for changes in the effect of temperature on the rates of photosynthesis that occur as a result of acclimation (Semikhatova, 1960; Mooney and West, 1964). Data on the rate of oxidation of mitochondria from various races of Sitanioiz hy.Ihix (Klikoff, 1966) suggest a similar possibility for respiration. Isozymic substitutions could result in a wide range of minor changes in the properties of those enzymes present in a given tissue. Isozymes may differ in response to inhibitors (Wieland et d.,1959), temperature, differing substrates, and differing concentrations of the same substrate (Plagemann et al., 1960; Plummer and Wilkinson, 1963) as well as in those properties mentioned below in connection with cold hardiness. Thus, isozymic substitutions could be used to make many kinds of delicate metabolic adjustments. The present hypothesis will require proof of this.

B. ISOZYMICSUBSTITUTIONS

AND COLD HARDENrNC

Isozymic substitions during cold hardening could help the plant to withstand low temperature by (1) increasing the tolerance of its proteins to high concentrations of salts and other metabolites, including other proteins; ( 2 ) reducing the low-temperature sensitivity of its proteins; (3) increasing the capacity of its proteins for binding water; ( 4 ) decreasing the chances of intraprotoplasmic ice nucleation ; ( 5 ) changing the metabolic balance. Changes in metabolic balance could help the plant withstand low temperatures by (1 ) increasing the accumulation of protective substances ; ( 2 ) reducing the accumulation of toxic metabolic intermediates; ( 3 ) moderating the effects of the differential reductions in the rates of enzymic reactions caused by drops in temperature. All these changes might occur without serious changes in the overall metabolic plan of a tissue. Comparable small modifications in structural proteins could also be involved in

320

D. W. A . ROBERTS

cold hardening. Such changes might modify the permeability and stability of cellular membranes or change the properties of ribosomes to increase cold resistance. Appreciable changes in the proteins of plants may be associated with winter hardiness (Terumoto, 1957; Heber, 1959; Hodges, 1964; Pauli and Zech, 1964; Vasil'eva et al., 1964; Meador, 1965; Simura and Sugiyama, 1965; Coleman et d., 1966) and preparation for autumn (Siminovitch, 1963). These preparations for autumn were associated with increased synthesis of RNA. Such synthesis is required by the proposed hypothesis since it postulates that some different proteins are synthesized. Work on the peroxidase isozymes indicates qualitative and quantitative changes in the isozymes with changes in temperature of growth (Gerloff, 1966; Olson et a/., 1967; Roberts, 1967a). Changes in the concentration of repressor proteins induced by changes in growth temperatures in bacteria (Marr et al., 1964) add to the complications. It has been difficult to correlate or rationalize many of the observed changes in protein content with cold hardiness. The hypothesis presented here suggests that, i f we are to understand the adaptive advantage to be gained by particular changes in proteins during cold hardening, we must study in detail the properties of the specific proteins involved and, where relevant, the metabolic pathways in which they operate in vizm. The details involved are likely to be species-specific. I do not wish to suggest that the hypothesis of isozymic substitution explains all facets of cold hardening but I do suggest that it is a part of the phenomenon. The hypothesis implicates nucleic acid metabolism since changes in the primary structure of the proteins produced requires changes in the base sequence of mRNA. In this field there are already indications that low temperatures cause upsets. For example, the relative rates of production of enzymes under the control of a single operon in bacteria may be altered by changing the temperature of growth (Nishi and Zabin, 1963). Such a phenomenon, if its occurrence is verified, could produce or prevent metabolic imbalance. In Drosophila it appears that exposure to low temperatures (14°C.) can induce or enhance heterochromatinizatiun of the chromosomes which induces inactivation of the affected genes (Hartmann-Goldstein, 1967). I n vifro tests suggest the possibility of errors occurring in the translation of the genetic code at low temperatures (Szer and Ochoa, 1964). It is not known if this detrimental possibility exists Z E z h o . In cold shock some authors (Byfield and Scherbaum, 1967) have observed that although the RNA is stable it loses its ability to synthesize protein, while other authors (Strange and Postgate, 1964) observe that chilling results in subsequent degradation of the RNA. There are indications that at low temperatures the operator may cease to function properly (Marr et al., 1964). At the high-temperature end of the temperature range for growth there is evidence that the

ISOZYMES A N D COLD HARDENING

321

ribosomes stabilize the RNA. This could be the result of differences in packing of the protein and RNA or of differences in primary structure of the ribosomal proteins (Saunders and Campbell, 1966). Such changes might be the result of the type of protein substitutions propounded in this article. C. PROBLEMS ASSOCIATED WITH TESTING THE HYPOTHESIS O F ISOZYMIC SUBSTITUTION

If the hypothesis suggested correctly describes an important facet of cold hardening, there is need to compare the properties of specific proteins from coldhardened plants with those of comparable proteins from similar tissues from cold-sensitive plants of the same species o r variety. Attention must be paid to the possibility that protein behavior in z i ~ r omay differ from that in oizw because other proteins and metabolites are present in the living organism but absent in uitro. Gel electrophoresis may fail to separate the isozymic forms involved if the molecules of the two forms do not differ sufficiently in shape, size, or electrical charge. Enzymes will be the easiest proteins to work with because they have catalytic properties that serve to identify them. Little work has been done on comparing in detail the properties of proteins from cold-hardy and cold-sensitive plants of the same species. Work also is required on the relative rates of turnover of metabolites in metabolic pathways of cold-hardy and cold-sensitive plants of the same species at different temperatures. Such experiments might help to pinpoint some of the proteins requiring detailed study.

D. IMPLICATIONS O F

THE

HYPOTHESIS O F ISOZYMIC SUBSTITUTION

Several implications arise from this hypothesis. First, the phenotypic expression of the genome should be dependent on the temperature at which the plant is grown. Consequently, plants must be in the cold-hardened condition before testing their cold resistance. This principle is now widely recognized. The morphogenic effects of low temperature (Resende, 195 1; Levitt, 1956; Roberts, 1967b) and the temperature sensitivity of some of the genes for rust resistance in cereals (Waterhouse, 1929; Gordon, 1933; Martens et ul., 1967) are additional temperature-sensitive expressions of the genome. Second, the genetics of cold resistance should be very complex since minor changes in many proteins will likely be involved. There is already evidence for such complexity. Transgressive segregation for cold resistance occurs in oats (Finknec, 1966). The F, generation from a cross of wheat plants differing in cold hardiness tends to be intermediate between parents in cold resistance (Martin, 1927). Third, there is unlikely to be an exact correspondence between the accumulation of specific metabolites and cold resistance except in protein protectants and metabolic regulators that trigger the production of those proteins involved in cold hard-

322

D. W. A . ROBERTS

ening. Even in these cases, exceptions to the general pattern are to be expected, especially if comparisons are made between species that are taxonomically widely separated. If it is true that a lowered temperature of growth may result in isozymic substitutions, a corresponding phenomenon is to be expected for plants grown at higher than “normal” temperatures. Other environmental influences should produce other isozymic substitutions provided they change the pool size of compounds with regulatory effects. The production of nitrate reductase (Afridi and Hewitt, 1964) by plants supplied nitrogen in the form of nitrate is a simple example of such a phenomenon. Genetic studies of maturity in Sorghum vulgare, floral initiation in Lolium rigidam, and date of flowering in races of Potentilla glandidosa (Clausen, 1959) suggest that other environmental factors produce isozymic substitutions and the unmasking of normally latent portions of the genome. There is no reason why poikilotherms should not behave similarly. Isozymic substitutions may explain a part of the phenomenon of acclimation in animals.

VI. Concluding Remarks A speculative hypothesis regarding one facet of cold hardening in plants has been described. The basis of this hypothesis is the substitution at hardening temperatures of a modified form of a protein for the form of the functionally identical protein present at higher temperatures. It is suggested that such substitutions are triggered by either increases or decreases in the pool sizes of metabolites with regulatory functions. These increases or decreases in pool sizes are supposedly induced by the environmental factors that cause cold hardening. Such a hypothesis suggests two lines of research which have been little investigated until very recently. One line is the search for one or more cold-hardening hormones. In such a search the possibility that hardening is partly triggered by a drop rather than an increase in the concentration of a compound with regulatory activity should not be excluded. The other line of investigation is a detailed comparison of the properties of specific proteins from cold-hardened plants with the properties of their counterparts from unhardened plants of the same species or variety. In those cases that suggest that isozymic substitution is operative, protein purification followed by detailed structural studies will be required to determine whether the substituted protein differs from normal protein in amino acid sequence, level of polymerization, conformation, or some combination of these types of difference. With the techniques now available it appears that some of the processes involved in cold hardening and acclimation to other environmental influences may be amenable to experimental attack.

ISOZYMES AND COLD HARDENING

323

REFERENCES Afridi, M. M. R. K., and Hewitt, E. J. (1964). J . Exptl. Botany 15, 2 5 1 . Altland, P. D., Highman, B., Nelson, B. D., and Garbus, J. (1966). Life Sri. 5, 375. Anderson, J. D., and Moore, T. C. (1967). Plant Pby.r.riol. 42, 1527. I J J503. Andrews, J. E., and Roberts, D. W. A. (1961 ) . Can. J . B O ~ U T39, Areshidze, I. V., and Podrazhanskaya, 2. L. (1956). Pratsi Odrs’k. Unir. 146, Zbirnik Siud. Robit No. -I, 91 IChem. Absir. 52, 6508h (1958).1 Asaliina, E. (1956). Conti-ib. I w t . Laic Temp. Sc-i. 10, 83. Asahina, I!. (1966). I n “Cryobiology” ( H . T. Meryman, ed.), p. ,451. Academic Press, New York. Audus, L. J., (1935). N e w Phytologbt 34, 386. Audus, L. J. (1939). Nrul Phytologist 38, 284. Axelrod, B. (1955). Methods Ertzytnul. 1, 19. Bagi, G.,and Farkas. G . L. ( 1967). Phytochemistry 6, 161. ~ 317. Bailin, G.,and Lukton, A. (1966). Bior-him. Biophys. A L .128, Baker, P. F. (1966). Endraiwur 25, 166. Beilinsson, A. (1929). Bioi-hem. Z.213, 399. Berendsen, H . J. C., and Migchelsen, C. (1965). Ann. N.Y. Acad. S i i . 125,Art. 2, 365. Bernal, J. D. (1965). Syrnp. Sor. Exptl. Biol. 19, 17. Blagoveshchenskii, A. V., and Gavrilova, L. P. (1954). Dokl. Akad. Nauk S S S R 95, 607. Bonner, J., and Huang, R. C. (1962). Can. J . Botany 40, 1487. Bonner, J., Huang, R. C., and Gilden, R. V. (1963). Proi.. Natl. A[-ad. Sri. U S . 50, 893. Bornstein, J. (1953). J . B i d . Chem, 205, 513. Bouroncle, B. A. (1965). Pror. Sol-. Exptl. Biol. Med. 119,958. Briggs, D . E. (1963). J . I w t . Brewing 69, 13. Brown, A. D. (1957). J . Gen. Miirobiol. 17, 6.40. Brown, D.K., Militzer, W., and Georgi, C. E. (1957). Arch. Bioi.hem. Biopbys. 70, 2-18. Byfield, J. E., and Scherbaum, 0. H . (1967). Pro[.. Natl. A c d . St-i. U.S. 57, 602. Campbell, L. L. Jr. (1954a). Arch. Bioc-hem. Biopbys. 54, 154. Campbell, L. L. Jr. (1954b). J . Am. Chem. Sor. 76, 5256. Campbell, L. L., Jr. and hlanning, G. B. (1961). J . B i d . Chem. 236, 2962. Cerbon, J. (1964). Biorhrm. Biophys. Arta 88, 444. Chambers, R.,and Hale, H. P. (1932). Pror. Roy. Sot.. (London) B110, 336. Chanutin, A,, and Curnish, R. R. (1966). Arch. Biorhem. Biophyr. 113, 114. Chilson, 0.P., Costello, L. A,, and Kaplan, N. 0. (1965). Federation Proc. 24, Suppl. IS, S-55.

Clausen, J. (1959). Proc. loth Intern. Congr. Genet. Montreal 1958 1, 4 3 f . Clever, U. (1964). I n “The Nuclrohistones” (J. Bonner and P. Ts’o, rds.), p. 317. Holden-Day, San Francisco, California. Coleman, E. A,, Bula, R. J., and Davis, R. L. (1966). Plant Physiol. 41, 1681. Das, T.M., Hildebrandt, A. C., and Riker, A. J. (1966). Am. J . Botany 53, 253. Dennis, D . T., and Coultate. T. P. ( 1966). Biorhem. Biophy-r. Res. Commun. 25, 187. Derjaguin, B. V. (1966). Discussions Farday Soc. 42, 109. Dodds, J. J. A,, and Ellis, R. J. (1966). Biorbem. J . 101, 31p. Dunican, L. K., and Seeley, H. W., Jr. (1963). 1. Bar-teriol. 86, 1079. Evans, L. F. (1966a). Nature 211, 281. Evans, L. F. (1966b). Nature 213, 384.

324

D. W. A. ROBERTS

Falk, M., and Ford, T. A. (1966). Carl. J . Chem. 44, 1699. Fel’dman, N. L. (1966). Dok1.-Botan. Sri. Sect. (English Transl.) 166-168, 45. Fincham, J. R. S. (1957). Biochem. J . 65, 721. Fincham, J. R. S., and Paternan, J. A. (1957). J . Genet. 55, 456. Finkner, V. C. (1966). Crop. Sri. 6, 297. Fowler, A. V., and Zabin, I. (1966). J. Burteriol. 92, 353. Fox, C. F., and Kennedy, I:. P. (1965). Pror. Nutl. Acud. Sci. U.S. 54, 891. Frank, H . S . (1965). Federution Pror. 24, Suppl. 15, S-I. Frank. H. S., and Wen, W . Y. (1957). D~XL~U.I.I~U~/.I F~ ~ n d Sol-. q 24, 133. Frankr, W . (1957). Plnntu 49, 345. Fredrick, J. F. ( 1953). Phy.siol. Plantnruin 6, 96. Gallant, J. A. (1962). Biorhem. Biophys. Res. Commun. 8, 506. Gartner, T. K., and Riley, M. (1965). J . Bacteriol. 89, 313. Gaskins, M. H . (1959). Proc. Florida Stale Hort. Sor. 72, 339. Gerloff, E. D . (1966). Proteins and lipids in alfalfa roots as related to cold hardiness. Ph. D. Thesis, LJniv. of Wisconsin (Agr.), Madison, Wisconsin. University Microfilms, Ann Arbor, Michigan. Glasziou, K. T., Waldron, J. C., and Hull, T . A. (1966). Phtt Phy.siol. 41, 282. Gordon, W . L. (1933). Sci. Agr. 14, 184. Gorham, P. R. (1955). Methods E m y m o l . 1, 22. Grant, N . H., Clark, D . I:... and Alburn, H. E. (1966). J . Am. Chenz. S o r . 88. .<071. Graves, D. J., Sealock, R. W., and Wang, J. H. (1965). Biochemistry 4, 290. Gruener, N., and Avi-Dor, Y . (1966). Biochenz. J. 100, 762. Halpern, Y . S. (1961). Biorhem. Biophys. Res. Commun. 6, 33. Harrington, J. F., and Kihara, G . M. (1960). Pror. Am. Sor. Hort. S1.i. 75, .185. Hartmann-Goldstein, I. J. (1967). Genet. Res. 10, 143. Hasegawa, S., Johnson, R. M., and Gould, W . A. (1966). /. Agr. Food Chem. 14, 165. Heber, U. (1959). Pldnta 54, 34. Hechter, 0. (196Sa). Ann. N . Y . Arud. Sri. 125, Art. 2, 625. Hechter, 0. (1965b). Federation Prac. 24, Suppl. 15, S-91. Hellman, A., Farrelly, J. G., and Martin, D . G. (1967). Nature 213, 9x2. Hendershott, C. H . (1962). Pror. Am. Sor. Hort. Sci. 80, 241. Hochachka, P. W. ( 1965). Arch. Biorhem. Biophys. 111, 96. Hodges, H. F. (1964). Characterization of proteins and carbohydrates associated with cold-hardening in winter wheat. Ph. D. Thesis, Purdue Irniv. ( A g r . ) , Lafayette, Indiana. Univ. Microfilms, Ann Arbor, Michigan. Hogeboom, G . H. ( 1955). Methods Enzyrnol. 1, 16. Hollander, D. H., and Nell, E. E. (1954). Appl. Mirrobiol. 2, 16.1. Hopkins, A. L. (1960). Biochini. Biophys. Actct. 37, 1zL8. Hori, T. (1956). Teion Kaguku A15, 34 [English transl., U. S. Army, Snow, Ice, and Permafrost Establishment. No. 62. Wilmette, Illinois.] Horiuchi, T., and Novick, A. (1961). Cold Spring Harbor Syinp. Quant. B i d , 26, 2.17. Horiuchi, T., Horiuchi, S., and Novick, A. (1961). J . M u / . B i d . 3, 703. Horowitz, N . H., and Fling, M. (1953). Genetics 38, 360. Horowitz, N. H., and Shen, S. C. (1952). J . Biol. Chem. 197, 513. Howell, R. W., and Collins, F. I. (1957). Agron. J. 49, 593. Huang, R. C. C., Bonner, J., and Murray, K. (1964). J. Mol. Biol. 8, 54. Hulrne, A. C., Smith. W. H., and Wooltorton. L. S. C. (1961). J. Sri. Food Agr. 15, 303. Hultin, H. O., Sun, B., and Hulger, J. (1966). J . Food S i i . 31, 320.

ISOZYMES A N D COLD HARDENING

325

Hurry, S. W. ( 196.d). "The Microstructure of Cells." Houghton-hfifflin, Boston, Massachusetts. Irving, R. M., and Lanphear, F. 0 . ( 1 9 6 s ) . Pluiit Phytrof. 43, 9. Jacob, F., and Monod, J. (1961). J. Mol. Biul. 3, 318. Jacob, F., and Monod, J. ( 1961). Zrt "Cytodifferentiation ;ind Macromolecular Synthesis" (M. Locke, ed.), p. 30. Academic Press, New York. Jarabak, J., Adanis, J. A . , Williams-Ashman, H. G.. and Talalay, P. (1962). J . Biol. Chem. 237, 345. Jarabak, J., Seeds, A . E., Jr., and Talalay, P. (1966). Biuf-herizis/r)5, 1269. Jockusch, H . (1966). Bior-hem. Biophys. Res. Commuiz. 24. 577. Kauzmann, W. (1959). Adtwn. Protein Chem. 14. 1. Kerr, I. M., Cohen, N., and Work, T. S. (1966). Biorhr7r/. J . 98, 826. Ketellapper, H . J. (1963). Plant Physiml. 38, 175. Khan, A . A,, and Faust, M. A . (1967). 1'hy.riol. I'lantururii 20. 671. Kiermeier, F., and Koberlein, W. (1957). Binihem. Z.329, 2/17, Kiovsky, T. I:., and Pincock, R. E. (1966). J. A m . Chem. Sot-. 88. 170 I . Kislyuk, I. M. ( 1 9 6 4 a ) . Biophj.rir.r ( U S S R ) (Ej/,qfi.rhTvaml.) 9, 499. Kislyuk, N. M. (1964b). Dak/.-Botan. St-i. 5 ' e l / . (EnRlith Trut/r/.) 157-159, 1 ~ f l . Kligman, A . M. (1965). J . A m . Med. Auot.. 193, 1.10. Klikoff, L. G . (1966). Narure 212, 5 2 9 . Klotz, I . M. ( 1 9 5 8 ) . Scieii[.e 128, 815. Kofler. H., Mallett, G . I:.. and Adyr, J. (1957). PWK. Nut/. A ~ u d S1i. . U . S . 43, .I6 I. Koga. S., Echigo, A . , and Oki, T. (1966). App1. Mit.rohiol. 14, ,466. Korn, E. D. (1966). Srietzre 153, 1491. Korner, A. (1966). Bull. SOL-.Chim. B i d . 48, 103 1. Kowalsky, A,, and Cohn. M. (1964). Ann. Rez'. Biocbmi. 33, i81. Krieg, A. F., Gorton, S., and Henry, J. B. (1967). Cliil. Chin/. Al-ia. 17. i 6 3 . Langer, L. J., and Engel, L. L. (1958). J . Biol. C'hem. 233, 583. Langridge, J. (1963). Ann. Rei,. P h t Phy.rio1. 14, 4.11. Langridge, J., and Grifhng, B. (1959). AuJlra/iatz J . B / d . S1;. 12, 117. Langridge, P., and Morita, R. Y. (1966). J . Bacterial. 92. 418. Leibo, S. P., and Jones, R . F. (1964). A n h. Biot-hem. Bjuphys. 106, 78. Levitt, J. (1956). Agronomy 6, 5 . Levitt, J. ( 1 9 5 s ) . I n "Protoplasmatoloeia" (L. V. Heilhrunn and F. Weber, eds.), Vol. 8, p. 1. Springer, Vienna. Levitt, J. (1962). J. Theuret. B i d . 3, 3 5 5 . Levitt, J. (1966). IN "Cryobiology" ( H . T. Meryniati. e d , ) , p. .f95. Academic Press, New York. Lieberman, M., Craft, C. C.. Audia. W. V., and Wiltox, M. S. ( 1 9 5 8 ) . Z'faut Phyi;o[. 33, 307. Ling, G. N. (1965). A N N N . . Y . A[-&. S t i . 125. Art. 2. ,101. Loginova, L. G., Karpukhina, S. Y., and Guzhova. E . P. (1967). Z m . Akud. Nauk S S S R Sev. Biol. No. 4 , 595. Lohmann, W., Moss, A. J., Jr., and Perkins, W. H . (196.1). J . Nurl. N e d . 5, 304. Lohmann, W., Moss, A . J., Jr., and Barker, S. L. (1966). Nuturwir.rerisrhaflen 53, 702. Lovelock, J. E. (1954). Biochem. J . 56, 265. Lovelock, J. E. (1957). Pror. Roy. Sor. (London) B147. 427. Lovelock, J. E., and Bishop, M. W. H.. Jr. (1959). Nature 183, 139.t. Luyet, 8.J,, and Keane, J. F. (1952). Biodyynamira 7, 119.

326

D. W . A. ROBERTS

Lvov, S. D., and Altukhova, L. A. (1951). Dokl. Akad. Naud SSSR 80, 113. Maas, W. K., and McFall, E. (1964). Ann. Rev. Mi(.robiol. 18,95. Macleod, R. A,, Smith, L. D. H., and Gelinas, R. (1966). Can. J. Microbiol. 12, 61. McNaughton, S. J. (1966). Plant Physiol. 41, 1736. Magasanik, B. (1961 ) . Cold Spring Harbor Symp. Quant. B i d . 26, 249. Markert, C. L. (1963). In “Cytodifferentiation and Macromolecular Synthesis” ( M . Locke, ed.), p. 65. Academic Press, New York. Marr, A. G., and Ingraham, J. L. (1962). J. Bacterd. 84, 1260. Marr, A. G . , Ingraham, J. L., and Squires, C. L. (1964). J. Barteriol. 87, 3 56. Marre, E., Cornaggia, M. P., Alberghina, F., and Bianchetti, R. (1965). Biochem. J. 97, 20p. Martens, J. W., McKenzie, R. I. H., and Green, G . J. (1967). Can. J. Botany 45, 451. Marth, P. C. (1965). J. Agr. Food Chem. 13, 331. Martin, J. H . (1927). J. Agr. Res. 35, 493. Massey, V., Curti, B., and Ganther, H. (1966). J. B i d . Chem. 241, 2347. Mazur, P. (1960). Ann. N . Y . Acad. Sri. 85. Art. 2, 610. Mazur, P. (1966). In “Cryobiology” ( H . T. Meryman, ed.), p. 213. Academic Press, New York. Meador, D. B. (1965). Disserhition Ah.rtr. 26, 2962. Meryman, H . T. (1957). Phy.rio1. Rez’. 37, 233. Meryman, H . T. (1966). IN “Cryobiology” (H. T. Meryman, ed.), p. 1. Academic Press, New York. Meyerhof, O., and Ohlmeyer, P. (1952). /. B i d . Chetn. 195, 1 I . Meynell, G . G . (1958). J. Gen. Microbiol. 19, 380. Milkman, R. (1963). /. Cen. Physiol. 46, 1151. Milkman, R., and Hille, B. (1966). Biol. Bull. 131, 331. Molotkovskii, Y. G., and Zhestkova, I. M. (1964). Plant Physiol. ( U S S R ) (English Tram).) 11, 254. Monod, J,, and Jacob, F. (1961). Cold Sprirtg Harbor Symp. Quant. B i d , 26, 389. Monod, J., Changeux, J. P., and Jacob, F. (1963). J. Mol. B i d . 6, 306. Mooney, H . A,, and West, M. (1964). Am. J. Botany 51, 825. Morris, D., and Smith, M. W. (1967). Biorhern. J. 102, 648. Moss, C. W. (1966). Dissertation Abrtr. 26, 4178. Nash, T. (1966). I n “Cryobiology” (H. T. Meryman, ed.), p. 179. Academic Press, New York. Ng, H., Ingraham, J . L., and Marr, A. G . (1962). J. Bac/eriol. 84, 331. Nishi, A,, and Zabin, I. (1963). Biorhem. Biophyr. Res. Commun. 13, 320. N o d , F. F. (1936). Nafuru)is.rei?.rrhaftevt 24, 481. Numa, S., and Ringelmann, E. (1965). Bioc-hem. 2. 343. 258. Ockerse, R., Siegel, B. Z., and Galston, A . W . (1966). Science 151, 452. O’Donovan, G. A,, and Ingraham, J. L. (1965). Pror. Natl. Acud. Sri. U.S. 54, 451. Olson, A. C . , Hawker, K. M., DeJong, D. W., and Jansen, 1:. F. (1967). Federation Pror. 26, 385. Parker, J. (1963). Botan. Rev. 29, 123. Pauli, A. W., and Zech, A . C. (1964). Crop Sri. 4, 204. Penefsky, H. S., and Warner, R. C. (1965). J. B i d . Chem. 240, 4694. Pentzer, W. T., and Heinze, P. H . (1954). Ann. Rev. Plant Physiol. 5, 205. Perkins, H . J., and Andrews, J. E. (1960). Natlrrwissenschaften 47, 608. Pincock, R. E., and Kiovsky, T. E. (1966). J. Chem. Educ. 43, 358.

ISOZYMES A N D COLD HARDENlNG

327

Pitman, M. G., and Saddler, H. D. W. (1967). Pror. Nail. Arad. Sci. U.S. 57, 44. Plagemann, P. G. W., Gregory, K. F., and Wrbblewski, F. (1960). J . Biol. Chem. 235, 2288. Plunimer, D. T., and Wilkinson, J. H. (1963). Biochem. J . 87, 423. Polge, C., Smith, A. U., and Parkes, A. S. (1949). Nature 164, 666. Potter, V. R. (195s). Methods Enzymol. 1, 13. Prosser, c. L. (1963). In “The Role of Cellular Reactions in Adaptation of Multicellular Organisms to Environmental Temperature,” Intern. Symp. Cytoecol. Leningrad (G. I . Poljansky, ed.), p. 49. Academy of Sciences of the LISSR Press, Leningrad. Pullman, M. E., and Monroy, G. C. (1963). J . B i d . Chem. 238, 3762. Purohit, K., and Stokes, J. L. (1967). J . Bat-teriol, 93, 190. Rapatz, G., and Luyet, B. (1965). Biodynamira 9, 3 3 3 . Reithel, F. J. (1963). Advaa. Protein Chem. 18, 123. Resende, F. (1951). Bol. Soc. Portuguesa Cienr. Nat. 18, 256. Ring, K. (1965a). Biorhim. Biophys. Acta. 94, 598. Ring, K. (1965b). Biorhrm. Biopbys. Res. Commun. 19, 576. Ritossa, F. (1963). Drmophila Inforni. Serv. 37, 122. Roberts, D. W . A. (1951). Can. J . Botany 29, 383. Ro+erts, D. W . A. (1967a). Plant Physiol. 42, Sl6. Roberts, D. W . A . (1967b). Can. J . Botany 45, 1347. Sadoff, H. L., Bach, J. A., and Kook, J. W . (1965). Spores, Symp., 3rd, Allerton Park, Ill., 1964, p. 97. Sakai, A. (1957). Nippot/ Sawshigaku ZaJJhi 26, 13. Salt, R. W . (1958). J . Insect PhyJiol. 2, 178. Salt, R. W . (1961). Ann. Rev. Etitomol. 6 , 55. Saunders, G. F., and Campbell, L. L. (1966). J . Bacieriol. 91, 332. Scharff, O., and Vestergaard-Bogind, B. ( 1966). SI-und. J . Clin. Lab. Invest. 18, 8 7 . Schwan, H. P. (1965). Am. N . Y . Arad. Sri, 125, Art. 2, 344. Schwartz, D. (1966). Proc. Null. Ac-ad. SLY. US.56, 1431. Schwemmle, B. (1953). Plawta 43, 98. Sellschop, J. P. F., and Salmon, S. C. (1928). J . Agr. Res. 37, 315. Semikhatova, 0. A. (1960). Botan. Zh. 45, 1488. Sherman, J. K. (1964). Pror. S o r . Exptl. B i d Med. 117, 261. Shikama, K. (1963). Sri. Rept. Tohoku Univ. Fourth Ser. 27, 91. Shikama, K., and Yamazaki, I. ( 1961). Nature 190, 83. Siminovitch, D. (1963). Can. J . Botany 41, 1301. Simpson. G . M., and Naylor, J. M. (1962). Can. J . Botany 40, 1659. Simura, T., and Sugiyama. N. (1965). Japan J . Breeding 15, 230. Sizer, I. W . (1943). Adt:an. Enzymol. 3, 3 5 . Skou, 3. C . (1965). Phyriol. Rev. 45, 596. Smith, M. W . (1966). Experirntia 22, 252. Smith, M. W. (1967). Biorhem. J . 105, 67. Smith, M. W., and Morris, D. (1966). Experientia 22, 678. Stewart, I., and Leonard, C. D. (1960). Pro[-.A m . Sol.. Hurt. Sci. 75, 253. Straka, R. P., and Stokes, J. L. (1959). J . Barterrol. 78, 181. Strange, R. E. (1964). Nature 203, 1304. Strange, R. E.. and Postgate, J. R. (1964). J . Gen. Mirrobiol. 36, 393. Sullivan, C. Y . , and Kinbacher, E. J. (1967). Crop Sci. 7, 241. Sultzer, B. M. (1961). J . Bacterial. 82, 492.

328

D. W. A . ROBERTS

Sumner, J. B., and Somers, G. F. (1947). “Chemistry and Methods of Enzymes,” 2nd Ed., p. 43. Academic Press, New York. Sussman, M. V., and Chin, L. (1966). Science 151, 324. Sussman, R., and Jacob, F. (1962). Compt. Rend. 254, 1517. Szer, W., and Ochoa, S. (1964). J . Mol. Biol. 8, 8 2 3 . Taketa, K., and Pogell, B. M. (1965). 1. Biol. Chem. 240, 651. Taylor, A. C.. and Grrstner, R. (1955). J . Cellular Comp. Phy.tiol. 46, 477. Terumoto, I. (1957). Teiorz Kagaku B15, 39. Terumoto, 1. (1965). Teion Kagaku B23, 11 Thompson, T. L., Militzer, W . E., and Georgi. C. E. (1958). J. Bacterial. 76, 337. Trunova, T. 1. (1964). Plan! Physiol. ( U S S R ) ( E n g l i s h T r a m l . ) 10, 495. Tumanov, I. I . , and Krasavtsev, 0. A. ( 1 9 5 9 ) . Plant Phpiol. ( U S S R ) (English Tranrl.) 6, 663. Tumanov, 1. I., and Trunova, T. I . (1957). Plant Physiol. ( U S S R ) ( E n g h h . Trans/.) 4, 379. LJdaka, S., and Horiuchi, T . (1965). Bior.hetn. BiophyJ.. Res. Comniun. 19. 156. IIllrich, H., and Heber, U. ( 1 9 5 8 ) . Planla 51, 399. IJllrich, H., and Heber, IJ. (1961). Planta 57, 370. Ilmbarger, H . E. (1961). Cold Spring Harho,. Symp. Quaat. Biol. 26, 301. “rnbarger, H. E. (1963). A N NRev. . Plant Physiol. 14, 19. Varner, J. E., Chandra. G. R., and Chrispeels, M. J. (1965). 1. Crllulav C6mp. P b p i o l . 66, Suppl. 1, 55. Vasil’eva, 1. M., Lebedeva, L. A,, and Rafikova, F. M. (1964). Plant Physiol. ( U S S R ) (Engli.ih Trans/.) 11, 761. Verzhbinskaya, N. A,, and Sidorova, A . I . (1964). Biuphy.ci~s ( U S S R ) (English T r a w l . ) 9. 375. Vos. 0. (1965). Intern. J. Radiation Biol. 9. 179. Wallach, D. F. H., and Zahler, P. H. (1966). Pror. Nail. Al.ad. Sci. U.S. 56. 1552. Wang, G. T., and Marquardt, W . C . (1966). J. Pro!ozool. 13, 1 2 3 . Wang, S. Y. (1961). Nature 190, 690. Warren, J. C., and Peterson, D. M. (1966). Srience 152, 1245. Waterhouse. W. L. (1929). Pror. Limean Sot-. N. S. Wales 54, 615. Wiberg, J. S., and Buchanan, J. M. (1964). Proc.. Natl. At-ad. Sci. U S . 51, 421. Wicke, E. (1966). Angew. Chem. lnteln. Ed. English 5, 106. Wieland, T., Pfleiderer, G., and Ortanderl, F. (1959). Biorhem. Z . 331. 103. Williams, K. I. H.. Burstein, S. H., and Layne, D. S. (1966). Ari-h. Biochem. Biophys. 117, 84. Wood, T. H., and Rosenberg, A . M . (1957). Biorhim. Biophys. Arta 25, 7 8 . Wiinsche, LJ. ( 1966). Natur!~iis.~eri.~cbaf~e?i 53, 386. Yasumatsu, K., Ohno, M., Matsumura, C.. and Shimazono, H. (1965). Agr. B i d . Chem ( T o k y o ) 29, 665. Youngner, V. €3. (1961 ) . Sc-ience 133, 577. Zondag, H. A. (1963). Sciewr-e 142, 965.

Author Index Numbers in italics refer to the pages on which the complete references are listed.

A Aaji, C., 179, 184. 189 Acs, G., 194, 225 Ada, G. L.. 81, 99 Adams, J. A.. 31 1 , ,325 Adamson, A. W.. 34. 60 Adiga, P. R., 196, 228 Adler, H. I., 193, 225 Adye, J., 317, 325 Afridi, M. M. R. K.. 322. 323 Afzelius, B. A., 109, 143, 145, I85 Aherne, W., 252, 253, 299 Ajtkhozhin, M. A,, 218, 219, 226, 2317 Akers, C. K., 70. 103 Alberghina. F., 312, .?26 Albritton, W . L., 205. 2-79 Alburn, H. I!., 305, 324 Alexander, M., 193, 195, 228 Allen, D . W., 224. 225 Allmann, D. W., 75, 101. 177, 181 Alpers, D. H., 210, 211, 225 Altamirano. M., 35, 5 9 Altland, P. D., 312, 323 Altukhova, L. A,, 316, 326 Ambrose, E. J., 79. 91, 97. 98, 99, 103 Ames, A., 96, 100 Amos, H., 194, 231 Anderson, J. D., 112, 323 Anderson, T. F., 95, 100 Andoh. T., 222, 225 Andrews, J. E., 311, 312, 316, 323, 3-76 Antonini, E., 51. 60 Areshidze, I. V..316, 323 Arlinghaus, R., 224. 230 Armentrout. S. A,, 204, 205, 126 Aronson, A. I., 212, 219, 225, 231 Artman, M., 222, 223, 225 Asahina, E., 304, 316, 32.3 Asakura, S., 56, J 8 Asconas, B. A., 197, 228 Astbury, W. T., 11, 36, 58 Astrachan, L., 197, 208, 226 232

Attardi, B., 170, 171, 173, 180, 181. 189 Attardi, G., 170, 171, 173, 180, 181, 189, 195, 208, 228, 244, 299 Audia. W. V.. 315, 325 Audus, L. J., 305. 323 Augustinson, K.. 221, 226 Avers, C. J., 140, 141. 142, 180, 181, 189 Avery, 0. T., 4, 58 Avi-Dor. Y.. 308. 32-5 Axelrod, A. E.. 194, 199, 206, 215, 232 Axelrod, B., 311, 323

B Baas, P. P.. 137. 184 Babcock. K. L.. 93, 102 Rach. G.. 257, 260. 299 Bach, J. A,. 316, 3-37 Bachmann, E.. 75, 101 Barlund, H., 2. 58 Bagi, G.. 305, 3173 Bailey, €..148, 181 Bailin, G., 313, 323 Baker. J. R., 68. 9 9 Baker, P. F., 308, 323 Bakerman. S.. 71, 74, 76, 99 Ballentine. R., 71, 103 Baltus, E., 179, 189, 214, 220. 226, 232 Bangham, A. D . , 65, 67, 85, 86, 94, 97, 99. 100, 101 Bank, A., 201, 206, 226 Banthorpe, D. V., 164, 182 BBrHny, M., 172, 182 Barany, W. C., 243, 252, 262, 263, 268, 273, 302 Bard. S. G., 218. 230 Barker, S. L., 312, 325 Barnard, E. A,, 83. 100. I01 Barnard, P. J., 88, 100 Barondes, S. H., 197, 222, 226 Bartley, W., 148, 156, 181, 186 Basford, R. E., 173, I 8 2 Bass, R., 146, 147, 149, 151, 181, 185 Baudhuin, P., 148, 183, 257, 259, 262, 293, 299

320

330

AUTHOR lNDEX

Bauer, G. E., 201, 227 Bauer, W., 120, 121, 124, 125, 126, 131, 133, 181, 186 Baum, H., 75, 101 Baumgarten, D., 121, 130, 188 Bear, R. S., 69, 104 Beard, N. S., Jr.. 204, 205, 226 Beattie, D. S., 173, 182 Becker, A,, 154, 182 Becker, F. F., 95, 102 Becker, Y.,198, 199, 201, 213, 230 Beebe, S . P., 96, 100 Beguin, S., 210, 211, 224, 229 Beilinsson, A,, 311, 323 Belitsina, N. V., 218, 219, 226, 232 Bell, E., 217, 219, 228, 231 Beller, B., 200, 231 Bello, J., 86, I05 Benditt, E. P., 298, 301 Bennett, H. S., 4, 58, 94, 100 Ben-Or, S., 87, 88, 100, 101 Benson, A. A,, 73, 74, 100 Berendsen, H. J. C . , 307, 323 Berger, S., 220, 231 Berlin, C. M.. 207, 231 Berman, G. R., 216, 226 Bernal, J. D., 307, 323 Bernardi, G., 180, 189 Bernstein, J., 28, 33, 58 Berthet, J., 257, 259, 262, 293, 299 Berwick, K., 96, 100 Berwick, L., 86, 96, 100, 101, 103 Betzinger, R. J., 56, 58 Beutner, R., 35, 58 Bianchetti, R., 312, 326 Biggs, D. R., 177, 184 Birnstiel, M. L., 114, 188 Bishop, J., 199, 226 Bishop, M. W. H., Jr., 311, 325 Bitensky, L., 91, 92, 100 Bladen, H. A,, 211, 226 Blagoveshchenskii, A. V., 316, 323 Blair, J. E., 158, 188 Blair, P. V., 75, 101 Blichfeldt, H. F., 244, 299 Blobel, G . , 199, 206, 214, 226, 232 Bloom, S., 216, 226 Bock, R. M., 74, 75, 100, 101, 177, 178, 182, 187

Bockstiegel, G., 257, 260, 299 Bode, V. C . , 130, 136, 182 Boedtker, H., 198, 226 Bogorad, L., 114, 184 Bolton, E. T., 198, 226 Bond, H. E., 165, 166, 183 Bond, S. B., 165, 166, 183 Bonner, J., 3 15, 323, 324 Bonner, W. D., Jr., 110, 114, 143, 144, 187 Booth, F., 79, 100 Booyse, F. M., 217, 226 Bornstein, J., 309, 323 Borsook, H., 216, 226, 229 Borst, P., 109, 110, 112, 115, 116, 118, 120, 122, 124, 125, 126, 127, 128, 129, 130, 131, 136, 138, 139, 140, 141, 143, 144, 145. 149, 151, 152, 153, 154, 158, 165, 166, 168, 173, 177, 179, 182, 1x4, 186, 188, 189 Bouroncle, B. A., 311, 323 Bowers, M. B., 91, 100 Boyle, P. J., 3, 4, J8 Brachet, J., 191, 218, 220, 226, 227, 230 Bradley, S., 20, 58 Bragg, J. K., 42, 61 Bragg, W. L., 42, 58 Branton, D., 96, 100 Bratton, C . B., 21, 58 Brawerman, G., 197, 226 Brdiczka, D., 172, 186 Bregman, J. I., 14, 58 Breidenbach, R. W., 114, 182 Bremer, H., 211, 226 Bremer, H. J., 220, 226 Brenner, S., 196, 230 Brewer, E. N. 152, 153, 182 Briggs, D. E.. 316, 323 Briggs, D. R., 53, 60 Brinton, C . C . , 79, 100, 198, 201, 232 Britten, R. J., 165, 167, 182 Bronsert, U., 172, 182 Brooks, D. E., 98, 104 Brown, A. D., 316, 323 Brown, D. K., 317, 323 Brown, D. M., 83, 100 Brunschwig, A., 96, 100 Buchanan, J. M., 317, 3-78

AUTHOR INDEX

Buck, C. A., 197, 228 Biicher, T., 172, 186 Biitschli, 0.. 11, 58 Buffon, G., 246, 299 Bula, R. J., 320, 323 Bulger, J., 310, 324 Bull, H., 20, 21, 58 Bull, T. A., 312, 324 Bullivant, S., 96, 100, 104 Bungenbeg de Jong, H. G., 64, 105 Burdon, M. G., 138, 183 Burge, R. E., 11, 59, 69, 101 Burgoyne, L. A,, 115, 182 Burka, E. R., 85, 100, 201, 206, 216, 224, 226, 230 Burn, G. P., 5, 59 Burnet, F. M., 81, 100 Burny, A,, 196, 199, 201, 226, 229, 230 Burr, H. E., 165, 166, 183 Burri, P., 268, 269, 278, 292, 299 Burstein, S. H., 312, 328 Byheld, J. E., 320, 323 Byrne, R., 2 11, 226

C Cahn, J. W., 244, 245, 275, 279, 284, 301 Cairns, J., 142, 162, I82 Caldwell, P. R., 262, 294, 301 Campbell, L. L., Jr., 317, 321, 323, 327 Campbell, P. N., 173, I83 Campbell, w., 173, I87 Caputo, A., 51, 60 Carnevali, F., 112, 158, 182 Carpenter, A . M., 243, 284, 298. 299, 301 Carruthers, C., 96, 100 Carstensen, E. L., 80, 100 Casby, J. U., 14, s y Casley-Smith, J. R., 68, 100 Castelfranco, P., 114, 182 Catalano, P., 96, 100 Cecere, M. A , , 199, 232 Cerbon, J., 307, 323 Cereijido, M., 26, 60 Chalkley, H. W., 244, 248, 249, 279, 299,

3 00 Chambers, R., 7, 21, 31, 58, 91, 100, 306, 323 Chandra, G. R., 316, 328

331

Chang, L. O., 147, 182 Changeux, 1. P., 49, 51, 52, 18, 60, 313, 326 Chantrenne, H., 196, 208, 216, 220, 226, 230 Chantrenne-Van Halteren, M. B., 220, 226 Chanutin, A , 309, 311, 312, 323 Chapman, D., 67, 100 Chapman, G., 7, 21, 23, 58 Chargaff, E., 136, 188, 195, 228 Chaudhuri, S., 82, 98, 100 Chayes, F., 270, 299 Chen, P. Y.,74, 101 Chilson, 0. P., 312, 323 Chin, L., 307, 328 Choules, E. A., 220, 221, 229 Chrispecls, M . J., 316, 328 Chun, E. H. L., 114, 116, 182 Church, R., 169, 182 Clark, B. F. C., 205, 226 Clark, D. E., 305, 324 Clark, G. L., 69, 104 Clark-Walker, G. D., 110, 174, 175, 177, 182, 184 Clausen, J., 322, 3-33 Clauss, H., 191, 193, 214, 220, 221, 226. 228, 229 Clawson, C., 243, 299 Clayton, D. A , , 115, 120, 122, 126, 128. 133, 134, 143, 180, 182, 189 Clever, U., 317, 323 Clifford, J.. 67, 100 Cline, M. J., 193, 197, 226 Clowes, G. H. A . , 96, 100 Cobble, J. W., 34, 60 Cochran, W. G., 269. 300 Coconi, F. M., 201, 206, 226, 230 Cohen, A,, 193, 221 Cohen, J. A., 136, 137, 186 Cohen, N., 204, 229, 3 15, 325 Cohen, S. S., 222, 223, 231 Cohn, M., 307, 325 Cohn, W. E., 83, 104 Cole, K. S., 3, 35, 58 Coleman, E. A,. 320. 323 Collander, R., 2, 3, 58 Collins, A,, 148, 185 Collins, F. I., 306, 316, 324 Coman, D. R., 95, 97, 100

332

AUTHOR INDEX

Conover, T. E., 172, I82 Conway, E. J., 3. 4, 5, 58 Cook, G. M. W., 73, 82, 86, 87, 88, 100, 104

Cook, W. H., 72, 100 Coombs, R. R. A.. 82, 92, 101, 105 Cooper, D., 164, 182 Cope, F., 26, 27, 58 Cordes, S.. 117, 185 Cornaggia, M. P., 312, 326 Corneo, G., 110, 112, 116, 150, 156, 165, 179, 182, 189 Cornfield, J., 248, 249. 279. 299, 300 Costello, L. A,, 112, 323 Cotman, C., 156, 184 Cottier, H., 257, 286, 300 Coultate, T. P., 315, 323 Counts, W. B.. 115, 148, 182 Coupland, R. E., 257, 300 Cousineau, G . H., 218, 228 Cowan, S. L., 35. 5 8 Cozzone, A,, 214, 215, 226, 230 Craft, c. C., 315, 325 Crane, F. L., 67, 100 Crawford, L. V., 1 2 5 , 126, 129, 182 Crestfield, A. M.. 83, 100 Criddle, R. S., 74, 75, 100, 101, 114, 177, 182

Crocco, R. M., 205, 227 Crocker, T. T., 221, 222, 227 Croft, J. H., 159, 161, 162, 187 Cudney, T. L., 86, 10.5 Cummins, J. E., 117, 182, 221, 326 Cundliffe, E., 178, 183 Cunningham, W. P., 67, 100 Curnish, R. R., 309, 311, 312, 323 Curti, B., 310, 326 Curtis, H. J., 35, 58 Curtis, P. J.. 138, 183 Cuzner, M. L., 148, 183

D Damadian, R., 54, 58 Danielli, J, F., 3, 58, 64, 65, 66, 93. 100 Dannenberg, M. A., 177, 183 Danon, D., 207, 216, 224, 226, 230 Darnell, J. E., 195, 198. 199, 201, 206, 213. 227, 229, 230, 231

Das, H. K., 211, 224, 226 Das, T. M., 315, 323 Dashman, T., 207, 227 Davern, C., 196, 230 Davern, C. I. ,196, 227 Davidson, N., 165, 188 Davis, R. L., 320, 323 Davison, A . N., 148, 183 Davson, H., 3, 58, 64, 100 Dawid, 1. B., 110, 118, 120, 122, 124, 127, 128, 131, 136, 137, 138, 145, 165, 166, 167, 168, 179, 183, 188, 189, 190 Dawson, D. M., 4, 58 Dawson, R. M. C., 94, 99, 101 Dean, R. B., 3, 58 DeBellis, R. H., 216, 227 de Boer, J. H., 20, 58 De Deken, R. H., 155, 183 de Duve, C., 91, 101, 148, 183 Defendi, V., 98, 101 De Gier, J., 67, 100 De Hoff, R. T., 251, 269, 300 DeJong, D. W., 320, 326 de Kloet, S . W., 82, 101 DeLander, A . M., 54, 59 Delesse, M. A . , 242. 300 Delius, H., 157, 178, 185, 188 DeMars, R. I., 195, 227 de Mello Mattos, C. M., 294, 301 Dennis, D. T., 315, 323 Denny, P. C., 218, 227 Derjaguin, B. V., 307, 323 D e Salle, L., 110, 186 Deutsch, D., 93, 101 D e Vries, A . , 152, 151, 182 D e Vries, H., 2 , 58 Diana, A . L., 65, 104 Dickson, R. C., 115, 152, 186 Dietz, G. W.. 207, 227 DiGirolamo, A,, 197, 227 DiMargio, E. A,, 42, 59 Dingle, J. T., 91, 92, 201, 10-7, 105 Dingman, C. W., 197, 226 Dintzis, H. M., 199, 227 Dodds, J. J. A,, 309, 323 Dorfler, G . , 298, 300 Doljanski, F., 82, 87, 88, 100, 101, 103

AUTHOR INDEX

Doty, P.. 41. 59, 109. 111, 158, 185, 187, 199. 218, 227, 232 Douglas, H . C.. 174, 181, Drach, J. C., 198, 227 Dreyfus, J. C., 200, 227 D u Buy. H. G., 166, 167, 168, 183 Duffin, R. J., 247. 300 Dumonde, D. C., 92. 101 Dunham, L. J., 96, 100 Dunican, L. K., 314, 323

E Earl, D. C. N.. 194, 227 Eason, R.. 193, 197, 227 Eastman, N. J., 54. 59 Easty, G. C.. 96, 101 Ihbesson, S. O., 264, 265, 300 Echigo, A,. 307, 324 Edelman, M . . 112, I83 Edwards, D. L., 74. 100 Eisenbeg, S., 8 2 , 87, 88, Jon, JOI, 103 Ilisenman, G., 14, IS, 17. 59 Eisenstadt, J.. 197, 226 Ekedahl. G., 221, 226 Elbers, P. F.. 67* 78. 101 Elias, H., 236, 240, 241, 255, 257, 259. 300, 302 Eljasson. I!. E., 201, 205, 227 Ellem. K. A. 0.. 197, 227 Ellis, R. J., 309, 323 Elson, D., 222, 223. 229 Emrich. 199, 232 Endter, F., 286, 300 Ingel, L. L., 311, 323 Ilngelbert, H., 222, 223, 22J Epstein, E., 28, 5 9 Epstein, H. T., 112, 183 Eranko, O., 244, 300 Erkan, Y . , 269, 270, 300 I%termann, E. F., 93, 103 Ivans, L. F., 306, 307, 323 Evans, T . E., 112, 117, 145, 182: 183 Eyer, J.. 170, 187 Eylar, E. H., 82, 89, 104 J.?

F Falk, G., 35, 59 Falk, M., 307, 324

333

Fan, D. P.. 224, 227, 230 Farber, I:., 201, 209, 214, 232 Farkas, G . L.. 3 0 5 , 323 Farquhar, M. G . , 91, 101 Farrelly. J. G, 312, 324 Faust, M. A,, 312. 321 Favelukes, S., 224. 230 Fawcett, D. W., 91, 101 Feigelson, P., 207, 227 Fel'dman. N. L., 318, 32.r' Fell, H. B.. 91, 92. loo, 101, 102 Fenichel, 1. R.. 7, 9. 5 9 Fernindez-Moran, H.. 75. 101 Fessenden, J. M., 177, 183 Ficq. A , , 218, 226 Fincham, J. R. S., 310, 324 Finean. J. B.. 69, 101 Finkner. V. C., 321, 324 Fischer, H., 12. 1.5, 59. 1'95. 227 Fischnieister, H. F.. 283, 300 Fisher, C., 283, 300 Fisher, E. H., 216. 226 Fisher, T. N., 208, 226 Fisher, W. D., 193, 2 2 j Flamm. W. G., 1 1 5 . 148, 165, 166, 182, 183 Fleck. A , , 201, 227 Fleischer, B., 68. 74, I 0 1 Fleischer, S., 68. 74. 101 Fletcher, M. J.. 147, 183 Fling. M., 314, 324 Folch-Pi, J.. 77, 101 Forchhammer, J., 209, 224, 227 Ford, T. A., 107. 324 Forssmann. W. G . , 262, 300 Fowler. A . V., 312, 324 Fox, C. F.. 308. 309, 324 Fraenkel-Conrat, H., 86, 101 Frank, H. S.. 107. 324 Franke. W.. 316, 324 Franklin. N. C..196 228 Franklin, R. M.. 193, 1'97, 227, 231 Fredrick. J. F., 314. 324 Freeman, J. A,, 66. 101 Freeman, K. B., 172. 173, 183 Freere, R. H.. 268, 287, 300 Freese, E.. 136, I86 Freifelder, D . , 131, 283

334

AUTHOR INDEX

French, E. L., 81, 99 Friesen, J. D., 195, 224, 227 Frisbie, W. S., 96, 100 Fritz, 0. G., 7, 21, 36, 59 Fuchs, A , , 269, 278, 292, 300 Fuhrmann, G. F.. 80, 88, 98, 100, 101, 103

Fukada, T., 197, 233 Fukuhara, H., 154, 1 5 5 , 170, 183, 187 Fuller, W., 193, 228 Furth, J. J., 193, 195, 228

G Gadaleta, M. N., 151, I87 Gallant, J. A., 314, 324 Galston, A. W., 316, 326 Gander, R. H., 268, 287, 300 Ganther, H., 310, 326 Garbus, J., 312, 323 Garofalo, M., 214, 231 Garren, L. D., 205, 227 Gartner, T. K., 314, 324 Gary-Bobo, C . M., 26 Gasic, G. J., 86, 98, I01 Gaskins, M. H., 312, 324 Gavrilova, L. P., 218, 226, 316, 323 Gebauer, H., 286, 300 Gefter, M., 154, 182 Geidushek, E. P., 165, 183 Geiling, E . M. K., 54, 59 Gelinas, R., 315, 326 Gellert, M., 154, 183, 185 Georgi, C . E., 316, 317, 323, 327 Gerard, R. W., 35, 59 Geren, B. B.. 69, 101 Gerloff, E. D., 316, 320, 324 Gerstner, R., 312, 328 Gesteland, R. F., 222, 227 Getz, G. S., 140, 144, 145, 146, 147, 148, 149, 155, 175, 176, 183, 186, 187 Giacomoni, D., 198, 227 Gibbs, J. H., 42, 59 Gibor, A,, 109, 132, 183, 220, 227 Gierer, A., 199, 227 Giger, G., 257, 286, 300 Giger, H., 238, 254, 257, 259, 260, 268, 269, 270, 292, 299, 300 Gil, J., 262, 300

Gilbert, W., 195, 199, 208, 224, 227, 228 Gilden, R. V., 315, 323 Gillespie, D., 198, 227 Gillespie, M. E., 210, 231 Gimigliano, A. F., 151, 187 Ginelli, E., 110, 116, 182 Girard, M., 195, 227 Girardier, L., 262, 300 Gittens, G. J., 80, 101 Glagoleff, A. A., 243, .?OO Glasziou, K. T., 312, 324 Glauert, A. M., 67, 102 Glick, M. C., 84, 104 Glick, P. M., 83, I 0 1 Gligin, M. V., 218, 227 Gliiin, V. R., 218, 227 Glover, J. C., 85, 99 Glowacki, E. R., 207, 227 Gluck, N., 216, 227 Gnagi, H. R., 237, 250, 262, 268, 270, 278, 281, 282, 285, 286, 287, 288, 289, 291, 292, 298, 299, 300, 301, 302 Goffeau, A,, 220, 227 Gold, L., 197, 226 Goldberg, B., 200, 216, 226, 227 Goldberg, 1. H., 193, 194, 217, 227, 231 Goldstein, A., 210, 211, 224, 226, 227 Goldstein, L., 221, 222, 227 Gomez, D. M., 252, 253, 301, 302 Gonzilez-Cadavid, N. F.. 173, 183 Goodman, H. M., 198, 227 Gordon, M. P., 114, I83 Gordon, W. L., 321, 324 Gordy, W., 38, 59 Goren, H. J., 83, 101 Gorham, P. R., 31 1, 324 Gorter, E., 64, 101 Gorton, S., 318, 325 Gottschalk, A , , 81, 101 Gould, B. S., 200, 229 Gould, W. A,, 316, 324 Graham, A. F., 196, 231 Grandchamp, S., 174, 185 Granick, S., 109. 132, I83 Grant, N. H., 305, 324 Grasso, J. A., 216, 228 Gratzer, W. B., 41, 59 Graves, D. J., 312, 324

AUTHOR INDEX

GrPce, M. A . , 1 1 5 , 172, 184 Green, B. R., 114, 183 Green, D. E., 68, 73, 74, 75, 100, 102, 103, 177, 182 Green, G. J., 321. 326 Green, H., 200, 216, 226, 227 Greenawalt, J. W., 151, 152, 183 Gregory, K. F., 319, 327 Gregson, N. A,, 148, 183 Grendel, F., 64, 101 Griffing, B., 310, 325 Groot, G . S. P., 151, 184 Gros, F., 195, 208, 211, 224, 228, 231, 233 Gross, N. J., 120, 130, 131, 141, 143, 147, 148, 149, 166, 180, 183, 187,

101,

230, 146, 189,

190

Gross, P. R., 218. 228 Grossman, L. J., 110, 112, 150, 156, 165, 182 Gruber, M., 118, 120. I88 Gruener, N., 308, 324 Grundfest, H., 35, 59 Guerineau, M., 180, 189 Guttes, E., 145, 183 Guttes, S., 145, 183 Guttman, L., 242, 246, 247, 301 Guzhova, E. P., 317, 325

H Haber, J. E., 51, 59 Hadjiolov, A . A . , 196, 228 Hammerling, J., 191, 220, 228 Hagen, C. E., 28, 59 Haldar, D., 172. 173, 183 Hale, H. P., 7, 21. 58, 306, 323 Hall, B. D., 197, 198, 228, 230 HaII, D. O., 151, 152, 183 Hall, J. B., 196, 228 Hallett, J., 21, 22, 59 Hally, A . D., 244, 282, 300 Halpern, Y.S., 314, 324 Hamburger, H . J., 2, 5 9 Hamilton, L. D., 193, 228 Hammett, L. P., 38, 59 Hanawalt, P. C., 112, 139, 186 Hanig, M., 81, 101 Hardesty, B., 201, 229

335

Hardigree, A . A , , 193, 225 Harkins, W. D., 12, 59 Harrington, J. F., 315, 324 Harrington, W. F., 41, 59 Harris, E. J., 5, 5 9 Harris, H., 213, 219, 221, 228, 232 Hartley, G. S . , 93, 101 Hartman, J. F.,243, 299 Hartmann-Goldstein, I. J., 320, 324 Hartridge, H., 93, I 0 1 Hartwell, L. H., 210, 211, 228 Harvey, E. B., 193, 228 Harvey, E. N., 64, 65, 100 Hasegawa, S., 316, 324 Haselkorn, R., 110, 116, 186, 187, 197, 229 Haslbrunner, E., 140, 144, 187 Haug, H., 248, 253, 286, 300 Haughton, G., 98, 102 Hauschka, T., 99, 102 Hawker, K. M., 320, 326 Hawthorne, D. C . , 174, I85 Hayashi, M., 196, 198, 208, 228 Hayashi, M. N., 198, 228 Haydon, D. A,, 65, 67, 79, 80, 81, 99, 102 Heard, D. H., 70, 79, 81, 82, 85, 88, 100, 102, 104 Hearst, J. E.. 114, 128, 188 Heber, LI., 309, 311, 320, 324, 327 Hechter, O., 307, 308, 324 Heckmann, K., 34, 59 Heiniger, H. J., 257, 286, 300 Heinle, E., 206. 232 Heinze, P. H., 315, 326 Heldt, H . W., 151, 152, 184, 186 Helge, H., 1 5 1 , 171, 172, I85 Helinski, D. R., 135, 186 Hellman, A , , 312, 324 Hellmann, W., 186 Hendershott. C. H., 312, 324 Hennig, A.. 244, 248, 256, 259, 265, 275, 282, 300 Henniker, J., 80, 102 Hennix, LJ., 144, I85 Henriques, V.,56, 59 Henry, J. B., 318, 32J Henshaw, E. C., 197, 227 Heppel, L. A,, 3, 59 Herbst, C., 97, 102

336

AUTHOR INDEX

Herbst, R., 144, 185 Hermann, L., 32, 59 Hess, F. A., 237, 250, 262, 278, 281, 282, 285, 286, 287, 288, 289, 291, 292. 298, 300, .301, 302 Hess, R., 237, 282, 285, 286, 293, 301 Heston, W. E., 207, 231 Hewitt, E. J., 322, 323 Hiatt, H., 208, 228 Hiatt. H . H., 195, 197, 208, 213, 214, 227, 228, 231 Hickler, S., 140, I86 Higa, A,, 197, 208, 209, 224, 227, 229 Highman, B., 312, 323 Hildebrandt, A . C . , 315, 323 Hill, A . V., 32, 50, 56, 59 Hill, H . Z . , 214, 233 Hille, B., 316, 326 Hilliard, J. E., 244, 245. 251, 253, 257, 260, 275. 277, 279, 284, 296, 297, 301 Hinke, J. A. M.. 26, 59 Hoagland, M. B., 197, 205, 206, 214, 228, 232, 233 Hochachka, P. W., 317, 324 Hochstein, P., 194, 229 Hodges, H . F., 320, 324 Hodgkin, A. L., 3, 6, 32, 33, 35, 36, 59 Hoffman, C . , 3, 60 Hoffman, J. F.. 64, 102 Hogeboom, G. H., 311, 32.1 Holland, J. J., 197. 228, 229 Hollander, D. H., 312. 324 Hollingshead, S., 85, 99 Hollmann, K. H.. 294, 301 Holmes, A . H., 256, 301 Holt, C. E., 180, I89 Holter, H., 4, 59 Holtzer, H., 4, 59 Holwill, M. E. J., 11, 59 Holzer, H., 175, I88 Honig, G. R., 194, 228 Hoover, E. F., 20, 60 Hopkins, A . L., 21, 58, 307, 324 Hopkins, J. W . , 197, 232 Hori, T., 307, 324 Horiuchi, S., 314, 324 Horiuchi, T., 314, 324, 328

Horowicz, P., 6, 35. 59 Horowitz, J., 195, 228 Horowitz, N. H., 314, 324 Horowitz, S. B., 7, 9, 59 Horrikawa, E., 247, 301 Howell, R. R., 196, 205, 227, 229 Howell, R. W., 306, 316, 324 Hradecna, Z . , 118, 184 Huang, M.. 177, 184 Huang, R. C., 315, 323, 32.1 Huberman, J. A., 142, 184 Hudson. B., 120, 126. 133, 134, 135. 180, 184, 189 Huez, G.. 196, 201. 229, 230 Hulcher, F. H.. 76, 102 Hulme, A. C . , 315, 324 Hultin, H. 0.. 74, 103, 310, 324 Hultin, T., 201, 218, 227, 2.33 Humm, D. G., 169, I84 Humm, J. H., 169, 184 Humphreys. T.. 217, 219, 228 Hurry, S. W.. 308, 325 Hunvitz. J., 154. 182, 193, 195, 228. 229 Huxley, H . E., 33, 53

I Infante. A. A., 218, 219, 228, 230 Ingraham, J. L.. 306, 313, 315. 316, 320, 326 Inouye, M.. 199, 2.?2 Irving. R . M., 312, .?25 [to. s., 94, 103 Izawa, M., 220, 227

J Jackson, R. J., 199, 200, 207, 230 Jacob, F., 210, 228, 313, 314. 315, 325, 326, 327 Jacobson, L., 4, 60 Jacques, M., 28, 59 Jahrisch, S., 93, 102 Jakob,H., 112, 145, 358, 159, 160, 161, 181, James, A. M., 79, 80, 101, 102 Janin, J,, 197, 209, 231 Jansen, E. F., 320, 326 Jansz. H . S., 136, 137, 184, 186 Jarabak, J., 310, 311, 321,

AUTHOR INDEX

Jayaraman, J., 156, 184 Jensen, W. A., 4, 60 Jensen, W. N., 206, 232 Jeon, K. W., 221, 228 Jinks, J. L., 109, 184 Jockusch, H., 311, 325 John, D. W., 215, 228 Johnson, R. M., 316, 324 Jones, A . , 54, 59 Jones, R. F., 309, 110, 325

K Kadenbach, B., 172, 179, 184, 192, 228 Kaempfer, R. 0. R.. 21 1, 229 Kagawa. Y . , 76, 102 Kahan, I;., 193, 2-79 Kahan, F. M., 193, 229 Kalant, H..97, I02 Kalf, G. F., 115, 172, 184 Kamat, V. B., 67, 82, 100, I 0 4 Kanner, L. C., 224, 226 Kanno, Y . , 96, 102 Kao, C. Y . , 33, 3 5 , 58. 59 Kapanci, Y., 294, 301 Kaplan, H. P., 294, 301 Kaplan, N. O., 312, 323 Karol, M. H., 180. 189 Karpukhina, S. Y . . 317, 325 Karreman, G., 7, 9, 54. 59. 60 Karrer, H. E., 67. 102 Kartha, G., 83, 102 Katchalsky, A., 97, 102 Katoh, T., 174, 184 Katz, J. H., 82, 103 Katz, M.. 98. 104 Katz, R. D.. 3 3 , 36, 59 Kauzrnann, W., 310, 3-35 Kavanau. J. L., 65, 66, 10-3 Kawade, Y . , 197, 233 Keane, J. F., 3 1 1 , 325 Keck, K., 191, 193. 21.1. 220, 221, 226, 228, 229 Keighley, G., 216, 226 Kemp, A., Jr., 151, 284 Kennan, A. L., 215, 216, 231 Kennedy, E. P., 308, 309, 324 Kennell, D., 198. 229 Kenney, F. T., 205. 207, 229

337

Kepes, A,, 210, 211, 223, 224, 229 Kerr, I. M., 204, 229, 315, 325 Ketellapper, H. J., 315, 325 Keynan, A., 197, 208, 209, 229 Keynes, R. D., 5, 32, 59 Khan, A. A., 312, 325 Kick, C., 85, 100 Kieras, K. J., 116, 187 Kiermeier, F., 311, 325 Kihara, G. M., 315, 324 Kimball, R. F., 222, 229 Kimberg, D. V., 291, 300. 302 Kimura, Y.,35, 59 Kinbacher, E. J., 318, 327 Kinoshita, J. A,, 217, 232 Kiovsky, T. E., 304, 305, 32>, 326 Kirk, J. T. O., 114, 184, 193, 229 Kirschbaum, J. B., 210, 227 Kirschner, R. H., 180, 189 Kislev, N., 114, 184 Kislyuk, I. M., 315, 32g Kislyuk, N. M., 315, 325 Kistler, G. S., 240, 253, 262, 266, 268, 273, 277, 279, 287, 292, 294, 301, 302 Kittel, C., 52, 58 Kitzinger, C . , 56, 60 Kivity-Vogel, T., 222, 223, 229 Kjeldgaard, N. O., 209, 224, 226 Klein, G., 96, 98, 100, 203 Klein, L. A,, 80, 100 Kleinschmidt, A. K., 131, 183. 186 Klenk, E., 81, 102 Kligman, A . M., 312, 325 Klikoff, L. G., 319, 325 Klima, J., 174, 175, I87 Klingenberg, M., 151, 152, 184, 186 Kloss, K., 144, 185 Klotz, I. M., 307, 325 Knight, B. W., 250, 252, 253, 261, 262, 296. 301, 302 Knijnmburg, C. M., 136, 137, 186 Knopf, P. M., 199, 200, 229, 232 Knox, W. E., 207, 229 Koch, J., 110, 115, 145, 284 Kdberlein, W., 311, 325 Koehn, P. V., 217, 230 Koffler, H., 317, 325 Koga, S., 307, 325

338

AUTHOR INDEX

Kohlmeier, V., 195, 228 Kohne, D. E., 165, 182 Koketsu, K., 35, 59 Kollin, V., 211, 229 Konrad, M. W., 211, 226 Kook, J. W., 316, 327 Kopaczyk, K., 75, 101 Koritz, S. B., 173, 182 Korman, E. F., 75, 101 Korn, E. D., 67, 68, 102, 307, 325 Kornberg, A,, 117, 154, 187 Korner, A,, 194, 198, 199, 200, 207, 227, 229, 230, 315, 325 Koshland, D. E., 51, 5 9 Kowalewski, V., 26, 60 Kowalsky, A., 307, 32$ Kraemer, P. M., 88, 102 Krasavtsev, 0. A,, 304, 328 Kretsinger, R. H., 200, 229 Krieg, A. F., 318, 325 Kroger, R. A., 151, 152. 186 Krogh, A., 3, 1 9 Kronau, R., 175, 188 Kroon, A. M., 109, 110, 112, 115, 120, 122, 124, 125, 127, 128, 136, 138, 139, 140, 143, 144, 149, 154, 165, 166, 168, 172, 177, 179, 182, 184, 188 Kruh, J., 216, 229 Kubinski, H., 118, 184 Kuff, E. L., 110, 144, 147, 149, 187, 229 Kurland, C. G., 208, 228 Kwan, S. W., 214, 229

L Lachmann, P. J., 82, I05 Laipis, P., 120, 127, 129, 188 Lamar, C., Jr., 215, 216, 232 Lamfrom, H., 200, 229 Lane, B. P., 95, 102 Langer, L. J., 311, 321 Langmuir, I., 19, 5 9 Langridge, J., 310, 316, 325 Langridge, P., 310, 325 Lanphear, F. O., 312, 321 Lansing, A. I., 82, 85, 102 Lark, K. G., 136, 184 Larsson, A., 151, 184

214,

118, 131, 145, 173,

200,

Laszlo, J., 194, 229 Latham, H., 195, 199, 206, 213, 227, 229, 23 1 Lauffer, M. A,, 79, 100 Lauwers, A., 177, 181 Layne, D. S., 312, 328 Lazarow, A , , 243, 284, 298, 299, 301 Leahy, J,, 199, -326 Lebedeva, L. A,, 307, 320, 328 Lebleu, B., 201, 229 Lebowitz, J., 120, 121, 125, 127, 129, 188 Lehman, I. R., 222, 229 Lehninger, A. L., 132, 172, 184, 186 Leibnitt, L., 298, 301 Leibo, S. P., 309, 310, 321 Leive, L., 209, 210, 211, 229 Lenard, J., 72, 73, 102 Lenz, F., 257, 301 Lenza, G., 177, 181 Leonard, C. D., 312, 327 Lerman, L. S., 162, 184 Lerman, M. I., 218, 232 Levi, H., 5, 19 Levin, J. G., 211, 22G Levinthal, C . , 195, 197, 208, 209, 210, 211, 224, 227, 229, 231, 233 Levitt, J., 303, 304, 305, 306, 307, 31 I. 313, 316, 321, 325 Lewis, G. N., 79, 102 Lewis, I. C.. 38, 61 Lewis, M. S., 26, 5 9 Lieberman, I., 82, 98, 100 Lieberman, M., 315, 3-35 Lin, E. C . C., 207, 229 Lin, S., 201, 229 Lindegren, C. C., 140, 186 Lindegren, G., 140, 186 Ling, G . N., 2, 3, 4, 5, 6 , 7, 8. 9, 11, 12, 14, 15, 16, 20, 21, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 35, 38, 43, 44, 46, 48, 49, 50, 52, 53, 54, 56, 57, 60, 307, 325 Lingrel, J. B., 198, 227 Linnane, A. W., 110, 155, 174, 175, 176, 177, 178, 181, 182, 184, 181, 188, 189, 190 Lipmann, F., 56, 60 Lipshitz, R., 136, 188 Lipton, S., 75, 101

AUTHOR INDEX

Little, J. W., 154, 185 Loeb, A. L., 79, 105 Loeb, J. N., 196, 229 Loewenstein, W. R., 88, 96, 102 Loginova, L. G., 317, 325 Lohmann, W., 311, 312, 325 Looney, W. B., 147, 182 Lorch, I. J., 221, 228 Loud, A. V., 237. 243, 252, 262, 263. 268, 273, 286, 292. 293. 301, 302 Lovelock, J. E., 309, 311, 325 Lowenstein, J. M., 151, 186 Lowney, L. I., 211. 226 Luck, D. J. L., 109, 112, 142, 149, 150, 185, 186 Lucy, J. A., 67, 91, 102 Lukins, H. B., 155, 175, 176, 185, lR9 Lukton, A., 313, 323 Lundsgaard, E., 56, 59, 60 Luria, S. E.. 195, 196, 227, 228 Luyet, B. J., 7, 21, 22, 60, 311, 312, 325, 327 Luzrati, V.,67, 70, 102 L’vov, S. D., 316, 326 Lyklema, J., 79, 102 Lyn, G., 154, 182

M Maas, W. K., 312, 326 McBain, J. W., 34, 60 McCarthy, B. J.. 169, 182, 197, 198, 218, 226, 228, 229, 232 McCarthy, K. S., 194, 229 McCarty, M., 4, 58 McClatchy, J. K., 212, 229 McCluskey, R. T., 91, 104 McConnell, D. G., 75, 101, 102 McCrea, J. F., 81. 100 McDonald, J. S., 35, 60 McFall, E., 312, 326 McHattie, L. A., 130, 182 McKenzie, R. I. H., 321, 326 Mackler, B., 112, 141, 144, 156, 168, 174, 185, I88 McLaren, A. D., 4, 60, 93, 102, 103 McLauchlan, K. A., 7, 21, 23, 58 MacLennan, D . H., 75, 101, 102 MacLeod, C. M., 4, 18 Macleod, R. A,, 315, 326

339

McNaughton, S. J., 319, 326 McQuillen, K., 178, 183, 210, 211, 231 Maddy, A. H., 76, 103 Magasanik, B., 195, 210, 211, 228, 229, 230, 312, 326 Maggio, R., 218, 229, 230 Mahler, H. R., 112, 141, 144, 156, 168, 174, 18$, 185, 188 Mahr, S. C., 68, 104 Maisel, G. W., 5, 59 Malamy, M., 193, 195, 228 Malcolm. B. R., 76, 103 Mallett, G. E., 317, 325 Malmgren, H., 92, 104 Mangiarotti, G., 209, 229 Manner, G., 200, 229 Manning, G. B., 317, 323 Mano, Y . , 218, 229 Manor, H., 197, 229 Marbaix, G., 196, 199, 201, 226, 229, 230 Marchis-Mouren, G., 214, 215, 226, 230 Marcker, K. A,, 205, 226 Marcus, P. I., 82, 98, 103 Margoliash, E., 173, I87 Margolis, F., 207, 227 Markert, C . L., 318, 326 Marks, P. A,, 201. 206, 207, 216, 224, 226, 227, 230 Marmur, J,, 109, 110, 111, 112, 117, 150, 156, 158, 165, 182, 185, 186 Maroudas, N. G., 180, 189 Marquardt, W. C., 312, 328 Marr, A. G., 306, 315, 316, 320, 326 Marre, E., 312, 326 Martens, J. W., 321, 326 Marth, P. C., 312, 326 Martin, D. G., 312, 324 Martin, J. H., 321, 326 Martin, W. G., 72, 100 Martinez, R. J., 212, 230 Marver, H. S., 148, 185 Massey, V., 310, 326 Master, R. W. P., 193, 221, 231 Matsuda, K., 198, 230 Matsumura, C., 311, 328 Mattern, C. T. F., 166, 167. 183 Matthaei, J. H., 197, 230 Mattoon, J. R., 185 Maturana, H. R., 69, 103

340

AUTHOR INDEX

Mayhew, E., 82, 84, 87, 88, 89, 90, 103, 105

Mazur, P., 304, 307, 312, 326 Meador, D. B., 320, 326 Meek, G . A , . 156, 186 Mellanby, E., 91, 101 Mellon, S. R., 20, 60 Mercer, E. H., 95, 96, 101. 103 Merishi, J. N., 89, 103 Merits, I., 193, 197, 230 Merker, H. J., 144, 151, 171, 172, 185 Meryman, H. T., 304, 307, 326 Merz. W. A , , 279, 281, 296, 301 Mesleson, M.. 149, IRS, 196, 227, 230 Meussner, R. A . , 247, 300 Meyer, R. R., 180, 189 Meyerhof, O., 311, 326 Meynell, G. G., 308, 326 Micou, J., 221, 222, 227 Migchelsen, C., 307, 323 Militzer, W., 316, 317, 323, 327 Milkman, R., 316, 326 Miller, A., 82, 103 Miller, D. S., 194, 229 Miller, L. L., 215, 228 Millette, R. L., 207, 227 Miura, K. I., 179, 189, 197, 230 Mizrahi, I. J., 204, 205, 230 Mizuno, D., 222, 225 Molotkovskii, Y . G., 311, 326 Mond, R., 3, 60 Monod, J., 49, 51, 52, 60, 210, 228. 3 1 3 , 315, 325, 326 Monroy, A,, 218, 229, 230 Monroy, G. C . , 308, 327 Montjar, M., 194, 199, 206. 213, 232 Mooney. H. A . , 319, 326 Moor, H., 263, 301 Moore, C., 110. 112, 150, 156, 165, 187 Moore, G . A., 12. 14, 59, 284, 301 Moore, P. B., 157, 178, 1 8 5 , 188, 197, 230 Moore, S., 83, 100 Moore, T. C., 312. 323 Morales, M. F., 56, Ij8 Morita, R. Y . , 310, 325 Morris, A,, 224, 230 Morris, D., 317, 326, 327 Morris, J. A,, 207, 231 Moss, A. J., Jr., 311, 312, 32s Moss, C. W., 315, 326

Mosteller, R. D., 201, 229 Mounolow, J. C., 112, 145, 158, 159, 160, 161, 1x5 Mous, W., 97, 102 Moustacchi. E., 112, 140, 156, 185 Mueller, G. C., 193, 197, 232 Mueller, P., 65, 103 Munkres, K. D., 176, 177, 1x5, IXX, 189, 192, 232 Munro, A . J., 198, 199, 200, 207, 214, 229. 230

Munro. H. N., 201, 227 Murray, K., 315, 324 Murray, R. K., 97, 102 Mustacchi, H., 67, 102 Muto, A . , 197, 230 Mysels, K. J., 34, 60

N Naaman, J., 82, 103 Nagano. H., 218, 229 Nakada, D., 195, 211, 224, 230 Naono, S., 195, 208, 211, 224, 228, 230. 233

Naota, H., 220. 230 Nash, T.. 312, 326 Nass, M. M. K., 84, 104, 109, 110, 118, 131, 140, 1 4 1 , 144, 145, 185 Nass, S., 109, 110, 132, 143, 144, 147, 148, 149, 1x5 Nathanson, A., 70, 103 Natori, S., 222, 225 Naylor, J. M., 316, 327 Negrotti, T., 156, 185 Nell, E. I!.. 312, 324 Nelson, B. D., 312, 323 Nemer, M., 218. 219, 2-78, 230, 232 Netter, H., 3, 60 Neubert. D., 146, 147, 149, 151, 152, 1 5 3 , 171, 172, 181, 185, 287 Neupert, W., 172, 182, 186 Ng, H., 315, 32G Nichols, S., 96, 100 Nielsen, J. M., 34, 60 Nirenberg, M. W., 197. 211, 222, 226, 230 Nishi, A , , 320, 326 Niyogi, S . K., 167, 186 Noll, A,, 213, 214, 232 Noll, H., 198, 199, 200, 201, 206, 209, 214, 220, 2.30, 232

AlJTHOR INDEX

Noller, H., 157, 178, 185. I88 Nomura, M., 196, 197, 230 Nord, F. F., 309, 326 Nordling, S., 82, 89. 10.3 Novick, A,, 314, 321 Nowell, P. C., 96, 100, 10.3 Numa, S., 310, 326 Nygaard, A . P., 198. 2.30

0 Oberdisse, E., 146, 147, 1-19. 151, 185 Ochoa, J., 199, 231 Ochoa, S., 320, 328 Ochsenfeld, M. M., 3 . 7, 9 , 25. 26, 27, 28, 29, 30, 31, 32, 60 Ockerse, R., 316, 326 Oda, T., 75, 101 O’Donovan, G. A,, 311. .326 O’Grady, E. A., 88. I03 Ogur, M., 140, 186 Ohlmeyer, P., 3 11 326 Ohnishi, T., 1 5 1 , 1 5 2 , I86 Ohno, M., 311, 328 Okada, Y . , 199, 232 Oki, T., 307, 325 Olcott, H. S., 86, 101 Oldenziel, H., 137, 181 Olivera, B. M., 121. 130, 188 Olson, A. C., 3 2 0 , 3-76 Oncley, J . L., 72, 103 Opara-Kubinska, 2.. 118. I84 Orci, L., 262, 300 Ortanderl, F., 319, 328 Oshinsky, C. K., 154, 18J Osterhout, W. J. V., 28, 60, 70, 103 Oura, H., 199. 206. 232 Overbeek, J. T. G., 79, 102, 103, 105 Overton, E., 2 , 3, 60, 95, 103

P Pack, B. A , , 241, 2 5 2 , 262. 263, 268, 273, 301

Palade, G. E., 91. 101 Pallansch, M. J.? 53. 60 Papa, S., 109, 187 Papaconstantinou, J.. 2 17, 230. 232 Pardee, A. B., 210, 230 Park, H., 248, 249, 279, 299 Parker, J.. 173, 187. 311. 326

34 1

Parkes, A . S., 311, 327 Parpart, A . K., 71, 103 Parsons, D. F., 67, 70, 74, 76. 77. 103, 171, 186 Parsons, J. A,, 115, 145, 152, 186 Parsons, P., 144, 1 5 2 , 153, 1 7 4 , 165, I86 Pasternak, G., 98, 103 Patrman. J. A., 310, 324 Pauli, A . W., 320, 326 Pauly, J. E.. 257, 300 Peaker, C . R., 34, 60 Pearson, P., 157, 178, 185 Penefsky, H. S., 310, 326 Penkett, S. A., 67, 100 Penman, S., 195, 198, 199. 201, 213, 217, 227, 230 Pentzer, W. T., 315, 326 Peraino, C., 215, 216, 231 Perdue, J. F., 73, 75, 101, 177, 183 Perez-Esandi, M. V., 82, I04 Perkins, H. J., 311, 312, 326 Perkins, W. H., 311, 325 Perl, W., 201, 206, 230 Perodin, G.. 159, 161, 162, 187 Prrutz, M. F.. 70, 103 Peterkofsky. €3.. 207, 230 Peters, R. A,, 69. 93, 101, I03 Petersen, T. G., 74, 100 Peterson, D. M., 305, 328 Pethica, B. A., 67, 85, 86, 97, 99, 100 Pfaff, E., 1 5 1 , 152, 184, 186 Pfeffer, C. R., 140, 181 Pfeffer, W., 2, 60 Pfefferkorn, L. C., 2 0 5 . 228 Pfleiderer, G., 319, 328 Pictet, R., 262, .?00 Piko, L,. 110, 120. 1 2 2 , 131, 180, 186, 189 Pincock, R. E., 304. 305, 3-35,, 326 Piperno, G., 112, 158. I82 Pitman, M. G., 309, 327 Pitot, H. C., 215, 216, 230, 231 Plagemann, P. G. W, 319, 327 Plaut, W., 221, 226, 231, 233 Plummer, D. T., 319, 327 Poche, R., 294, 301 Podolsky, R. J.. 56, 60 Podrazhanskaya, 2. L., 316, 323 Pogell, B. M., 313, 328 Polakis, E. S., 156, 186 Polge, C.. 311, 327

342

AUTHOR lNDEX

Polli, E., 110, 116, 182 Pollock, M. R., 212, 231 Porter, K. R., 91, 103 Postgate. J. R., 308, 320, 327 Potter, J. L., 91, 104 Potter, V. R., 199, 206, 214, 226, 232, 311, 327 Pouwels, P. H., 136, 137, 184, 186 Preer, J. R., Jr., 112, 144, 187 Preisig, R., 262, 301 Prescott, D. M., 191, 221, 222, 229, 231 Prestidge, L. S., 210, 230 Previc, E. P., 210, 231 Prince, L. M., 70, 104 Prose, P. H., 92, 101 Prosser, C. L., 317, 327 Pullman, M. E., 172, 186, 308, 327 Pulvertaft, R. J. V., 85, 103 Purdom, L., 98, 103 Purohit, K., 317, 327 Purvis, J. L., 151, I86

Q Quagliariello, E., 109, 187

R Rabinowitz, M., 110, 120, 130, 131, 141, 142, 143, 144, 145, 146, 148, 149, 155, 165, 166, 171, 183, 186, 187, 193, 194, 200, 227, 23 1 Racker, E., 76, 102, 177, 183 Radloff, R., 120, 124, 126, 127, 129, 133, 186, 188 Rafelson, M. E., 217, 226 Rafikova, F. M., 307, 320, 328 Rake, A . V., 196, 231 Ramanis, Z., 163, 187 Rampersad, O., 200, 231 Rancourt, M. W., 140, 181 Randall, M., 79, 102 Rapatz, G., 7, 21, 22, 60, 312, 327 Rapp, F., 98, 104 Ratcliffe, T. M., 88, 89, 91, 100, I05 Ray, D. S., 112, 139, I86 Rechcigl, M., Jr., 148, 185, 207, 231 Reeder, R., 217, 231 Reich, E., 112, 142, 149, 150, 185, 193, 194, 225, 227, 228, 231

140, 147, 176, 228,

131,

186,

Reichard, P., 151, 184 Reilly, C., 156, 175, 186 Reinders, 93, 103 Reiss-Husson, F., 67, 70, 102 Reithel, F. J., 310, 327 Rembarz, H. W., 294, 301 Rendi, R., 77, 78, 103 Resende, F., 321, 327 Revel, J.-P., 94, 103, 213, 231 Revel, M., 211, 213, 214, 231 Rhaese, H . J., 136, 186 Rhines, F. N., 247, 251, 300 Rich, A,, 114, 116, 182, 198, 199, 200, 227, 229, 232 Richardson, C. C . , 154, 187 Richardson, S. H., 74, 103 Richter, G., 191, 220, 228 Rickenberg, H. V., 212, 229 Riedwyl, H., 257, 259, 260, 286, 300 Rieske, J. S., 75, 101 Rifkind, R. A., 201, 206, 207, 224, 230 Riggs, A. D., 142, 184 Riker, A. J., 315, 323 Riley, F. L., 166. 167, 168, 183 Riley, M., 314, 324 Rinaldi, A. M., 218, 229, 230 Rinaldi, L. M., 97, 103 Ring, K., 308, 327 Ringelmann, E., 310, 326 Risebrough, R. W., 208, 228 Ritossa, F., 317, 327 Roberts, D. W. A,, 305, 310, 315, 316, 320, 321, 323, 327 Roberts, N. E., 200, 229 Robertson, J. D., 66, 69, 101, 103 Robinson, F. R., 294, 300 Roe, J. W., 93, 101 Rogers, H. J., 94, 10.3 Roman, A,, 210, 227 Roodyn, D. B., 92, 101, 109, 132, 160, 172, 175, 177, 179, 186, 192, 231 Rosas del Valle, M. R., 212, 219, 225, 231 Rosenberg, A. M., 307, 328 Rosenthal, T. B., 85, 102 Rosiwal, A,, 243, 301 Ross, R., 298, 301 Ross-Fanelli, A., 51, 60 Roth, T. F., 91, 103, 135, 186 Rotunno, C. A., 26, 60

AUTHOR INDEX

Rouiller, C., 262, 300 Rouvitre, J.. 2 1 1, 230 Rowley, P. T., 207, 231 Rownd, R., 165, 18T Rubin, H., 98, 104 Rudin, D. O., 14, 59, 65, 103 Ruhenstroth-Bauer, G., 88, 103 Ruhland, W., 3, 60 Runner, C. M., 110, 118, 120, 131, 139, 140, 143, 288 Rusch, H. P., 117, 152, 153, 1x2 Rush, M. G., 135, 186 Rustad, R. C., 221, 231 Ruttenberg, G. J. C. M., 109, 110, 115, 118, 120, 122, 124, 125, 127, 128, 129, 130, 131, 136, 139, 140, 141, 143, 144, 145, 154, 158, 165. 166. 168, 173, 179, 182, 184. 186, 188, 1x9 Ryser, H. J. P., 4, 60

343

Schellman, J. A,, 41, 42, 59, 61 Scherbaum, 0. H., 320, 323 Scherle, W. F., 240, 253, 266, 268, 273, 277, 279, 287, 292, 302 Schemer, K., 198, 199, 201, 213, 230, 231 Schiff, J. A., 112, 183 Schildkraut, C. L., 111, 154, 158, 165, I R T , 186

138,

Schimke, R. T., 207, 231 Schlessinger, D., 196, 209, 216, 222, 223,

112, 126, 138, 149, 177,

Schmidt, W. J., 70, 103 Schmieder, M., 152, 153, 187 Schmitt, F. O., 69, 97, 103, 104 Schneider, W. C., 110, 144, 147, 149, I87 Schreml, W., 85, 100 Schulman, H. M., 219, 231, 233 Schulman, J. H., 70, 104 Schuurmans Stekhoven, F.M.A.H., 110, 118, 120, 131, 138, 139, 140, 143,

Saccone, C., 151, 287 Saddler, H. D. W.. 309, 327 Sadoff, H. L., 316, 327 Sager, R., 163, 187 Sakai, A,, 312, 327 Salas, M., 199, 231 Salb, J. M., 82, 103 Salmon, S. C . . 315, 327 Salser, W.. 197. 209, 231 Salt, R. W., 306, 3 1 1 , 327 Saltykov, S. A.. 242, 246, 248, 257, 259, 260, 286, 301 Sanadi, D. R., 110. 112, 147, 150, 156, 165,

Schwan, H. P., 307, 327 Schwartz, D., 319, 327 Schwartz, H. A,, 259, 301 Schwartz, H. S., 214, 231 Schwartz, V. G., 82, 103 Schweet, R., 199, 206, 224, 226, 230, 232 Schweiger, E., 220, 226, 231 Schweiger, H. G., 193, 220, 221, 226, 231 Schwemmle, B., 316, 327 Schwindenvolf, U., 34, 61 Scornik, 0. A,, 205, 228 Scott, R. B., 217, 231 Sealock, R. W., 312, 324 Seaman, G. V. F., 73, 79, 81, 82, 85, 86, 87, 88, 89, 98, 99, 100, 102, 103,

229, 230, 232

S

182, 183

Sanford, B. H., 103 Sanghavi, P., 120, 140, 141, 142, 165, I87 Sanukida, S., 174, 184 Saroff, H. A,, 26, 5 9 Saunders, G. F., 321, 327 Saunders, G . W., 177, 178, 188 Schaechter. K., 210, 231 Schaechter, M., 210, 211, 231 Schapira, G., 200, 227 Scharff, O., 308, 327 Schatz, G., 140, 144, 155, 172, 174, 175, 186, 187

Scheil, E., 259, 301

188

104

Sebald, W., 172, 180, 186, 189 Sedat, J. W., 196, 228 Seed, R W., 217, 231 Seeds, A. E., Jr., 310, 31 1, 32J Seeley, H. W., Jr., 314, 323 Sekiguchi, M., 222, 223, 231 Sellschop. J. P. F., 315, 327 Semikhatova, 0. A., 319, 327 Shah, D. O., 70, 104 Shapiro, H. S., 136, 188 Shapiro, L., 180, 189 Sharp, C. W., 156, 184

344

AUTHOR lNDEX

Shatkin, A . J., 193. 231 Shen, S. C., 314, 324 Shepherd, J., 201, 227 Sheridan, J. W., 197, 227 Sherman, F., 156, 173, 175, 185, 186, 187 Sherman, J. K., 311. 327 Shikama, K., 310, 311, 316, 327 Shimazono, H., 3 1 1 , 328 Shipp, W. S.. 116, 187 Sidorova, A . I., 307, 328 Sidransky. H., 206, 231 Siegel, A., 198, 230 Siegel, B. Z., 316, 326 Siegrist, G., 262, 300 Siminovitch, D., 320, .327 Simon-Reuss, I . , 88, 102, 104 Simpson, G. M.. 316, 327 Simpson, M. V., 144, 152, 153, 154, 165, 180, 186. 189, 207. 227 Simura, T., 320, 327 Sinclair. J. H., 110, 1 1 2 . 114, 118, 120, 130, 131, 139, 140, 1 4 1 , 142, 143, 165, 166, 186, 187 Singer, M. F., 223. 231 Singer, S. J., 72, 73. 102 Sitte, H., 238, 279, 296, 208, 301 Sizer, I. w., 314, 327 Sjiistrand, F. S., 67, 68, 76, 78, 204, 262, 301 Skou, J. C . , 308 327 Skoulios, A., 67, 102 Slater, D. W.. 218, 2.31 Slater, E. c., 109, 187 Slayter, H. S., 200, 229 Slonimski, P. P., 112, 1.45, 155, 158, 159, 160, 161, 162, 174. 185, 187 Smearing, R. W., 80, 100 Smellie, R. M. S., 138, 283, 193, 197. 227 Smit, E. M., 122, 186 Smith, D., 180, 189 Smith, A. U., 311, 327 Smith, C . S., 242, 246. 247, 301 Smith, D. S., 67, 104 Smith, E. G., 67, 100 Smith, L. D. H., 315, 326 Smith, M. A,, 199, 231 Smith, M. W., 309, 317. 326, 327 Smith, W. H., 315, 324 Sodergren, J. E., 214, 23J Soeiro, R., 194, 231

Soffer, R., 195, 208, 224, 233 Solomon, A . K., 26 Somers, G. F., 311, 328 Somlo, M., 154, 155, 187 Sonenshein, G. E., 180, 189 Sordat, B., 257, 286, 300 Sorrentino, M., 86, 101 Spahr, P. F., 195, 208, 222, 223, 228, 231, 23 2 Spector, A., 217. 232 Spencer, D., 117, 286 Spencer, T., 219, 221, 232 Spiegehan, S., 196. 197, 198, 208, 218, 227, 228, 230. 231, 233 Spirin, A . S., 217, 218, 219, 226, 2317 Spiro, D., 293, 302 Sporn, M. P., 197, 226 Squires, C. L.. 320, 326 Staehelin, T., 198, 199, 200, 201, 206. 209. 213, 214, 2.30, 232, 232 Staubli. W., 237, 250, 262, 278, 281, 282, 285, 286, 287, 288, 289, 291. 292, 293, 298, .?01, 302 Stahl, F. W., 149, 185 Standish, M. M., 65, 100 Stanford, S. C . , 38, 5 9 Stanley, W. M., Jr., 178, 187. 199, 231 Starr, M. P., 10, 61 Steens-Lievens, A., 220, 2.?2 Stein, W. D., 83, 100 Steinbach. H. B., 3 , 61 Steinbeg, M. S., 97. 104 Sternberg, S. S., 214, 231 Stevens, B. J., 110, 118, 120, 130, 131, 140, 141, 142, 143, 165, 166, 187 Stewart, G. A., 201, 232 Stewart, I., 312, 327 Stewart, J. A,, 217, 230, 232 Stewart, J . W., 173, 187 Still, J. L., 174, 185 Stoeckenius, W., 68. 70, 74, 101, 104 Stoepel, K.. 294, 301 Stokes, J. L.. 315, 317, 327 Stokstad, I:. L. R.. 110, 115, 1.45, 184 Stone, J. D., 81, 99, 100 Straka, R. P., 315, 327 Strange, R. E., 308, 320, 327 Straus, W., 92, 104 Streisinger, G., 199, 232 Studier, F. W., 128, 187

AUTHOR INDEX

Stumpp, S., 198. 226 Stutz, E., 220, 232 Sueoka, N., 111, 114. 187 Sugiyama, N., 320, 327 Sullivan, C. Y.. 318, 327 Sullivan, J. F., 82, 103 Sultzer, B. M., 316, 3-77 Summer, J. B., 111, 3-78 Sun, B., 310, 3-74 Sundararajan. T. A,. 199, 3-32 Suntzeff, V., 96, 100 Sussman, M. V., 207, 232, 107, 328 Sussman, R. R.. 207, 232, 3 1 4 , 327 Suttie, J. W., 172, 186, 192, 231 Suyama, Y . , 110, 112, 114, 139. 141. 1.14, 169, 170, 179, 187, 189 Svensmark, 0.. 221, 1-32 Swartz, M., 117, 187 Sweeney, E. W., 207, 231 Swetly, P., 174, 117, 18X Swift, H . , 109, 110, 114, 120, 140* 141, 143, 144, 1-15, 1 5 5 , 175. 176, 181. 18$, 186, 187, 189, 216, 2-78 Swift, T . J., 7, 21, 36, S9 SylvCn, B., 92, 104 Symons, R . H., 1 1 5 , 18-7 Sypherd, P. S., 197. 23-1 Szer, W., 320, 3-78 Szybalski, W., 118, 1 8 i

T Taft, R. W., 38, 61 Tager, J. M., 109. 187 Takrta, K., 313, 3-78 Talalay, P., 310, 311, 3-15 Tamaoki, T., 197, 197, -73-7 Tamm, C., 136, 187 Tang, D. B., 264, 265, 300 Tatum, E. L., 193. 23Z Taylor, A . C., 312, 328 Taylor, C. B., 148, 181 Taylor, J. H., 67, 102, 142, 187 Tecce, G., 112, 158, 182 Tencer, R., 218, 226 Ter Schegget, J., 152. 153, 154, 188 Terumoto, l., 312, jZU, 328 Terzaghi, E., 199, 2.32 Tevethia, S. S., 98. 104 Tewari, K. K., 112, 116, 141, 144, 156, 161, 165, 168, 188

345

Thach, R. E., 199, 232 Tham, S. H., 155. 175, 185 Thiery. J., 5 2 , 58 Thomas. C. A,, Jr., 128. 167. 186, I88 Thomas. D. Y . , 161, 164, 177, 178, 181, 188, 189, 190 Thomas, L., 91, I 0 4 Thompson, T. L., 316, 328 Tien, H. T., 65, 103, I04 Tisdak, H. D., 7.1, 75, 100, 101, 177, 182 Tissisres, A , , 178, 188, 197, 23-7 Tobias, J. M., 35, 61 Todd, A. R., 83, 100 Tolbert, G., 223, 231 Tomkeieff, S. I., 247, 301 Tomkins, G. M., 196, 205, 207, 210, 211. 225, 227, 229, 230 Trakatellis, A. C., 194, 199, 206, 213, 232 Traut, R . R., 157, 178, 185, 188 Trautner, T., 117, 187 Triplett, E. L., 220, 232 Troschin, A . S., 12, 18, 20, 24, 60 Truman, D. E. S., 172, 188 Trunova, T. I.. 311, 3 1 2 , 328 Tschudy, D. P., 148, 185 Tsugita, A . , 199, 232 Tumanov, I. I., 304, 312, 3-17 Tung, Y.,52, SjX Tuppy, H. 109, 140, 144. 174, 175. 177, 187, 188

Tiistanoff, E. R., 181 Tyler, A,. 110, 120, 122. 111, 186, 218, 219, 232 Tzagoloff, A,, 75, 100, 102

U Lrdaka, S., 3 1 4 , 328 Iiemura, I . , 196, 228 Llllrich, H., 309, 3 1 1 , 328 [1mbarger, H. I:., 3 1 3 , 328 Ilnderwood, E. E., 240, 241, 242, 250, 256, 301, .?02

Ussing, H. H., 5, 59

V Valanju, S., 194, 22S VanBruggen, E. F. J., 110, 115, 118, 120. 122, 124, 125, 127, 128. 130, 131. 136, 137, 138, 139, 140, 141, 143, 158, 182, 184, 18G, 188

346

AUTHOR INDEX

Vanderhaeghe, F., 220, 226, 232 van Rotterdam, J., 136, 137, 186 Varner, J. E., 316, 328 Vasil’eva, I. M., 307, 320, 328 Vassar, P. S., 98, 104 Vatter, A. E., 77, 78, 103 Vaughan, N. H., Jr., 114, 116, 182 Verney, E., 206, 231 Vernier, R., 243, 299 Verzhbinskaya, N. A., 307, 328 Vestergaard-Bogind, B., 308, 327 Viehhauser, G., 181, 190 Villa-Trevino, S.. 209, 214, 232 Vinograd, J., 110, 114, 115, 120, 121, 122, 124, 125, 126, 127, 128, 129, 131, 133, 134, 135, 136, 143, 180, 181, 182, 184, 286, 188, 189 Vittorelli, M. L., 218, 229 Votsch, W., 112, 141, 144, 156, 168, I R R Vogt, P. K., 98, 104 Volkin, E., 83, 104, 197, 232 Von der Decken, A,, 204, 232 Vos, O., 311, 312, 328

W Waldron, J. C., 312, 324 Wallace, P. G., 155, 175, 181, 185, 188, 190 Wallach, D. F. H., 82, 89, 104, 307, 328 Wallis, 0.C.,151, 152, 183 Wang, G. T., 312, 328 Wang, J. C., 121, 130, I88 Wang, J. H., 312, 324 Wang, S. Y., 305, 328 Waring, M. J., 125, 126, 129, 161. 165, 167, 182, 188 Warner, J. R., 199, 200, 232 Warner, R. C., 135, 186, 310, 326 Warren, J. C., 305, 328 Warren, L., 84, 104 Wasemiller, G., 71, 74, 76, 99 Waterhouse, W. L., 321, 327 Watkins, J. C., 65, 100 Watkins, W . M., 94, 104 Watson, J. D., 195, 199, 208, 228, 232 Watson, R., 120, 121, 127, 129, 188 Watts, J. w . , 213, 228 Webb, T. E., 206, 214, 229, 232

Weibel, E. R., 236, 237, 239, 240, 241, 242, 250, 252, 253, 255, 261, 262, 266, 268, 269, 270, 273, 275, 277, 278, 279, 281, 282, 284, 285, 286, 287, 288, 289, 291, 292, 293, 294, 296, 298, 299, 300, 301, 302 Weibull, C., 10, 11, 61 Weil, R., 136, 188 Weinberg, J. W., 21, 58 Weinstein, R. S., 96, 104 Weisman, R. A., 67, 102 Weiss, L., 63, 65, 67, 81, 82, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95. 97, 99, 100, 101, 102, 103, 104, 105 Weissmann, G., 91, 92, 104, I05 Wells, R., 114, 188 Wells, R. D., 158, 188 Wen, W. Y., 307, 324 Wenner, C. E., 116, 188 Werz, G., 191, 193, 220, 221. 228, 2?l Wescott, W. C.,65, 103 West, M., 319, 326 Wetmur, J. G., 165, 188 Wettstein, F. O., 198, 199, 200, 201, 206, 209, 213, 214, 230, 232 Whabba, A. J., 199, 231 Wheeldon, L. W., 172, 186 Whiteley, A. H., 218, 232 Whiteley, H. R., 218, 232 Whitfeld, P. R., 117, 186 Wiberg, J. S., 317, 328 Wicke, E., 307, 328 Wicksell, S. D., 257, 259, 302 Wieland, T., 319, 328 Wiener, J., 293, 300, 302 Wiersema, P. H., 79, 105 Wigglesworth, V. B., 68, 105 Wilcox, M. S., 315, 325 Wildman, S. G., 116, 161, 165, I88 Wilkie, D., 109, 132, 155, 156, 160, 161. 164, 175, 177, 178, 179, 180, 181, 182, 185, 186, 188, 189, 190 Wilkinson, J. H., 319, 327 Will, S., 174, 185 Williams, E. J., 42, 5 8 Williams, K. I. H., 312, 328 Williams, M. A,, 298, 302 Williams, R. C., 10, 61 Williams-Ashman, H. G., 311, 325

347

AUTHOR INDEX

Williamson, A. R., 206, 232 Williamson, D. H., 112, 140, 156, 1RS Wilson, S. H., 206, 214, 232, 233 Wilt, F. H., 218, 219, 233 Winkler, K. C., 64, 105 Winnick, T., 196, 228 Wintetsberger, E., 109, 152, 153, 172, 175, 181, 188, 190 Wischnitzer, S., 262, 302 Witt, J., 175, 188 Woese, C., 195, 208, 224, 233 Wolstenholme, D. R., 110. 118, 120, 1 2 2 , 124, 127, 128, 1 3 1 . 136, 137, 118, 165, 166, 167, 168. 179, 180, 181, 183, 189, 190, 221, 233 Wood, T. H., 307, 328 Woodard, J. W., 216, 228 Woodbury, J. W., 35, 60 Woodward, D. O., 176, 177, 18S, 188, 189, 192, 223 Wool, I. G., 200, 231 Wooltorton, L. S. C . , 315. 321 Work, T. S., 172, 173>183. 186, 189, 192, 204, 229, 231, 315, 325 Wroblewski, F., 319. 327 Wiinsche, IJ., 312, 328 Wyman, J., 49, 51, 52, 60, 61

Y Yamazaki, I., 311, 316, 327 Yamoto, T., 67, 105

Yankofsky, S. A,, 198, 233 Yasumatsu, K.. 311, 328 Ycas, M., 175, 189 Yoshikawa, H., 136, 189 Yoshikawa, M., 197, 233 Yoshikawa-Fukada, M., 197, 233 Yotsuyanagi, Y., 156, 174, 189 Youngner, V. B., 315, 328 Yu. R., 175, 176, 189 Yudkin, M. D., 209, 212, 225, 233

z Zabin, I., 312, 320, 324, 326 Zahlet, P. H., 307, 328 Zak, R., 200, 231 Zamecnik, P. C., 224, 2 2 j Zech, A. C., 320, 326 Zehavi-Willner, T., 216, 226 Zeidman, I., 97, 103 Zetterqvist, H., 66, I05 Zhestkova, I. M., 311, 326 Zierler, K., 4, 60 Zimm, B. H., 42, 61 Zimmerman, E. F., 201, 233 Zimmerman, R. A,, 195, 209, 210, 211, 233 Zimmerman, S. B., 154, 185 Zondag. H. A., 310, 328 Zucker, W . V.,219, 233 Zwikker, C., 20, 58

Subject Index A Amino groups, cell surface charge and, 85. 86 Animal tissues, mitochondrial deoxyribonucleic acid, I 18121 alkali and, 136-137 closed circular duplex, 121-128 composition, 137-139 number of superhelical turns, 128- I 30 oligomers of, 133-136 size and circularity of, 130-133 Anisotropic systems, morphometric cytology of, 294-298 Anucleate state, definition of, 191 Association-induction hypothesis, solute distribution and, I 1-12 Autolysis, sublethal, cell periphery and, 9192

C Calcium binding, malignant cells, 96-97 Cell(s), ionic absorption, answers to criticisms, 31-36 malignant, calcium binding, 96-97 fine structure, 95-96 peripheries of, 94-99 surface charge, 97-99 physical state of water in, 7 search for better model, 10.11 Cell memhrane, barrier function of, 7-10 Cell periphery, enzyme activity and. 9 1-94 lipid biiayers and, 64-70 other models, 70-78 sublethal autolysis and, 91-92 surface pH and, 92-94 Cell surface, charge, amino groups and, 85-86 distribution, 90-91 dynamic aspects, 87-90 electrophoretic mobility and, 78-81 348

other ionogenic groups and, 86-87 ribonuclease and, 82-85 sialic acid and, 81-82 Charge, cell surface, 78-91

D Deoxyribonucleic acid, mitochondrial, addendum, 179-181 alkali and, 136-137 amount of, 143-145 anaerobiosis and, I 53-1 5 5 animal tissue, 118-139 base composition, 109-117 dosed circular duplex. 1 2 1-128 complementary strand differences, I 17118

composition of. 137-139 evolution and relation to nuclear, 167168 genetic function, 168-179 glucose repression and, 155-1 56 mechanism of synthesis, 149-15 2 mutagenic agents and, 156-163 nearest-neighbor frequencies, 1 17 nucleoside incorporation into. 152-I 5 4 number of superhelical turns, 128- 130 oligomers of, 133-136 plant and microorganism, 139-143 recombinations of, 163-164 renaturation studies, 164-166 size and circularity of. 130-133, 139-

143 timing of replication, 145-146 turnover of, 146-149

E Electron microscopy, sampling of tissue, 263-273 specimen preparation, 261-263 stereological analysis and, 273-286 Electrophoretic mobility, cell surface charge and, 78-81 1;nergy requirement, membrane punlps and, 5-6

349

SUBJECT INDEX

Enucleation, inhibition of ribonucleic acid synthesis, 193-196 physical. 192-193 Enzymes, petite mutants and, 174-176 reaction rates, temprraturr and, 314-316 Euraryotes. anucleate. d x a y of messenger ribonucleic acid and protein synthesis in, 2 1 2 22-7

Experimental pathology, morphometric cytology and. 293-294

G Genetic function, mitochondria1 deoxyribonucl-ic acid, 168. 178-179 changes in proteins and. 176-178 enzymes in peritc mutants, 17-1-176 extraniitochondrial protein and, 173174 hybridization experiments, 169-172 protein synthesis and, 172-173

I Ice. formation. extracellular, 304-305 intracellular, 305-309 Isozymic substitution, cold hardening and, 319-321 hypothesis, implications of. 321-322 nature of, 318-319 problems of testing, 321

L Lipid bilayers, cell periphery and, 64-70 Liver cell, niorphometric cytology. correlation with biochemical data, 293 general concepts, 286-287 specific methods, 287-292

M Membrane theory, energy requirement of pumps, 5-6 history, 2 - 5

Metabolic imbalance, prevention of, 316-318 temperature changes and, 3 13-3 14 Microorganisms, niitochondtial deoxyribonucleic acid of, 139-143 Mitochondria, deoxyribonucleic acid, addendum, 179-181 alkali and, 136-137 amount of, 143-145 anaerobiosis and, 154-1 5 5 animal tissue, 118-139 base composition. 109-117 closed circular duplex, 121-128 complementary strand differences, I 17118

composition of, 137-139 evolution and relation to nuclear. 167168

genetic function, 168-179 glucose repression and, 1 5 5 - 1 5 6 mechanism of synthesis, 149-15 2 mutagenic agents and, 156-163 nearest-neighbor frequencies, 117 nucleoside incorporation into, 1 5 2 - 1 54 number of superhelical turns, 128-130 oligomers of. 133-136 plant and microorganism, 139-143 recombination of, 163-164 renaturation studies, 164-166 size and circularity of, 130-133, 119143 timing of replication. 145-146 turnover of, 146-149 Models, cell periphery and, 70-78 Morphometric cytology, anisotropic systems, assessment of structure, 297-298 effect on stereological measurements, 295-297 sampling of tissues. 294-295 application to electron microscopy, sampling of tissue, 263-273 specimen preparation, 261 -263 stereological analysis, 273-286 classification of structures. 237-238 experimental pathology and, 293-294

350

SUBJECT INDEX

fundamental stereological principles, assessment of aggregate structures, 242256 basic parameters characterizing structures, 238-240 size distribution of particles, 257-261 terminology and symbolism, 240-242 variation in thickness of sheets, 261 liver cell, correlation with biochemical data, 293 general concepts, 286-287 specific methods, 287-292 present state and future possibilities, 298299 problem of measuring structures on sections, 236-237 purpose and aims of, 235-236

P Plants, cold hardening, effects and prevention of ice formation. 304-309 effects of low temperature on proteins, 309-313 metabolic imbalance and, 313-318 isozymic substitution and, 318-322 mitochondrial deoxyribonucleic acid of, 139-143 Polysomes, messenger ribonucleic acid and, 198-201 Procar yot es , anucleate, decay of messenger ribonucleic acid and protein synthesis in, 208212 Proteins, effect of low temperature. description of, 309-31 1 protection from, 311-313

synthesis, decay in anucleate cells, 208-222 messenger ribonucleic acid and, 201207 mitochondrial deoxyribonucleic acid and, 168, 178-179 Protoplasm, integrative functions, mechanisms, 36-57

R Ribonuclease, cell surface charge and, 82-85 Ribonucleic aicd, messenger, decay in anucleate cells, 208-222 enzymatic degradation, 222-223 initiation of decay, 223-224 quantitation by direct methods, 196.198 relationship to polysomes, 198-201 relationship to protein synthesis, 201209 synthesis, inhibition of, 193-196

S Sialic acid, cell surface charge and, 81-82 Solute distribution, mechanism, experimental evidence, 20-31 theoretical aspects, 12-20 Surface charge. malignant cells, 97-99

W Water, physical state in cell, 7

Y Yeast, mitochondrial deoxyribonucleic acid, anaerobiosis and, 154-15 5 glucose repression and, 155-156 mutagenic agents and, 156-163