INTERNATIONAL
REVIEW OF CYTOLOGY VOLUME92
Nuclear Genetics
ADVISORY EDITORS DONALD G. MURPHY H. W. BEAMS ROBERT G . E. MURRAY HOWARD A. BERN RICHARD NOVICK GARY G. BORISY ANDREAS OKSCHE PIET BORST MURIEL J . ORD BHARAT B. CHATTOO STANLEY COHEN VLADIMIR R. PANTIC W. J. PEACOCK RENE COUTEAUX MARIE A. DIBERARDINO DARRYL C. REANNEY LIONEL I. REBHUN CHARLES J. FLICKINGER JEAN-PAUL REVEL OLUF GAMBORG M. NELLY GOLARZ DE BOURNE JOAN SMITH-SONNEBORN WILFRED STEIN YUKIO HIRAMOTO HEWSON SWIFT YUKINORI HIROTA K. TANAKA K. KUROSUMI DENNIS L. TAYLOR GIUSEPPE MILLONIG TADASHI UTAKOJI ARNOLD MITTELMAN ROY WIDDUS AUDREY MUGGLETON-HARRIS ALEXANDER YUDIN
INTERNATIONAL
Review of Cytology EDITED BY G. H. BOURNE
J . F. DANIELLI
St. George's University School of Medicine St. George's. Grenada
Danielli Associates Worcester, Massachusetts
West Indies
ASSISTANT EDITOR K. W. JEON Department of Zoology University o j Tennessee Knoxville. Tennessee
VOLUME92
Nuclear Genetics EDITED BY J . F. DANIELLI Danielli Associates Worcester, Massachusetts
ACADEMIC PRESS, INC. 1984 (Harcourt Brace Jovanovich, Publishers) Orlando San Diego New York London Toronto Montreal Sydney Tokyo
COPYRIGHT @ 1984, BY ACADEMIC PRESS, INC. ALL RIGHTS RESERVED. NO PART OF THIS PUBLICATION MAY BE REPRODUCED OR TRANSMITTED I N ANY FORM OR BY ANY MEANS, ELECTRONIC OR MECHANICAL, INCLUDING PHOTOCOPY, RECORDING, OR ANY INFORMATION STORAGE AND RETRIEVAL SYSTEM, WITHOUT PERMISSION IN WRITING FROM THE PUBLISHER.
ACADEMIC PRESS,INC.
Orlando, Florida 3 2 8 8 7
United Kingdom Edition published b y
ACADEMIC PRESS, INC. (LONDON) LTD. 24/28 Oval Road, London NWI 7DX
LIBRARY OF CONGRESS CATALOG C A R D NUMBER: 52 - 52 03 I S B N 0-12-364492-5 PRINTED IN THE UNITED STATES OF AMERICA
84 85 86 87
9 8 7 6 5 4 3 2 1
Contents CONTRIBUTORS
......................................
vii
Nitrate Assimilation in Eukaryotic Cells NIGELS . DUNN.COLEMAN. JOHNSMARRELLI. JR., AND REGINALD H . GARRET^
I. I1 . 111. IV . V.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Biochemical Aspects of Nitrate Assimilation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Genetic Regulation of Nitrate Assimilation in Fungi . . . . . . . . . . . . . . . . . . . . . . . Genetics of Nitrate Assimilation in Plants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cell and Molecular Biological Advances in Nitrate Assimilation . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1
2 12 26 39 45
Endocytosis and Exocytosis: Current Concepts of Vesicle Traffic in Animal Cells MARKC . WILLINGHAM A N D IRAPASTAN
I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. Events at the Plasma Membrane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
111. IV . V. VI . VII .
Endocytic Vesicles: Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Kinetic Classes of Endocytosis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Compartmentalization in the Golgi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Functions of Endocytic and Exocytic Membrane Traffic . . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
51 53 66 72 12
79
85 85
Stability of the Cellular Translation Process T. B . L. KIRKWOOD. R . HOLLIDAY. A N D R . F. ROSENRERGER I. I1. 111. IV . V. VI . VII . VIII . IX .
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Models of Error Propagation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Evolutionary Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nature of Proof or Disproof . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Measurement of Error Rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Turnover of Proteins and Messages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Evidence from Prokaryotes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Evidence from Eukaryotes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
"
93 95 103 107 111
118 119 121 126 128
vi
CONTENTS
Chromosome and DNA-Mediated Gene Transfer in Cultured Mammalian Cells A . J . R . DE JONCEAND D . BOOTSMA
I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Technical Aspects of DMGT and CMGT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Uptake and Expression of Donor Genetic Materials in Recipient Cells . . . . . . . .
I1. 111. IV . V.
Applications of Gene Transfer in the Genetic Analysis of Mammalian Cells . . . . Conclusions and Future Prospects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
133 135 139 147 154 156
DNA Methylation in Eukaryotic Cells AHARON RAZINA N D HOWARD CEDAR
I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I1. The Methylation Pattern of Eukaryotic DNA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111. DNA Methylation and Gene Expression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
159 160 166 178 181
INDEX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . CONTENTS OF RECENTVOLUMES. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
187 191
Contributors Numbers in parentheses indicate the pages on which the authors' contributions begin.
D. BOOTSMA(133), Department of Cell Biology and Genetics, Erasmus University, 3000 DR Rotterdam, The Netherlands HOWARDCEDAR(1 59), Department of Cellular Biochemistry, The Hebrew University-Hadassah Medical School, Jerusalem 91010, Israel
A. J. R. DE J O N G E ~(133), Department of Cell Biology and Genetics, Erasmus University, 3000 DR Rotterdam, The Netherlands NIGELS. DUNN-COLEMAN (l), Central Research and Development Department, Experimental Station, E. I . DuPont Nemours and Co., Inc., Wilmington, Delaware I9898 REGINALDH. GARRETT( I ) , Department of Biology, University of Virginia, Charlottesville, Virginia 22901 R. HOLLIDAY (93), National Institute for Medical Research, The Ridgeway, Mill Hill, London NW7 I A A , England T. B. L. KIRKWOOD (93), National Institute for Medical Research, The Ridgeway, Mill Hill, London NW7 I A A , England IRA PASTAN(5l ) , Laboratory of Molecular Biology, National Cancer Institute, National Institutes of Health, Bethesda, Maryland 20205 AHARONRAZIN (159), Department of Cellular Biochemistry, The Hebrew University-Hadassah Medical School, Jerusalem 91010, Israel R. F. ROSENBERGER(93), National Institute for Medical Research, The Ridgeway, Mill Hill, London M 7 IAA, England
'Present address: Department of Microbiology and Parasitology, Free University of Amsterdam, Medical Faculty, Amsterdam, The Netherlands. vii
...
Vlll
CONTRIBUTORS
JOHN SMARRELLI, JR.* (l), Department of Biology, University of Virginia, Charlottesville, Virginia 22901
MARKC. WILLINGHAM (5l), Laboratory of Molecular Biology, National Cancer Institute, National Institutes of Health, Bethesda, Maryland 20205
2Present address: Department of Biology, Loyola University of Chicago, Chicago, Illinois 60626.
We regret to announce the death of Professor James F. Danielli on April 22, 1984, co-editor with Dr. Geoffrey H. Bourne of International Review of Cytology since publication of the first volume in 1952.
This Page Intentionally Left Blank
INTERNATIONAL REVIEW OF CYTOLOGY. VOL 92
Nitrate Assimilation in Eukaryotic Cells NIGELS. DUNN-COLEMAN,* JOHNSMARRELLI, JR., ? . I REGINALD H. GARRETT?
AND
*Central Research and Development Department Experimental Station, E.I. DuPont Nemours and Co. Inc., Wilmington, Delaware, and fDepartment of Biology, University of Virginia, Charlottesville, Virginia I
Introduction . . . . . . . . . . . Biochemical Aspects of Nitrate Assimilation. . . . . . . . . . . . . , . . . . . . A. General Features of Assimilatory Ni B. Fungal Nitrate Reductases . . . . . . . . C. Algal Nitrate Reductases.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D. Higher Plant Nitrate Reductases . . . . . . . . . . . . . . . . . . . . . . . . . . E. Assimilatory Nitrite Reductases . . . . , . . . . . . . . . . . . . . . . . . . . . F. Biochemical Regulation of Nitrate Assimilation . . . . . . . . . . . . . G. I n Vitro Complementation Studies on Nitrate Reductase . . . . . . 111. Genetic Regulation of Nitrate Assimilation in Fungi .... A. Genetic Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Autogenous Regulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Nitrogen Metabolism Repression of Nitrate Assimilation. . . . . . D. Possible Translational Control of Nitrate Assimilation . . . . . . . . E. Evidence for the Turnover of Nitrate Reductase . . . . . . . . . . . . . F. The Effect of Carbon Metabolism on Nitrate Assimilation. . . . . ____ IV. Genetics of Nitrate Assimilation in Plants . . . . . . . . . . .
I. 11.
.................................. V.
B . Conclusions .................................. Cell and Molecular Biological Advances in Nitrate Assimilation. . . . A. Somatic Hybridization Studies in Nitrate Assimilation . . . . . . . . B. Cloning of a Molybdenum Cofactor Gene from E . coli . . . . . . . C. Fungal Transformation Systems . . . . . . . . . . . . . . . . . . . . . . . . . . D. Plant Transformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E. Cloning Nitrate Assimilation Genes-Concluding Remarks. . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2
6 6 7 9 11 12 12 18 19 23 25 25 26 26 38 39 39 41 42 43 43 45
I. Introduction Nitrate is the predominant form of combined nitrogen available within our oxidative environment, and its assimilation is achieved through its biological reduction to ammonium. The subsequent utilization of ammonium to form amino 'Present address: Department of Biology, Loyola University of Chicago, Chicago, Illinois 60626. 1 Copyright $8 1984 hy Academic Press. Inc All rights of reprodualon in any form reserved ISBN 0-12-364492-5
2
NIGEL S. DUNN-COLEMAN ET AL.
and amido-N compounds provides the link between the plethora of pathways of organic nitrogen metabolism and the pathways of inorganic nitrogen assimilation of which there are only two: nitrate assimilation and dinitrogen (N,) fixation. Losada and colleagues (Losada et al., 1981; Guerrero et al., 1981) have calculated, on the basis of relative carbon and nitrogen content of plants and the levels of CO, fixed by plants, that roughly 2 X lo4 megatons of inorganic nitrogen are assimilated annually. Since Bums and Hardy (1975) estimated that the overall chemical and biological fixation of dinitrogen was 2 X lo2 megatons, it follows that nitrate' assimilation exceeds nitrogen fixation by over 100-fold. Thus, the preponderance of nitrogen acquisition by the biosphere occurs via nitrate assimilation. Its significance to agriculture is enormous. The capability to assimilate nitrate is possessed by certain bacteria, some fungi, and virtually all algae and higher plants. It is absent from the animal kingdom. Nitrate assimilation represents a substantial energy expenditure by the cell when compared with ammonium utilization since eight reducing equivalents are consumed in the reduction of nitrate to ammonium. Consequently, cells which assimilate nitrate regulate this pathway to avoid wasteful use of reducing power when the end product, ammonium, is available. The form of regulation adopted varies in accordance with the metabolic pattern and status of the cell type but the fundamental purpose of the regulation is the same: to effect an economy of existence. For example, photosynthetic tissues assimilating nitrate show a rapid biochemical inactivation of this process when ammonium is presented. On the other hand, more rapidly proliferating fungal cells regulate through nitrate assimilation-specific gene expression. This review addresses the biochemistry, genetics, and regulation of nitrate assimilation in eukaryotic cells; particular emphasis is placed on the genetics of nitrate assimilation with an attentive eye to emerging molecular biological studies. Nitrate assimilation both invites the investigation of molecular biologists wishing to understand the regulation of metabolic potentialities in eukaryotic cells and intices manipulation by genetic engineers seeking to enhance plant productivity through application of recombinant DNA technology. Neither group has yet met fulfillment but the opportunities for both seem most promising.
11. Biochemical Aspects of Nitrate Assimilation
As indicated, the assimilation of nitrate is achieved through the eight-electron reduction of this oxidized inorganic anion, resulting in the formation of ammonium. This transformation requires two enzymatic steps, the two-electron reduction of nitrate to nitrite followed by the six-electron reduction of nitrite to ammonium:
NITRATE ASSIMILATION IN EUKARYOTIC CELLS
NO?
2e
~
niuarc reductase
3
6r -
NOT -N H$ nitme rcduclaw
The enzymes mediating these steps are nitrate reductase (EC 1.6.6.1-3) and nitrite reductase (EC 1.6.6.4 and 1.7.7. l ) , respectively. OF ASSIMILATORY NITRATEREDUCTASES A, GENERALFEATURES
The biochemical aspects of assimilatory nitrate reductases have been extensively reviewed recently (Garrett and Amy, 1978; Hewitt and Notton, 1980; Beevers and Hageman, 1980; Guerrero et al., 1981; Vennesland and Guerrero, 1979; Losada et al., 1981). This section will initially focus on features shared by all assimilatory nitrate reductases and subsequently detail unique characteristics of the enzyme from algal, fungal, and higher plant sources. Assimilatory nitrate reductases are soluble, electron-transferring proteins, 200,000-300,000 in molecular weight. Electron transfer is generally regarded to be mediated by enzyme-bound heme iron, flavin adenine dinucleotide (FAD), and molybdenum cofactor. These components function as electron carriers between the physically separated pyridine nucleotide oxidation site and nitrate reduction site (Campbell and Smarrelli, 1978, 1983). The physiological activity involves the reduction of nitrate to nitrite occurring at the molybdenum site, with reducing power generated by pyridine nucleotides donated to the catalyst probably via a specific enzyme sulfhydryl group (Amy et al.. 1977). In addition to this physiological activity, apparent nonphysiological activities can be shown in vitro. The first is termed a dehydrogenase (diaphorase) activity in which nitrate reductase can mediate the pyridine nucleotide-linked reduction of one- or twoelectron acceptors such as ferricyanide, cytochrome c, or dichlorophenolindophenol. The molybdenum moiety is not involved in these reactions. The second type of activity involves the reduction of nitrate with reducing power generated either by reduced flavins or viologen dyes. The pyridine nucleotide reduction site is not involved in this reductase activity and electrons are added probably to either heme or molybdenum cofactor. The study of these partial activities has provided valuable insights into the overall nature of the enzyme since the characteristic response of these activities reflect intrinsic properties of the nitrate reductase. For example, these reactions have been found to be differentially inhibited. Sulfhydryl binding agents such as p-hydroxymercuribenzoate inhibit both the pyridine nucleotide-linked nitrate reductase and dehydrogenase reactions (Garrett and Nason, 1969; Schrader et al., 1968). Metal-binding agents such as cyanide inhibit only the reactions involving the molybdenum moiety, while having no influence over the dehydrogenase activity of nitrate reductase (Garrett and Nason, 1969; Hewitt and Notton, 1980).
4
NIGEL S . DUNN-COLEMAN ET AL.
Further, polyvalent monospecific antisera against nitrate reductase differentially inhibit the partial activities (Amy and Garrett, 1979; Funkhouser and Ramadoss, 1980; Smarrelli and Campbell, 1981). From these and similar studies, the general features of the typical assimilatory nitrate reductase have been revealed, as depicted in the following scheme: MVH
L
NADPH + [FAD + cytochrome-bsY7+ (Mo)] + N O 1 I cytochrome c , Fe3(CN),
Additional information and analogy provides a more complete description of overall electron flow from the binding and oxidation of reduced pyridine nucleotide to the terminal reduction of nitrate to nitrite. Pyridine nucleotides are thought to bind to a supersecondary structure of the enzyme called the dinucleotide fold (Solomonson, 1975). Dinucleotide folds are typically composed of about 120 amino acid residues in five or six parallel strands to form a P-sheet core, the strands being connected by a-helical intrastrand loops located above and below the P-sheet (Rossman et al., 1974). Following A-side oxidation of the reduced pyridine nucleotide (Guerrero et al., 1977), the flavin region of the enzyme becomes reduced. Flavin adenine dinucleotide appears noncovalently bound to nitrate reductase. The FAD is easily dissociable from the enzyme from Neurospora crassa (Garrett and Nason, 1967), while showing very tight binding in other assimilatory nitrate reductases. Amy et nl. (1977) described an important sulfhydryl group which apparently mediates electron transfer between NADPH and FAD. From FAD, electrons are subsequently shuttled to a 6-type cytochrome. First identified by Garrett and Nason (1967) for the enzyme from N . crassa, and subsequently confirmed for other nitrate reductases (Solomonson et al., 1975; Guerrero and Gutierrez, 1977; Notton et al., 1977; De la Rosa et al., 1981; Minagawa and Yoshimoto, 1982), it has been termed cytochrome bSs7 since it displays an a-peak at 557 nm in a reduced versus oxidized difference spectrum. The terminal electron acceptor of the nitrate reductase protein is thought to be the molybdenum cofactor. Molybdenum cofactor was once regarded as a proteinaceous component (Nason et d . ,1970). However, more recent studies reveal that it is most likely a low-molecular-weight, molybdenum-binding, urothionelike molecule termed molybdopterin (Johnson et al., 1980; Johnson and Rajagopalan, 1982). The actual oxidation states of molybdenum during nitrate reduction have not been established unambiguously (Jacob and Orme-Johnson, 1980). Although many of the biochemical features of assimilatory nitrate reductases appear well established, the actual structure and role of the components involved in electron flow are only poorly understood. Elucidation of these aspects, however, involves the use of large quantities of protein, a formidable task given the low
NITRATE ASSIMILATION IN EUKARYOTIC CELLS
5
cellular concentration and relative instability of all nitrate reductases. The low cellular concentration of this enzyme is attributed to its remarkable catalytic efficiency; its turnover number or molecular activity is 18,000 (as a point of comparison, the molecular activity of succinate dehydrogenase is 1200) (Garrett and Amy, 1978). B. FUNGAL NITRATE REDUCTASES
The best characterized fungal nitrate reductases are those derived from Neurosporu crussa (Garrett and Nason, 1969; Pan and Nason, 1978; Homer, 1983), Penicillium chrysogenum (Renosto et al., 198l), Rhodotorula glutinis (Guerrero and Gutierrez, 1977), and Aspergillus nidulans (Minagawa and Yoshimoto, 1982). Table I summarizes several of the important physical properties of these enzymes. With the exception of the enzyme from A . nidulans (see also Downey and Steiner, 1979), all appear to be homodimers, with molecular weights in excess of 200,000. All are specific for NADPH as pyridine nucleotide electron donor, display pH optima of approximately 7.5, and possess an easily dissociable FAD. All apparently homogeneous enzyme preparations except that from A . nidulans display specific activities greater than 100 unitslmg protein (1 unit being 1 kmol of nitrate reduced per minute). The best preparations of these fungal nitrate reductases were obtained using affinity chromatography (FAD-
TABLE I FUNGAL NITRATEREDUCTASES AND HIGHER PLANTNADH-NITRATE REDUCTASES
Source
Native molecular weight ( X 103)
Subunit molecular weight ( X lo3)
Reference
Fungi
N . crassa
230
N . crassa P . chrysogenum R . glutinis A . nidulans A . nidulans
290 200 230 180
Higher plants Squash Barley Tobacco Spinach
-
230 220 220 190
115 145
100 118 59 + 38 90
115 110 110
120 (major)
Garrett and Nason (1969); Pan and Nason ( 1978) Homer (1983) Renosto e t a l . (1981) Guerrero and Gutierrez (1977) Minagawa and Yoshimoto (1982) Tomsett (1983, personal communication)
Redinbaugh and Campbell (1983) Kuo ef a / . (1982) Mendel and Miiller (1980) Notton and Hewitt (1979)
6
NIGEL S . DUNN-COLEMAN ET AL
Sepharose or Blue Dextran-Sepharose) in combination with conventional techniques. C. ALGALNITRATEREDUCTASES The best characterized algal nitrate reductases are those from Chlorella vulgaris (Solomonson et al., 1974, 1975; Giri and Ramadoss, 1979; Howard and Solomonson, 1982) and Ankisfrodesmus braunii (De la Rosa, 1981). However, the literature reveals discrepancies regarding the subunit composition of the enzyme from Chlorella. Initial reports (Solomonson et al., 1975) described the Chlorella vulgaris nitrate reductase as a trimer with a native molecular weight of 356,000. Later studies (Giri and Ramadoss, 1979) concluded the enzyme to be a trimer of molecular weight 280,000. The most recent data (Howard and Solomonson, 1982) conclude that the Chlorella nitrate reductase is a homotetramer with a molecular weight of 360,000. The nitrate reductase from Ankistrodesmus braunii has been described as an octamer with a molecular weight of 370,000. In other significant features, pH optima, pyridine nucleotide specificity, prosthetic group involvement, and molecular activity, these two representative green algal nitrate reductases are not significantly different from each other or from fungal nitrate reductases. FAD is not easily dissociable from either algal enzyme, in contrast to the situation found in fungal nitrate reductases. D. HIGHERPLANTNITRATEREDUCTASES Immunological data suggest a degree of similarity between algal, fungal, and plant enzymes (Smarrelli and Campbell, 1981), but structural and catalytic diversity of the enzyme exists even within a single plant. However, the basis of this diversity is obscured by the difficulty in obtaining a protein unmodified by the purification procedures. Thus, discerning intrinsic differences is problematical. Well characterized enzyme has been obtained from corn, barley, wheat, spinach, squash, and tobacco (Campbell and Smarrelli, 1978; Redinbaugh and Campbell, 1983; Kuo et al., 1982; Sherrard and Dalling, 1979; Notton et al., 1977; Mendel and Muller, 1980). There is not only a diversity of structure between the well characterized leaf nitrate reductases, but also different enzymes in the same tissue (e.g., soybean leaves) and tissue-specific enzymes. Typically these enzymes have been distinguished with some reliability by their pyridine nucleotide specificity (e.g., corn leaves and roots). The majority of leaf nitrate reductases are highly specific for NADH. The physical properties of representatives of this group are also shown in Table I . This group of higher plant enzymes can thus be characterized as homodimeric with subunit molecular weights of approximately 115,000. All leaves thus far examined apparently contain this type of nitrate reductase, except those from Erythrina senegalensis (Stewart and Orebamjo,
NITRATE ASSIMILATION IN EUKARYOTIC CELLS
7
1979). This tropical legume contains an enzyme which can accept electrons from either NADH or NADPH. The bispecific NAD(P)H-nitrate reductase has been described in three general cases. First, it can accompany the typical NADH-nitrate reductase in the same tissue. This pattern was found in soybean leaves and cotyledons (Jolly et al., 1976; Orihuel-Iranzo and Campbell, 1980). In rice seedlings, the two enzymes respond to different induction conditions (Shen et al., 1976). Despite their pyridine nucleotide specificity differences, the presence of two nitrate reductases in one tissue has not yet been adequately explained. The second case of the bispecific nitrate reductase was reported for corn roots and scutella (Redinbaugh and Campbell, 1981; Campbell, 1978). Since corn leaves contain only the typical NADH-nitrate reductase, there appears to be tissue-specific control over synthesis of the bispecific enzyme. The third case is the synthesis of the bispecific enzyme in response to mutation of the NADH-nitrate reductase (Dailey et al., 1982a). Mutant barley seedlings which lack NADH-nitrate reductase activity (nar-la) have been found to contain a bispecific nitrate reductase which is absent from normal tissue. Thus, multiple genes for nitrate reductases apparently exist in higher plants. The total purification of higher plant nitrate reductase has required the application of affinity chromatography, the most prevalent matrix being blue-Sepharose. Rapid purification is essential since numerous higher plant proteases have been identified (Hageman and Reed, 1979). Proteinase inhibitors have been somewhat effective with leupeptin being the best inhibitor (Wray and Kirk, 1981). E. ASSIMILATORY NITRITEREDUCTASES Nitrite reductase, the second enzyme of the nitrate assimilatory pathway, catalyzes the six-electron reduction of nitrite to ammonium. The reduction proceeds as follows, apparently with no intermediates released in the reaction NO,
+
6e-
+ 8H+
+
NHa
+ 2H20
Although catalyzing the same reaction, the nitrite reductases from photosynthetic organisms (EC 1.7.7.1) are significantly different from those from nonphotosynthetic sources (EC 1.6.6.4), particularly in molecular weight and electron donor specificity. Nitrite reductases from several photosynthetic sources, including Chlorella fusca, spinach, squash, and Porphyra yezornsis, have been purified to homogeneity (Vennesland and Guerrero, 1979, and references therein). The nitrite reductase from the fungus N . crassa also has been purified to homogeneity (Greenbaum et al., 1978; Prodouz and Garrett, 1981). The two best characterized nitrite reductases have been obtained from spinach leaves (Vega and Kamin, 1977; Lancaster et al., 1979) and N . crassa (Garrett, 1978; Prodouz
8
NIGEL S. DUNN-COLEMAN ET AL.
and Garrett, 1981) and will be discussed as representative examples to compare nitrite reductases from photosynthetic and nonphotosynthetic organisms. Assimilatory nitrite reductases have in common the possession of iron-sulfur center and siroheme prosthetic groups. The iron-sulfur center in spinach nitrite reductase is organized as a tetranuclear Fe,S, cluster (Lancaster et al., 1979); the Neurospora nitrite reductase is thought to contain two such Fe,S, centers (Prodouz and Garrett, 1981). Siroheme is an iron tetrahydroporphyrin of the isobacteriochlorin type, containing eight carboxyl groups and having two adjacent pyrole rings reduced (Scott et al., 1978). Siroheme is also found in assimilatory sulfite reductase (Siege1 et al., 1974). The siroheme function likely serves as the site of binding and reduction of nitrite in Neurospora (Vega et al., 1975; Garrett, 1978) and spinach (Vega and Kamin, 1977). Siegel’s laboratory (Wilkerson et al., 1983) has evidence from Mossbauer spectroscopic studies that the Fe,S, cluster and the siroheme of spinach nitrite reductase undergo exchange interactions, indicating that these two centers are chemically linked. It thus seems probable that the nitrite-reducing center of assimilatory nitrite reductases is the chemically coupled Fe,S,/siroheme prosthetic pair: 6e-
4
[Fe4S4/siroherne]
G
NO, NHJ
+ 8H+ + 2H20
This basic nitrite-reducing unit is found in association with a 61,000-MW protein in photosynthetic organisms where ferredoxin serves as the electron donor (Vega and Karmin, 1977). In contrast, the enzyme from nonphotosynthetic organisms as typified by Neurospora C ~ U S S Uis a 290,000-MW homodimeric flavoprotein of 140,000-MW subunits which utilizes reduced pyridine nucleotide (either NADPH or NADH) as electron donor (Lafferty and Garrett, 1974; Greenbaum et al., 1978; Prodouz and Garrett, 1981). Thus, the following electron transfer schemes aptly represent the nitrite reductases of photosynthetic and nonphotosynthetic cells: Spinach.
6 Fdred-+ [Fe4S4/siroheme]+ NO,
Neurospora
3NAD(P)H + [FAD -+ Fe4S4/siroheme]
--f
NO?
It is interesting to note that nonphotosynthetic cells apparently require a relatively larger and more complex nitrite reductase in order to utilize the reduced pyridine nucleotides as electron donors since they lack ferrodoxin and the photochemical means to reduce it (Garrett, 1978). The comparative properties of spinach and Neurospora nitrite reductase are summarized in Table 11. Both classes of enzyme are also inhibited by sulfhydryl agents such as pHMB, the anions CN- and sulfite which are substrate analogs, and CO which binds avidly to nitrite reductase in which the siroheme moiety is reduced (Greenbaum et al., 1978). The Neurosporu nitrite reductase, like the nitrate reductase in this organism,
NITRATE ASSIMILATION IN EUKARYOTIC CELLS
9
TABLE I1 COMPARATIVE PROPERTIES OF ASSIMILATORY N I T R I TREDUCTAS~S ~ Property Molecular weight Subunits Subunit molecular weight Electron donor Flavoprotein Iron-sulfur clusters Siroheme Specific activity (+ma1 NO,
Spinacho 61,000 1
-
reducedimg protein)
61,000 Ferredoxin No One Fe4S4 Yes 108
N . crassab
290,000 2 (a2-type homodimer) 140,000 NAD(P)H Yes (FAD) Two Fe4S4 Yes 27
“Vega and Kamin (1977) and Lancaster el al. (1979). bGreenbaum et al. (1978) and Prodouz and Garrett (1981).
has the capacity to catalyze a number of partial electron transfer activities in vitro. These include a FAD-dependent NAD(P)H-diaphorase activity for which cytochrome c, ferricyanide, or dichlorophenol indophenol serve as electron acceptor, and a FAD-independent dithionite-nitrite reductase activity (Lafferty and Garrett, 1974; Vega, 1976). Further this enzyme can catalyze the two-electron reduction of hydroxylamine to ammonium in a reaction using NAD(P)H and requiring FAD. This NADPH-hydroxylamine reductase activity has no physiological significance because of the high K , for NH,OH, namely 3 mM.
F. BIOCHEMICAL REGULATION OF NITRATEASSIMILATION As previously mentioned, nitrate assimilation is an energetically expensive process, requiring four equivalents of reduced pyridine nucleotide per nitrate reduced. Consequently, regulation of this pathway, particularly by inactivation when sufficient ammonium levels are available, would be economically advantageous. Further, the efficient site of regulation would be the first step, nitrate reductase, effectively halting assimilation and preventing accumulation of the toxic metabolite, nitrite. Ammonium is not a direct feedback inhibitor of nitrate assimilation, nor are any of the primary amino compounds such as glutamine or glutamate. However, biochemical regulation of nitrate reductase activity through oxidation-reduction interconversion of the enzyme has been suggested (Losada, 1974). Losada demonstrated that algal and higher plant nitrate reductases could be inactivated in vitro by preincubation with reductants such as NADH or dithionite and then reactivated by oxidants such as ferricyanide. Losada suggested nitrate reductase could therefore exist in two interconvertible forms, an oxidized, active form and a reduced, inactive species. The generality of this mechanism is not universal; the N . crclssa nitrate reductase is not subject to reductive inactiva-
10
NlGEL S. DUNN-COLEMAN ET AL.
tion in vivo. Using purified nitrate reductase from Chlorella vulgaris,Lorimer et al. (1974) demonstrated that two components were required for rapid inactivation of nitrate reductase, NADH and cyanide. Cyanide reacted stoichiometrically with NADH-reduced enzyme to give a stable enzyme-cyanide complex having an association constant to 1Olo M . Further, stoichiometric amounts of cyanide were released when physiologically inactivated nitrate reductase, isolated from cells exposed to ammonium, was reactivated. This observation suggested a physiological role for cyanide in regulating nitrate reductase activity. These observations were expanded into a model for the metabolic regulation of nitrate assimilation (Solomonson and Spehar, 1977) in which cyanide was recognized as the simplest carbon-nitrogen compound and postulated to be pivotal in integrating and regulating carbon and nitrogen assimilation. Cyanide in vivo was hypothesized to arise from hydroxylamine and glyoxylate, hydroxylamine in turn being an intermediate product of nitrite reduction by nitrite reductase. The origin of CN- remains unclear and thus the model, though attractive, remains speculative. It is clear however that nitrate reductase is present in an inactive, cyanide-bound state in vivo in algal cells treated with ammonium or in nitrategrown cells in late log phase. Several high-molecular-weight inhibitors have been reported to affect higher plant nitrate reductases. These inhibitors can be classified as either proteolytic enzymes or binding proteins. Nitrate reductase proteases have been implicated in wheat leaves and maize roots (Wallace, 1974, 1975; Sherrard et al., 1979; Yamaya et al., 1980). Both inhibitors were found to be heat and EDTA sensitive. Other reports have implicated two nitrate reductase inactivators in N . c r a m (Walls et al., 1978; Horner, 1983). These inactivators which were apparently proteolytic and could be inhibited by EDTA and/or phenylmethylsulfonyl fluoride may be involved in the turnover of nitrate reductase (Section 111,E). Inhibitors which bind nitrate reductase have been reported for a number of plant species (Yamaya et al., 1980, and references therein; Jolly and Tolbert, 1978). Although proteases and other proteinaceous inhibitors do affect nitrate reductase activity, their functions remain generally uncharacterized and their roles in the regulation of physiological nitrate reductase activity thus are unclear. These agents increase the difficulty in obtaining an unaltered nitrate reductase protein during purification and thereby complicate the characterization of the native enzyme form. Reversible inactivation of nitrate reductase from higher plants, algae, or fungi has been achieved by either dialysis or gel filtration, with reactivation possible by EDTA or various amino acids (Smarrelli and Campbell, 1983; Ketchum et al., 1977). The biological significance of these effects is uncertain. In addition, higher plant nitrate reductase was found to have an affinity for heavy metals (Smarrelli and Campbell, 1983; Nason and Evans, 1953). However, this inhibi-
NITRATE ASSIMILATION IN EUKARYOTIC CELLS
11
tion-reactivation is unlikely to have physiological significance and is important only in the handling of nitrate reductase in vitro.
G. In Vitro COMPLEMENTATION STUDIESON NITRATEREDUCTASE In 1970, Nason and his collaborators (Nason et al., 1970) reported that NADPH-nitrate reductase activity could be reconstituted by combining extracts of a particular nitrate reductase-deficient mutant strain, nit-1 , with extracts of various other nit mutants strains or wild-type extracts. Subsequent experimentation (Ketchum et al., 1970; Nason et al., 1971) gave evidence that a molybdenum-containing component or molybdenum cofactor was provided by the wild-type or other extract added to the nit-1 extract, and this endogenous molybdenum cofactor complemented apo-nitrate reductase in the nit- 1 to form active NADPH-nitrate reductase molecules. These observations essentially gave birth to research on the nature of the molybdenum cofactor and the biological role of molybdenum, a rapidly expanding area of inorganic biochemistry beyond the scope of this review. Nevertheless, these observations had important bearing on our understanding of the structure and function of nitrate reductase. The nit-1 mutant contains a NADPH-cytochrome c reductase (diaphorase) activity displaying a sedimentation coefficient of 4.5 (4.5 S). This activity is increased in nit-1 mycelia grown on nitrate (Nason et al., 1970) and is cross-reactive to antinitrate reductase antibodies (Amy and Garrett, 1979). The conclusion is this NADPH-cytochrome c reductase is apo-nitrate reductase, i.e., molybdenumfree enzyme subunits, and that in the presence of a low-molecular-weight (< 1000) molybdenum-containing cofactor (molybdopterin), these apoprotein subunits dimerize to yield native NADPH-nitrate reductase (8 s). Further, it has recently been confirmed that the reconstructed enzyme exhibits properties similar to the wild-type enzyme (Homer, 1983). The in vitro complementation reaction has been instrumental in elucidating the nature of molybdenum cofactor and researchers have also utilized it as a means to characterize mutants deficient in molybdenum cofactor or apoprotein. Acidified xanthine oxidase is generally used as a source of molybdenum cofactor (Tachiki and Nason, 1983), while extracts of nit-1 are utilized as the source of apo-nitrate reductase to test for the presence of active molybdenum cofactor (for example, see Mendel et al., 1981, 1982a,b). Further, high concentrations of molybdate have been shown to restore the activity of some molybdenum cofactor mutants (Tachiki and Nason, 1983; Mendel et al., 1981). Experiments analogous to those described for N . crassa have been done with niaD and c m mutants from Aspergillus nidulans, again resulting in the formation of a functional nitrate reductase (Garrett and Cove, 1976). In vitro reconstruction of active nitrate reductase has also been performed with
12
NIGEL S. DUNN-COLEMAN ET AL
higher plants. Active nitrate reductase from molybdenum-deficient spinach leaves could be obtained by addition of acid-treated spinach nitrate reductase (Rucklidge et al., 1976) and, as cited, cnx-type mutants in tobacco were reconstituted by acid-treated xanthine oxidase (Mendel et al., 1981, 1982a,b), suggesting that they are defective in molybenum cofactor. Recently, barley mutants were analyzed by reconstitution experiments using spinach apoenzyme or molybdenum cofactor (Notton et al., 1983).
111. Genetic Regulation of Nitrate Assimilation in Fungi The regulation of nitrate assimilation in fungi occurs predominantly at the level of gene expression. In the presence of nitrate or nitrite, the enzymes of nitrate assimilation are induced, provided ammonium is absent. In the presence of ammonium, or reduced nitrogenous metabolites such as glutamine, nitrate assimilation is repressed even if nitrate is available. Thus, both induction and repression are operant here. These conclusions are based on the genetic analysis described below. A. GENETICANALYSIS 1. Aspergillus nidulans
A great deal of the understanding about the genetics and the biochemistry of nitrate assimilation in eukaryotic organisms has come from the study of this pathway in the ascomycete fungi Aspergillus nidulans and Neurospora crassa. For example, Birkett and Rowlands (1981) have recently undertaken a genetic analysis of nitrate assimilation in Penicillium chrysogerzum and shown it to be identical to that of A. nidulans (see Table V). Pateman and Cove (Cove and Pateman, 1963, 1969; Pateman et al., 1964; Pateman and Cove, 1967) initially screened A. nidulans for strains unable to utilize nitrate as a nitrogen source. These experiments led to the isolation of both structural and regulatory mutants of nitrate assimilation (Table 111). For example, niaD mutants although unable to grow on nitrate could utilize nitrite as a nitrogen source. MacDonald and Cove (1974) isolated several temperature-sensitive niaD mutants and showed that their nitrate reductase activity was thermolabile. These results substantiated the view that the niaD locus encodes the nitrate reductase apoprotein. A second class of mutants isolated, niiA, were unable to utilize nitrate or nitrite as the nitrogen source. These niiA mutants lacked nitrite reductase activity and the niiA locus is believed to be the structural gene for nitrite reductase (Pateman et al., 1964). A third class of mutants were unable to utilize nitrate or hypoxanthine as a nitrogen source since they lacked both nitrate reductase and xanthine dehydrogenase
TABLE TABLE111 111 FUNGAL NITRATE FUNGAL NITRATEASSIMILAIIUN ASSIMILATIONMUTANTS MUTANTS A.A niduluns . .crussu . nidulans N N crassa P .P.chwsogenum chwsogenum niaD niaD
nir-3 nit-3
niuA niaA
Nitrate reductare apnprotein structural gene
niiA niiA nirA nirA
nit-6 nit-6 nir-4/5 nit-4/5
niiA niiA nlrA nirA
areA areA
nit-2 nit-2
--
-
nit-1 nit- 1 nit-9ABC nit-9ABC
cmABC cmABC
Nitrite reductase apoprotein structural gene trans-acting reguhtoiy gene required for the expression of nitrate reductase and nitrite reductase under conditions of induction by nitrate trans-acting regulatory gene required for the expression of a wide range of ammonium-generating enzymes, including nitrate reductase and nitrite reductase Encodes Encodesananenzyme enzymefor forMoCo MoCosynthesis? synthesis? Possibly Possiblya asingle singlegene genewith with3 3intracistronic intracistroniccomplementation complementation groups contiguousgenes genescnxA cnxAand andcnxC, cnxC,with withcnxB CUB groupsoror2 2contiguous mutants mutantsbeing beinga adouble doubleloss lossclass class Encodes Encodesananinzyme enzymefor forMoCo MoCosynthesis? synthesis? cnxE CUEmutants mutantsarc aremolybdatc molybdatercpairable repairableand andmay maybebedcfective defective inininscnion insertionofofInolybdate molybdateinto intothc thecofactor cofactor
cmABC cmABC
-
-
I
I
cnrE clUE
-
-
cnxF cnxF
-
UL TH clUH
cnxG CnxG cnxJ cmJ
~~
cnxD cnxD cnxE cnxE
ClZIF” cnxF”
--
C fCI XU GU CP
nil-7 nit-7 nit-8 nit4
--
-
crnA crnA
-
-
-
---
~
Function Function
~
--
Encodes Encodesenzyme enzymefor forMoCo MoCobiosynthesis? biosynthesis? Structural MoCo? Structuralcomponent componentofofMoCo? Encodes Encodesananenzyme enzymefor forMoCo MoCobiosynthesis? biosynthesis? Possible regulatory locus, affecting MoCo levels It is not certain to which A. nidulans c m loci nit-7 and n i t 4 correspond Nitrate permease locus
ReTerence Reference
of N . c r a m , see Tomsett and Garrett (1980, 19x1) For reviews of A . nidulans, see Cove (1976a.b, 1979) and Tonisett and Cove ( 1979) Fnr reviews
P . ckqsogenum work by Birkett and Rowlands (1981) 3
I
+
Molybdenum cofactor (MoCo) loci. Mutations Mutationsininany anyone oneol’ ofthcsc theselac1 loci leads leadstotoaaloss lossofofnitratc nitratercductasc, reductase, purine purinehydrvxylase hydroxylase1,I ,and andpurine purine hydmxylasc activity hydroxylaseI1I1activity
I
Arst et al. ( I 982a)
Tonisett and Cove (1979); Brownlee and Arst 11983)
~
aIt is isnot notknown knownif ifthe thenomenclature nomenclatureused usedwith withthe theP.P.chtysogenum chrysogenumcn.r cnxmutants mutantsisisstrictly strictlyequivalent equivalenttotothe thecnrF cnxFand andcnxG cnxGmutants mutantsofofAA. . niduluns. nidulans.
14
NIGEL S. DUNN-COLEMAN ET AL
(XDH) activity (XDH activity in A . nidulans resides in two molybdoenzymes, called purine hydroxylases I and 11) (Pateman el al., 1964). These mutants were designed as cnx, indicating that they were defective in the biosynthesis of a cofactor needed for both nitrate reductase and xanthine dehydrogenase activity. It was found that these mutants could be further classified into seven complementation groups (cmABC, cnxE, c m F , c m H , and c m G ) at five genetic loci, the cmABC locus being possibly a single gene with three intracistronic complementation groups or two contigous genes, cnxA and cnxC, with cmB mutants being a double loss class. Arst et al. (1970) demonstrated that mutants at the cnxE locus were molybdate repairable, since high concentrations of exogenous molybdate partially restored both nitrate reductase and xanthine dehydrogenase activity. Similar molybdate-repairable mutants have been found in P . chrysogenum (Birkett and Rowlands, 1981) and in higher plants such as Nicotiana tabacum (Mendel and Muller, 1979). MacDonald and Cove (1974) have isolated temperature-sensitive alleles of cnxE, cnxF, and cnxH, but only cnxH temperaturesensitive mutants have a thermolabile nitrate reductase activity. They tentatively concluded that the cnxH gene product may have a structural role in the nitrate reductase molecule. However, the function of any cnx locus is presently unknown; it is thought that they encode for enzymes concerned with molybdenum acquisition and molybdenum cofactor biosynthesis. Arst et a f . (19824 isolated mutants at a sixth locus, cmJ. Two allelic mutants, cnxJl and 52, have reduced molybdenum cofactor levels, rendering these mutants more sensitive to toxic analogs such as methylammonium when grown on nitrate. They speculated that the cmJ locus has a regulatory role in molybdenum cofactor regulation. They also reported the isolation of cryosensitive cnxC mutants unable to utilize nitrate or hypoxanthine as a nitrogen source at 25°C but able to grow at 37°C. On this basis Arst et al. (1982a) suggested that the c r d gene product may also have a structural role in the molybdenum cofactor. Garrett and Cove (1976) undertook a biochemical study of the molybdenum cofactor mutants ofA. nidulans by in vitro complementation (see Section I1,G). They found that cohomogenization of mycelia from the niaD26 structural gene deletion mutant with mycelia from either cnxA6, cnxE29, cnxF12, cnnG4, or cmH3 resulted in the formation of NADPHnitrate reductase activity. Using this in vitro complementation of nitrate reductase activity to study the regulation of molybdenum cofactor synthesis, it was found that more molybdenum cofactor was present in mycelia grown with nitrate as a nitrogen source than with ammonium. Further, the results with niaD; nirA- , and niuD; nirAc double mutants indicated that the nirA gene (see Section II1,B) did not play a role in the regulation of molybdenum cofactor snythesis. Many mutations in the cnx and niaD genes lead to the loss of NADPH-nitrate reductase activity and to the constitutive synthesis of nitrite reductase in the absence of the inducers, nitrate or nitrite. A minority of cnx and niaD mutants retain an inducible nitrite reductase activity (Pateman et al., 1964). MacDonald
NITRATE ASSIMILATION IN EUKARYOTIC CELLS
15
et al. (1974) examined the subunit structure of nitrate reductase in wild-type, niaD and c m mutants by studying the sedimentation coefficients of the NADPH- cytochrome c reductase (NADPH-CR) activity of nitrate reductase. They found that the wild-type NADPH-CR had a sedimentation coefficient of 7.6 S (200,000 MW), whereas certain mutant strains had NADPH-CR of 4.5 S (90,000-100,000 MW). To explain these results, they proposed that the niaD gene specifies an NADPH-CR of 4.5 S and the nitrate reductase molecule is a dimeric molecule of two 4.5 S niaD products. cnx mutants which constitutively synthesized nitrite reductase were shown to have a NADPH-CR with a 4.5 S sedimentation coefficient, whereas most c m mutants with an inducible nitrite reductase had both the 7.6 S and 4.5 S species of NADPH-CR. The only exceptions to these findings were mutants at the cnxE locus. Mutants of this locus had either inducible or constitutive nitrite reductase activity, but only the 7.6 S species of NADPH-CR. Since cmE mutants are the only mutants in which exogenous molybdate will partially restore NADPH-nitrate reductase (Arst et al., 1970), it seems likely that cmE mutants are capable of synthesizing the metal-free cofactor which is then able to cause dimerization of two 4.5 S subunits to form 7.6 S enzyme. To explain the findings that cllx and niaD mutants have either inducible or constitutive synthesis of NADPH-CR and nitrite reductase, it was proposed that the nitrate reductase molecule somehow regulated its own synthesis. Such an autoregulatory model has been presented by Cove and Pateman (1969). Mutations which affect the integrity of the nitrate reductase enzyme such as c m or niaD mutants produce nitrate reductase constitutively. Mutants that do not affect the structural integrity of the nitrate reductase show wild-type patterns of inducibility for NADPH-CR and nitrite reductase (Section II1,B). The crn nitrate transport mutant ofA. nidulans is discussed in Section IV,A,8.
2. Neurospora crassa Recently, Tomsett and Garrett (1980, 1981) have undertaken an extensive genetic study of nitrate assimilation in N . crassa, complementing the biochemical study of Coddington (1976). Both studies indicate that the genetics and biochemistry of nitrate assimilation are very similar between N. crassa and A . nidulans (Table 111). The nit-3 gene encodes the nitrate reductase apoprotein and nit-6 apparently encodes the nitrite reductase apoprotein. The nit- 1, nit-7, nit-8, and nit-9 genes lack both nitrate reductase and xanthine dehydrogenase activity and appear very similar to cnx mutants of A. nidulans. nil-9 mutants fall into three complementation groups in a manner identical to that found in cnxABC mutants of A. nidulans. The biochemistry of molybdenum cofactor loci is less well known in N. crassa, and no molybdate-repairable locus similar to that of cmE or regulatory locus such as cnxJ has yet been identified. It is not known if one of the molybdenum cofactor loci in N . crassa shares a structural role similar to cmH, as tentatively proposed for A. niduluns. The nit-3 and nit-6 loci encod-
16
NIGEL S. DUNN-COLEMAN ET AL
ing nitrate reductase and nitrite reductase in N . crussu are on different linkage groups, unlike the situation in A. nidulans where niaD and niiA are very tightly linked. The nitrate assimilation enzymes in both these fungi have been shown to be inducible, and the increase in activity observed when inducer (nitrate or nitrite) is present is almost certainly the result of de n o w protein synthesis. Bahns and Garrett (1980) transferred uninduced cultures of N . crassa to medium containing 90% deuterium oxide plus nitrate and found that the nitrate reductase formed was of a uniformly heavy density. In the reciprocal experiment where mycelia were grown under noninducing conditions on media prepared in 80% D,O and then induced with nitrate in H,O-based media, nitrate reductase displayed a normal density. This result indicates that there can be little, if any, inactive or precursor nitrate reductase protein in cultures not exposed to nitrate or nitrite. Amy and Garrett (1979) used monospecific antibodies against N . crassa nitrate reductase and showed that uninduced wild-type cultures had no detectable cross-reacting material, again arguing against any inactive or precursor nitrate reductase being present under uninduced conditions. It is obvious from these results that nitrate reductase is under strict regulatory control. Several genes have been identified which affect the expression of nitrate assimilation enzymes (Table IV). The nirA and nit-4/5 loci of A . nidulans and N . crassa, respectively, play a role in the induction of the nitrate assimilation pathway (see also Section III,B), whereas the areA and nit-2 loci of A. nidulans and N . crassa, respectively, are concerned with nitrogen metabolite repression of a variety of pathways including nitrate assimilation (discussed in Section 111,C). Recessive mutations in the nirA or nit-4/5 genes result in noninducibility of nitrate reductase and nitrite reductase. A second class of semidominant mutant (nirA") at the nirA gene results in constitutive expression of these enzymes in the absence of inducers. (nirA" mutants are still subject to nitrogen metabolite repression of nitrate reductase and nitrite reductase.) nirA" mutants have a higher constitutive expression of nitrate reductase compared with nitrite reductase. This is an important finding because the niaD and niiA loci are very tightly linked in A. nidulans and there is the possibility that they may be coordinately regulated in the manner of an operon such as seen in prokaryotes. The differential expression of nitrate reductase and nitrite reductase in nirA" mutants limits this possibility. In addition, Arst et al. (1979) described a nonreciprocal translocation mutant, nis-5, which has a portion of linkage group I1 translocated into the region between the niaD and niiA genes (linkage group VIII). This mutant synthesizes 40% of the wild-type level of nitrite reductase upon induction but also constitutively synthesizes 8% of induced enzyme levels. Arst et al. explained these results by suggesting that the normal promoter for nitrite reductase is between the niaD and niiA genes and the translocation in nis-5 inserted a second promoter responsible for the low constitutive expression of nitrite reductase.
TABLE IV FUNGAL MUTANTSWHICHAFFECTTHE EXPRESSION OF NITRATE REDUCTASE A N D NITRITE REDUCTASE ~~~~
~~~
N . crussa
A. niduluns
Function
nit-415
nirA
Positive regulator required for the induction of nitrate reductase and nitrite reductase
nirA-/nit-415 nirAC nirAd
nit-2
areA
nmr- 1
Positive regulator required for expression of nitrate reductase and nitrite reductase. Action of this regulator mitigated by reduced nitrogen Two- to threefold elevation of nitrate reductase and nitrite reductase activities
urn
gdhA
NADP-GDH structural gene
gln- 1
glnA
ma-1
meaA, meaB
Probable glutamine synthetase structural gene Methylamine resistant, possibly altered in ammonium uptake and/or compartmentation
~
Comments
ureAr areAd
Noninducible mutants Constitutive expression of the enzymes, mutants of this class still subject to nitrogen metabolite repression Constitutive expression of the enzymes. Also insensitive to nitrogen metabolite repression. Selected as a suppressor of an areA'; nirACmutant Noninducible mutants Insensitive to nitrogen metabolite repression but require induction for expression
Four allelic mutants isolated with higher maximal levels of these enzymes and decreased sensitivity to nitrogen metabolite repression. Possibly altered for translational or hierarchical control of nitrate reductase and nitrite reductase synthesis Mutants at this locus grow poorly on ammonium and are not subject to ammonium repression of nitrate reductase and nitrite reductase Mutants may be insensitive to nitrogen metabolite repression Mutants at these loci are insensitive to ammonium repression of nitrate reductase and nitrite reductase
18
NlGEL S. DUNN-COLEMAN ET AL
3. Yeast At present, no nitrate reductase mutants have been reported in any yeast. The most widely studied yeast, Sacckarornyces cerevisiae and SckizoSaccharornyces pornbe, lack the ability to assimilate nitrate. However, the regulation of nitrate assimilation has been studied in Torulopsis nitru-atophila(Rivas et al., 1983) and Candida utilis (Prabhakara et al., 1976). Actinomycin D, an inhibitor of DNAdependent RNA synthesis, was shown by Prabhakara et al. to prevent the induction of nitrate reductase in C . utilis, in accord with the supposition that nitrateinduced synthesis of the enzyme requires transcription. Prabhakara's results also indicated that nitrate is not required for the translation of nitrate reductase mRNA. Thus, these results contrast with those of Sorger and Davies (1973) who concluded on the basis of studies employing inhibitors of macromolecular synthesis that nitrate was essential for translation of nitrate reductase mRNA, but not for the transcription of the nitrate reductase gene. Rivas et al. (1973) also showed that the T. nitratopkila system was similar to that of A . nidulans and N . crassa in that NADPH was the electron donor for nitrate reductase and nitrite reductase, in contrast to most plants where NADH and ferredoxin are favored. Zauner and Dellweg (1983) have recently reported the purification of nitrate reductase from Hansenula anornula.
B. AUTOGENOUS REGULATION To explain the findings that mutants such as niaD or cnx, which alter the structural integrity of nitrate reductase, result in constitutive expression of nitrate reductase, Cove and Pateman (1969) proposed an autoregulatory model. The recessive nature of nirA mutants in nirA- :nirA heterozygous diploids led Cove and Pateman (1969) to suggest that the nirA gene product plays a positive role in regulating the expression of both nitrate reductase and nitrite reductase. The finding that nirA" mutants are semidominant in nirAC:nirA heterozygous diploids (Cove, 1969) indicated that the nirA gene product was produced in limited quantities. Autogenous regulation can be rationalized as follows. Although nitrate reductase is an inducible enzyme, small but detectable quantities of the enzyme are present in cells in the absence of nitratdnitrite. In the absence of nitratdnitrite, nitrate reductase binds to the nirA gene product, thereby preventing the positive action of this gene product on niuD and niiA gene transcription. In the presence of nitrate, the nitrate reductasemitrate complex is unable to interact with the nirA gene product. As a consequence, transcription of nitrate reductase and nitrite reductase is activated. Mutations in either niaD or cnx genes which alter the overall structural integrity of nitrate reductase prevent the nirA gene product from binding, resulting in constitutive synthesis of both nitrate reductase and nitrite reductase. Tomsett and Garrett (198 1) found that certain N . crussa nit-3 mutants as well as molybdenum cofactor mutants such as nit-8 also ~
+
+
-
NITRATE ASSIMILATION IN EUKARYOTIC CELLS nit-3
&
nit-^
t
NaR apoprotein
19
nit-6
I
+
NiR apoprotein
I
y
Positive regulator J
Inactive complex
Nitrate reductase
+4
Nitrate reductase: NO; complex
FIG. I . Model for autogenous regulation of nitrate reductase in N . crussa. The nif-4/5 gene produces a positive regulatory element necessary for nit-3 and nit-6 expression. The nit-3 gene produces a protein which in conjunction with a functional molybdenum cofactor produces the native nitrate reductase molecule. If nitrate is absent, the nitrate reductase binds and sequesters the nit-415 product. Mutations in nit-3 or the molybdenum cofactor genes (nit 1,7,8,9) can result in an aberrant nitrate reductase incapable of interaction with the nif-4/5 regulator. Then, the nit-3 and nit-6 genes appear constitutively expressed.
have constitutive expression of nitrite reductase and nitrate reductase activities. The model for autogenous regulation is depicted in Fig. 1. As Marzluf (1981) points out, the proposed autoregulatory role for nitrate reductase, although accounting for the biochemical and genetical data, must remain speculative in the absence of direct molecular biological evidence.
C. NITROGENMETABOLISM REPRESSION OF NITRATEASSIMILATION The ammonium generated from a variety of pathways, including nitrate assimilation, is used by cells to form glutamine. First, glutamate is generated from 2oxoglutarate and ammonium via the action of NADP-specific glutamate dehydrogenase (NADP-GDH). Another equivalent of ammonium is then consumed through the action of glutamine synthetase (GS) converting glutamate to glutamine. IN. crassa (Hummelt and Mora, 1980; Dunn-Coleman et al., 1981b) and A . nidulans (Garrett and MacDonald, unpublished observations) also possess low levels of glutamate synthase (GOGAT) which forms glutamate from glutamine, 2-oxoglutarate and NADH. GOGAT is apparently important in these fungi only when NADP-GDH is absent or ammonium is limiting.] In addition to regulatory control over the expression of nitrate assimilation under conditions of induction, a second control mechanism limits the expression of this pathway and other ammonium-producing pathways when cells are grown with metabolites such as ammonium or glutamine. This phenomenon is called nitrogen metabalite repression and has been extensively studied in both A . nidulans and N . crassa. The nit-2 locus of N . crassa and the equivalent areA locus of A . nidulans play
20
NlGEL S. DUNN-COLEMAN ET AL.
key regulatory roles in the mediation of nitrogen metabolite repression in these fungi (Table IV). Both the nit-2 and areA genes are believed to encode positive trans-acting regulatory products required for the expression of genes which are subject to nitrogen metabolite repression. Pateman and Kinghorn (1977) had proposed that the tamA locus of A . nidulans played a role equally important as the areA locus in nitrogen metabolite repression. However, Arst et al. (1982b) have recently shown that tumA plays no major role. Two types of mutants have been isolated at the areA locus. The most common class, designated areA' (r for repressed), leads to an inability to utilize nitrogen sources other than ammonium or glutamine. A much rarer class, designated areAd (d for derepressed), results in specific derepressed expression of certain of the enzymes, such as nitrate reductase, which are usually subject to nitrogen metabolite repression. No derepressed mutants at the nit-2 locus have yet been isolated. This apparent lack may be the result of greater technical difficulties in isolating mutants in N . crassa. The isolation of several types of auxotrophic mutants impaired in the utilization of nitrogen has provided insights ino the mechanism of nitrogen metabolite repression and possible modes of action of the areA and nit-2 loci (see Table IV). Mutants possibly impaired in the uptake of ammonium are resistant to its toxic analog, methylamine; examples are meaA and meaB mutants in A. nidulans (Arst and Cove, 1969) and mea-1 in N . crassa (Dunn-Coleman and Garrett, 1984). These mutants have derepressed levels of enzymes such as nitrate reductase when grown with ammonium. Further, structural gene mutants for NADPGDH in both A . nidulans (gdhA) and N . crassa (am) are also insensitive to ammonium repression. These results led to the proposal by Pateman et al. (1973) that the NADP-GDH protein itself plays a regulatory role in ammonium repression. However, Dantzig et al. (1978) showed that am mutants, although insensitive to ammonium repression of nitrate reductase, have repressed levels of this enzyme when grown with amino acids such as glutamate. From these and other findings, it was evident that ammonium had to be assimilated before repression could occur. Dantzig et al. proposed the term nitrogen metabolite repression to accommodate these results. Glutamine synthetase in N . crassa has been extensively studied by Mora and Palacio's group (Davila et al., 1978, 1980, and references therein). They found that the oligomeric structure of this enzyme was dependent on the nitrogen status of the cell. When N . crassa was grown under conditions of nitrogen limitation, a tetrameric form of glutamine synthetase was present in the cell. Cells grown under conditions of nitrogen excess had an octameric form of the enzyme. Recently these researchers have shown that two similar but distinct monomers, a and p, may be involved in the formation of the different oligomers of glutamine synthetase. N . crassa grown under ammonium limitation has a tetrameric form of the enzyme being composed of a-subunits, whereas the octameric form of the GS found in cultures grown under nitrogen
NITRATE ASSIMILATION IN EUKARYOTIC CELLS
21
sufficiency consists of the P-subunit polypeptide (Davila et al., 1980). The subunit structure of glutamine synthetase in the two mutants gln- l a and gln-lb has been shown to be devoid of the P-polypeptide (Sanchez et ul., 1980). However, both of these mutants retain significant levels of glutamine synthetase activity, either as the tetrameric form in gln-la or both tetrameric and dimeric forms in the more extreme gln-lb mutant. Sorger’s group (Premakumar et al., 1979) and Garrett’s group (Dunn-Coleman et al., 1981a) have examined nitrogen metabolite repression in gln-la and gln-lb. Both gln-la and gln-lb have derepressed levels of nitrate reductase and nitrite reductase when grown in the presence of ammonium or glutamate. However, in gln- la, glutamine represses these enzymes. In contrast, gln- l b had varying levels of derepressed expression of these two enzymes when grown with glutamine. These findings with gln-lb (Dunn-Coleman et al., 1981a) stand in contradiction to those of Premakumar et al. (1979) who found that gln-lb had repressed levels of nitrate reductase when grown with glutamine. Two models have been suggested to explain nitrogen metabolite repression in N . crassa and A. nidulans. In the first, it has been proposed that glutamine itself is the effector of nitrogen metabolite repression by directly preventing the positive action of the nit-2 (or areA) gene product. One can envision a concentration-dependent interaction of glutamine and the nit-2 gene product, which is hypothesized to be a glutamine-binding protein. Presumably, once the concentration of glutamine reaches a sufficient value, its interaction with the nit-2 product would lead to a complex no longer capable of causing an activation of the genes encoding enzymes of pathways under nit-2 control. Evidence to support this view comes from Grove and Marzluf (1981) who reported the isolation of a nonhistone chromosomal protein from N . crussu nuclei having properties consistent with such a role for the nit-2 gene product. Grove and Marzluf radioactively labeled isolated nonhistone chromosomal proteins from N . crassa nuclei and chromatographed them on a N . crassa DNA-cellulose affinity column. After washing the column to remove nonspecifically bound material, they employed glutamine to elute any glutamine-binding proteins and found a single glutaminebinding, DNA-binding protein (MW 22,000 polypeptide). Two nit-2 mutant strains were shown to have diminished amounts of this protein. Marzluf (1981) postulated that glutamine, by directly interacting with the nit-2 gene product, reduced its binding efficiency for DNA enough that transcription of structural genes under nit-2 control (such as nitrate reductase) would not be activated. A second model to explain nitrogen metabolite repression has been put forward by Dunn-Coleman et al. (1979, 1980, 1981a). This model implicates a role for glutamine synthetase itself in mediating nitrogen metabolite repression. The principal justification for this model resides in the observation that the mutant gln-lb has substantially derepressed levels of both nitrate reductase and nitrite reductase when grown with glutamine. If glutamine were the direct effector of
22
NIGEL S. DUNN-COLEMAN ET AL.
nitrogen metabolite repression, one would expect gin-1 b to behave like wild-type and gln- l a and have repressed levels of nitrate assimilation enzymes when grown with glutamine. Dunn-Coleman et al. (1979, 1980, 1981a) proposed that the octameric form of glutamine synthetase which is found under conditions of nitrogen sufficiency directly or indirectly inactivates the nit-2 product. Nitrogen metabolite repression is thus understood as follows. When nitrogen becomes limiting, the tetrameric form of glutamine synthetase supercedes the octameric form in the cell, thereby freeing the nit-2 gene product to activate nitrate reductase gene expression. The summary diagram for the regulation of nitrate assimilation in N . CYUSSU presented in Fig. 2 encompasses these various interpretations. One of the difficulties in interpreting the role of glutamine and glutamine synthetase in nitrogen metabolite repression through gln mutant studies is that both gln-la and gln-lb have substantial amounts of glutamine synthetase activity. Recently, MacDonald (1982) has isolated a glutamine synthetase mutant
+
-iH4
NO~-NO;
I ,~[r
g
2-oxoglutarate
y
NADP-GOH
nit-1
&-7
*-8
G-9ABC
nit-4/5
fi-6
I’
&-2; 8
w
w Trans-acting positive regulatory gene for NO;/ NO2induction
G-31
I I/
tet--rttt-cF--c-c
Trans-acting positive regulatory gene, inactivated by nitrogen sufficiency (nitrogen metabolite repressed)
FIG. 2 . Model for the regulation of nitrate assimilation in N . crussu. Expression of the structural genes, nir-3 and nir-6 which encode the nitrate assimilation enzymes, nitrate reductase (NaR) and nitrite reductase (NiR) respectively, requires the action of two regulatory genes nit-415 and nit-2. nit-4/5 acts when nitrate (or nitrite) is present to serve as inducer; nit-2 acts only when nitrogen metabolite repression is relieved, i.e., the reduced nitrogen reserves are low. Glutamine and perhaps glutamine synthetase (GS) are essential for nitrogen metabolite repression. n i f - I , 7, 8, and 9 function in molybdenum acquisition and molybdenum cofactor (MoCo) biosynthesis. Nitrate reductase is dependent on MoCo for its physiological activity.
NITRATE ASSIMILATION IN EUKARYOTIC CELLS
23
in A . nidulans, &A-I, which lacks all glutamine synthetase activity. MacDonald found that this mutant had derepressed levels of nitrate reductase when grown with nitrate and glutamine. MacDonald’s results also tend to indicate that glutamine itself is not simply the corepressor of the areA or nit-2 gene product and would implicate glutamine synthetase as having a role in mediating nitrogen metabolite repression. However, as Marzluf (1981) points out, glutamine could be sequestered in a pool or compartment where it plays a role in nitrogen metabolite repression and this pool may be inaccessible to exogenously supplied glutamine. A genetic way to resolve the role of glutamine synthetase in nitrogen metabolite repression is as MacDonald (1982) suggested; that is, to map mutations which both abolish glutamine synthetase activity and relieve nitrogen metabolite repression, and see if they lie within the glutamine synthetase structural gene. Very recently, Gonzalez et al. (1983) studied the intracellular distribution of glutamine. Only 12% of the glutamine present in wild-type cells was sequestered in vesicles. In contrast, 50-60% of cellular arginine is in vesicles. These results tend to indicate that most glutamine is available for nitrogen metabolite repression. D. POSSIBLETRANSLATIONAL CONTROLOF NITRATEASSIMILATION Recently, another locus was identified in N . crassa which affects the expression of both nitrate reductase and nitrite reductase. Mutations at this locus, termed nmr-1 , were characterized by their sensitivity to the toxic analog chlorate even when glutamine is the nitrogen source (Table IV). [Under these growth conditions, wild-type cells have repressed levels of nitrate reductase and are resistant to chlorate (Dunn-Coleman et al., 1981a; Tomsett et al., 1981).] These mutants have up to threefold higher levels of nitrate reductase and nitrite reductase compared with wild-type. Immunoelectrophoretic determinations showed that the higher nitrate reductase activities in nmr-1 mutants were due to greater enzyme concentrations. Also, the half-life of nitrate reductase in these mutants was unaltered. Reduced nitrogen metabolites such as ammonium, glutamate, and glutamine retained the capacity to repress nitrate reductase in these mutants (Table V). Similar repression of nitrite reductase also occurs in nmr-1 mutants. Double mutants constructed between nmr-1 mutants and the regulatory gene mutants nit-415 and nit-2 indicated that these genes were epistatic to nmr-I. A reasonable interpretation is that the nit-2 and nit-4/5 genes exert their positive effects at the level of transcription of the nitrate reductase and nitrite reductase structural genes (nit-3 and nit-6, respectively). One possible explanation is that the nmr-1 gene product also acts on nit-3 and nit-6 expression at the transcriptional level. Several considerations militate against this view. First, all four known mutations in nmr-1 lead to enhanced enzyme levels, the inference being that loss of function in nmr-1 is the likely consequence of mutation and that such
24
NIGEL S. DUNN-COLEMAN ET AL. TABLE V THERESPONSE OF NITRATE REDUCTASE I N WILD-TYPE A N D nmr-l MUTANTS TO VARIOUS NITROGEN SOURCES" NaN03
Wild type nmr-1 (114) nmr-1 (183) nmr-1 (304) nmr-1 (319)
100 181 145 245 154
NaN03
+ NH4CI 40 45 49 66 38
NaN03
+ [.-glutamate 55 46 69 43 41
NaN03
+ L-glutamine 5 3 6 13 3
OThe results are expressed as the percentage wild-type nitrate reductase specific activity under conditions of maximum induction, 37.0 nmol nitrite producedlminlmg protein. Results adapted from Dunn-Coleman et al. (1981a).
loss in function does not reconcile with enhanced transcriptional activity. Second, a transcriptional role for nmr- 1 would implicate a third positive control element in the genetic expression of nitrate assimilation which would not be attended by any underlying physiological basis for its action (as in nitrate induction for nit-4/5, or nitrogen metabolite repression for nit-2). The possibility that a hierarchical arrangement of regulation exists between nmr-1 , nit-2, and or nit-4/5 merit consideration. Although the original studies on nmr- 1 afford no insight in this regard, the possibility that nmr-1 encodes a repressor antagonistic to either nit-2 or nit-4/5 would be compatible with this observation. The epistasy of nmr-1 to nit-3 is not only evidenced by enhanced nitrate reductase expression with the wild-type nit-3 allele but also with a mutant nit-3 allele (no. 14789) which has enhanced levels of the partial nitrate reductase activity, MVH-nitrate reductase. Another possible explanation for nmr-1 gene function is that its gene product acts at a posttranscriptional level on nitrate assimilation. Possibly, enhanced rates of translation of nitrate assimilation enzyme-specific mRNAs or increased rounds of translation of these messages would account for the results. Premakumar et ul. (1978) have shown that N . crussu nitrate reductase mRNA has a half-life of only 8.5 minutes. The higher nitrate reductase and nitrite reductase in nmr- I mutants may result because their specific mRNAs are more stable in these mutants and greater amounts of enzyme can be synthesized. That is, the nmr-1 gene product may govern nitrate assimilation mRNA degradation. This possibility has not been tested and the precise role of nmr-1 remains unknown. Nevertheless, the specificity of its effects in enhancing the enzymes if nitrate assimilation is striking. Using an experimental strategy suggested by the studies of Tomsett et al. (1981), Sorger's laboratory (Premakumar et al., 1980) subsequently isolated an unusual N . crussu mutant also altered in nitrate reductase regulation. This single mutant, MS5, had elevated levels of nitrate reductase, nitrite reductase, histidase, and acetamidase even in the presence of glutamine.
NITRATE ASSIMILATION IN EUKARYOTIC CELLS
25
(The latter two enzymes were not examined in nmr-1 mutants by Dunn-Coleman et al., 1981a.) Sorger’s results also showed that the half-life of nitrate reductase in MS5 was significantly altered. The MS5 mutant has now been mapped was shown to be an allele of nmr-1 (Debrusk, unpublished).
E. EVIDENCE FOR
THE
TURNOVER OF NITRATE REDUCTASE
Cove (1966) showed that the absence of inducer and the presence of ammonium caused rapid loss of nitrate reductase activity in wild-type cells of A . nidulans. Hynes (1973) and Dunn-Coleman and Pateman (1977) confirmed and extended these findings, showing that cycloheximide, a protein synthesis inhibitor, did not significantly reduce this loss of nitrate reductase activity. Amy and Garrett (1980) obtained similar results in N . crassa and showed that when mycelia were exposed to ammonium plus cycloheximide, nitrate plus cycloheximide or nitrogen-free media or media lacking a carbon source, the amount of material cross reactive with anti-nitrate reductase antibodies declined in concert with nitrate reductase activity. Amy and Garrett concluded that the loss of nitrate reductase activity was attributable to the loss of this protein. It seems likely that proteolytic degradation of nitrate reductase occurs. Walls et al. (1978) and Sorger et al. (1978) have attempted to examine the nature of this proteolysis in N . crassa. They presented evidence for two different decay mechanisms, a general protein turnover system (designated N) and a second system (designated A) which increased in proteolytic activity with mycelial age. Sorger et al. also presented in vitro evidence for two different nitrate reductase inactivator systems, I and 11. Inactivator I was present in all mycelia and may correspond to system A. Inactivator I1 was nitrogen repressible and missing in nit-2 mutants and may correspond to system N. Lewis and Fincham (1970) studied the regulation of nitrate reductase in the basidiomycete Ustilago maydis. Ammonium ions were shown to repress nitrate reductase activity. Cycloheximide and actinomycin D (an inhibitor of mRNA synthesis) led to maintenance of nitrate reductase activity even in the presence of ammonium. Lewis and Fincham further showed that the loss of nitrate reductase activity in U . maydis upon exposure to ammonium was attributable to turnover of this protein, not a reversible enzyme inactivation. F. THEEFFECTSOF CARBON METABOLISM ON NITRATEASSIMILATION Both nitrate reductase and nitrite reductase in fungi use NADPH as electron donor. Hankinson and Cove (1974) examined the regulation of the pentose phosphate pathway (PPP) in A . nidulans, which generates NADPH via the action of glucose-6-phosphate dehydrogenase (G6PDH) and 6-phosphogluconate dehydrogenase (6PGDH). Growth of wild-type A . nidulans in media containing urea plus nitrate resulted in a twofold elevation of G6PDH and 6PGDH activities
26
NIGEL S. DUNN-COLEMAN ET AL
compared with levels in mycelia grown with urea alone. Hankinson and Cove (1974) also examined various nitrate assimilation mutants and showed that the mutant noninducible for nitrate assimilation, nirA , did not show any elevation of PPP activities. On the other hand, niaD nitrate reductase apoprotein mutants and cnx molybdenum cofactor mutants could be divided into two classes. One class of niaD and c m mutants had elevated pentose phosphate pathway enzymes when grown with nitrate. The second class did not show such enzyme elevation. Hankinson and Cove (1974) proposed that the nirA+ product is responsible for the increases in the activities of the pentose phosphate pathway enzymes. Hankinson (1974) also described two mutants @ppA andpppB) which were impaired in their ability to reduce NADP+. These two mutants utilized nitrate poorly. Dunn-Coleman and Pateman ( 1979) examined hexokinase and phosphoglucomutase (PGM) activities in A . nidulans. Hexokinase activities did not vary with changes in carbon or nitrogen source. However, PGM activities were low in wild-type cells grown with nitrate and higher when cells were grown on ammonium. These findings seemed surprising at first, for it was anticipated that nitrate-grown cells would have elevated hexokinase activity to supply increased amounts of glucose 6-phosphate to the pentose phosphate pathway for NADPH generation. The findings were rationalized by a consideration by PGM regulation. PGM activity is lower in nitrate-grown cells, thus the rate of conversion of glucose 6-phosphate to glucose 1 -phosphate is lowered, allowing greater utilization of glucose 6-phosphate by the pentose phosphate pathway and a consequent enhancement of NADPH production. Schwytzer and Garrett (unpublished) examined pentose phosphate pathway activity regulation in N . crassa and were unable to find elevated enzyme levels in mycelia grown with nitrate. They concluded that the flux of glucose through the pentose phosphate pathway could vary with the nitrogen source such that cellular NADPH demands were met without changes in enzyme activities. ~
IV. Genetics of Nitrate Assimilation in Plants A. GENETICANALYSIS Two approaches have been successfully used to isolate nitrate reductase mutants in different species of plants (Tables VI and VII). The first has been the use of plant tissue culture, in which cell suspensions or protoplasts are screened for resistance to chlorate, the toxic analog of nitrate (Alberg, 1947). The second approach has been to mutagenize seed (M1 generation) and screen seedlings in the segregating M2 generation for low nitrate reductase activity. This approach, in the case of cereals such as barley (Hordeum vulgare), overcomes one of the serious difficulties with cereal tissue culture, that is, the lack of success in
NITRATE ASSIMILATION IN EUKARYOTIC CELLS
27
regenerating plants from cell suspension or protoplasts. Both approaches have yielded nitrate reductase structural gene mutants and molybdenum cofactor mutants, but no loci analogous to the regulatory loci seen in fungi have yet been described in higher plants. 1. Nicotiana tobacum and Nicotiana plumbaginijolia Nitrate Reductuse Mutants Miiller’s group at Gatersleben (GDR) was the first to successfully use tissue culture to select for nitrate reductase mutants in allodihaploid ( n = 24) cell cultures of Nicotiana tobacum (Muller and Grafe, 1978; Mendel and Muller, 1979, 1980). Their approach was to either mutagenize with ENU or select for spontaneous resistance to chlorate in wild-type cell suspensions. To date, they have isolated 40 variant cell lines, 36 of which are defective in the nitrate reductase apoprotein (niaA) and 4 lines defective in the synthesis of the molybdenum cofactor (cnxA). From the work with A . nidulans and N . c r a m , it is known that at least 9 genes are involved in nitrate assimilation. To overcome a possible problem of isolating a wide range of recessive mutants in N . tobacum, Maliga’s group in Szeged (Hungary) have recently used N. plumbaginifolia as a model system to isolate nitrate reductase mutants. Mutant isolation was carried out using haploid protoplast cultures, and 4 different types of nitrate reductase mutants ( 1 apoprotein mutant and 3 cofactor mutants) were isolated (Marton et al., 1982). It is interesting to note that Muller and Grafe (1978) treated 2 X lo7 cells with 0.25 mM ENU and isolated 5 nitrate reductase-defective lines. Maliga’s group used 8 X lo4 protoplast-derived colonies with 0.3 mM ENU and found 92 chlorate-resistant colonies (Marton et al., 1982). Only a small fraction of the chlorate-resistant clones in N . plumbaginijolia completely lack nitrate reductase, whereas most of the N . tabacum niaA mutants have no detectable enzyme activity. Miiller and Mendel (1982) have now regenerated fertile mutant plants from 15 niaA lines. They found that the nitrate reductase deficiency in these mutants is the result of two unlinked recessive mutations. Thus they concluded that NADH-nitrate reductase in N . tobacum is encoded by duplicate structural genes which they named niaAl and niaA2. The presence of duplicate genes for nitrate reductase in N. tobacum is probably the result of its allodihaploid nature (Miiller, 1983). Six chlorate-resistant lines (NA-type) of N. plumbaginijolia have similar characteristics to niaA mutants of N . tobacum. They are strictly auxotrophic on nitrate media and have xanthine dehydrogenase activity. Two of the mutants (NA8 and NA14) retain a residual amount of nitrate reductase activity. Unlike the niaA mutants of N . tobacum, the N . plumbaginijolia mutants are probably the result of a single mutation. Mendel and Miiller (1979) have also studied the biochemistry of nitrate assimilation in wild-type and mutant cells of N . tobacum. The electron donor for nitrate reductase is NADH; NADPH will not function. There does not appear to
28
NIGEL S. DUNN-COLEMAN ET AL
be either a biphasic NAD(P)H-nitrate reductase or a NADPH-nitrate reductase present in N. tobacum. None of the niaA mutants had any detectable NADH-, FADH,-, or BVH-nitrate reductase activity, with the exception of niaA-95 which retained some FADH,- and BVH-nitrate reductase activity. Cytochrome c reductase (NADH-CR) activities were low in all niaA mutants and this activity was not inducible by nitrate. Mendel and Miiller (1979, 1982) have shown that nitrate reductase and nitrite reductase activity are inducible in wild-type cells, confirming the original observation of Filner (1966). Cycloheximide prevented induction indicating that the increase in activity was likely due to de novo protein synthesis. Nitrite reductase levels in the niaA mutants were similar to those of the wild-type, indicating that nitrate reduction is not needed for the induction of nitrite reductase. Miiller and Mendel (1982) also examined nitrate reductase activity in the niaA1/2 double mutant seedlings. Ten niaA1/2 mutant lines were incapable of growth with nitrate as the sole nitrogen source and showed a nitrogen-starved appearance. These results indicate that NADH-nitrate reductase is essential for the utilization of nitrate as the sole nitrogen source. Two other niaA1/2 mutants (129 and 134) retained a residual nitrate reductase activity. Seedlings of these mutants grew on nitrate at reduced rates, initially having 5% and eventually 60% of the wild-type in vivo nitrate reductase activity (determined by nitrite accumulation). Yet, Miiller and Mendel observed that cultured cells derived from these mutants behaved like strict auxotrophs on nitrate medium. One possible explanation which may account for the nitrate reductase activity observed is in vivo complementation between defective nitrate reductase polypeptides produced by the mutated niaA1 and niaA2 genes. Cultured cells may not have sufficient time to produce enough nitrate reductase to grow on nitrate, whereas seedlings may slowly accumulate nitrate reductase and become capable of utilizing nitrate. More recently, Evola (1983) has isolated a chlorate-resistant cell line (clr 19) which lacks nitrate reductase activity and thus may be allelic to the niaA loci. Two other variants, clr 15 and clr 30, had lower resistance to chlorate and could still assimilate nitrate. 2 . Molybdenum Cofactor Mutants in Nicotiana and Datura Two types of molybdenum cofactor mutants have now been isolated in N . tobacum, the cnxA mutant isolated by Muller’s group and a second cnx type recently isolated by Buchanan and Wray (1982) (Table VII). Mutants in either cnxA or the second cnx locus (called cnxB in Table VII) lack both nitrate reductase and xanthine dehydrogenase. However, cnxA mutants are molybdate repairable both in vivo and in vitro; that is, when grown with high concentrations of molybdate, they develop 30% of the fully induced wild-type level of nitrate reductase (xanthine dehydrogenase activity is also partially restored). Grafe and Miiller (1983) have recently shown that the four cnxA mutants are alleles of the
NITRATE ASSIMILATION IN EUKARYOTIC CELLS
29
same pair of duplicate loci (cnxA1 and cnxA2). The cnxB-type mutants are not molybdate repairable. The molybdenum cofactor from cmA, but not cnxB, mutants is capable of restoring NADPH-nitrate reductase activity in cell-free extracts of the N . crassa nit-l molybdenum cofactor mutant. The cnxA mutant also differed from the wild-type with respect to the regulation of the enzymes in nitrate assimilation. The inactive nitrate reductase and the nitrite reductase were both found to be constitutively expressed (Muller and Mendel, 1982). These observations are similar to those found in the fungi A . nidulans and N . crassa (Cove, 1979; Tomsett and Garrett, 1981). This result indicates that the tobacco nitrate reductase is autogenously regulated as in fungi (see Section 111,B). Further, it regulates expression of nitrite reductase as well. This conclusion derives from the crucial observation that mutants which affect the structural integrity of the enzyme alter the pattern of regulation of those two enzymes. The molybdenum cofactor mutants of N . plumbaginifolia are less well characterized. The NX 1 mutant is molybdate repairable and will complement the N . crassa nit- 1 mutant in vitro; it therefore appears similar to the cnxA mutant of N . tobacum. The two other molybdenum cofactor mutants of N . plumbaginifolia, NX 21 and NX 24, are not molybdate repairable and are not allelic since somatic hybrids derived from fusing protoplasts have restored nitrate reductase activity (Marton et al., 1982). NX 24 is equivalent to cnxB of N . tobacum (Xuan et al., 1983). Mendel et al. (1982~)studied the regulation of molybdenum cofactor synthesis in N . tobacum. Molybdenum cofactor content was determined by the ability of a cell-free extract of tobacco to restore NADPH-nitrate reductase activity to an extract of N . crassa nit- 1. The addition of nitrate to an amino acidgrown culture of wild-type cells resulted in a concurrent increase in both nitrate reductase and molybdenum cofactor content of the cells. This result was unlike that found by Garrett and Cove (1976) in A . nidulans, where nitrate did not influence molybdenum cofactor content. But as Mendel et al. (1982~)point out, Garrett and Cove measured “free” cofactor, while Mendel et al. assayed total molybdenum cofactor content following heat treatment of cofactor-containing extracts. Four cnxA mutants were examined and all were shown to have a constitutively maintained high molybdenum cofactor content. The niaAl/2 mutants could be divided into two groups depending on molybdenum cofactor content. Group 1 niaA1/2 mutants had a low molybdenum cofactor content, and the levels were not inducible by nitrate. Group 2 niaA1/2 mutants had molybdenum cofactor levels similar to wild-type. Recently, Evola ( 1 983) has isolated a chlorate-resistant variant (clr0) in N . tobacum lacking both nitrate reductase and xanthine dehydrogenase activity. This variant may be allelic to either cnxA or CUB. A possible molybdenum cofactor mutant has been isolated in Datura innoxia by King and Khanna (1980). The variant cell line, N R l , was selected from a haploid cell culture grown in the presence of chlorate. The NR1 mutant was
30
NlGEL S . DUNN-COLEMAN E T AL
unable to grow on nitrate as the sole nitrogen source and lacked NADH-nitrate reductase activity. The mutant did however retain NADH-cytochrome c reductase activity typical of that seen in cnxA mutants of N . tobacum. Recently, molybdenum cofactor mutants also have been isolated in Hyoscyamus muticus by King’s group ( S t r a w et al., 1981; Gebhardt et al., 1982). A total of five clones lacking nitrate reductase activity were isolated from MNN9-mutagenized haploid protoplasts. Unlike previously described methods for isolating nitrate reductase mutants by resistance to chlorate, King’s group let the protoplasts from colonies on enriched media and then plate-tested a sample from each colony on selective media. This procedure also lead to the identification of various amino acidrequiring clones, e.g., tryptophan. All four clones lacked nitrate reductase activity. Xanthine dehydrogenase activity was examined and shown to be absent in three clones: VIC2, I,D12, and XIVE9. The clone I,D12 was also shown to have molybdate-repairable nitrate reductase and xanthine dehydrogenase and is therefore similar to cnxA of N . tobacum. Gebhardt et al. (1982) tentatively concluded that the four clones belonged to at least two different complementation groups. 3. Barley and Pea Nitrate Assimilation Mutants Kleinhofs and Warner’s group (Washington State University) have isolated nitrate reductase structural gene and molybdenum cofactor mutants in both barley (Hordeum vulgare) and pea (Pisum sativum) (Tables VI and VII). The strategy used to isolate the mutants in both plants was similar. Mutations were induced by incubating seeds with the mutagen NaN, (M1 generation). The treated seed was then planted, selfed, and the segregating M2 generation seedlings individually assayed in vivo for nitrate reductase activity (Warner er al., 1977, 1982; Kleinhofs et al., 1978; Warner and Kleinhofs, 1981). Three nitrate reductase-deficient mutants were isolated in pea. Two of the mutants which were shown to be allelic, A317 and A334, retained less than 6%of the wild-type level of nitrate reductase and had normal levels of xanthine dehydrogenase activity. These mutants appear to be nitrate reductase apoprotein mutants and the designation nar-1 was given to them. The third mutant (A300) lacked xanthine dehydrogenase activity and retained only 20% of the wild-type level of nitrate reductase. This mutant was designated as nar-2 and appears to be a molybdenum cofactor mutant. Warner et al. (1982) undertook a genetic analysis of the mutants and showed that nar-1 and nar-2 are nonallelic. F, hybrids between each mutant and the wild-type (Juneau) had intermediate levels of nitrate reductase activity, indicating incomplete dominance of the deficiency. The F, segregation of the mutants indicated that the nitrate reductase deficiency was controlled by a single nuclear gene. Feenstra and Jacobsen (1980) and Feenstra et al. (1982) have also isolated a nitrate reductase mutant in pea. They used an isolation procedure similar to that of Warner and Kleinhofs and also showed the mutant to be chlorate resistant. As in the case of the nar-1 mutants, Feenstra’s mutant (El)
TABLE VI NITRATEREDUCTASE STRUCTURAL GENEMUTANTSI N PLANTS
Plant cell tissue culture
Locus or mutant number
Method of isolation
Comments
References
N . tobacum cell cultures are allodihaploid and Miiller’s group has recently shown that niaA mutants are in fact unlinked double mutants niaAl, niaA2. niaA mutants are strictly auxotrophic on nitrate as the sole nitrogen source NA (equivalent to niaA mutants of N . tobacum) were isolated in haploid protoplast cultures and are likely to be single mutants. NA mutants are strict auxotrophs on nitrate as the sole nitrogen source nar-1 mutants retain 3-10% of wild-type NADH-nitrate reductase activity. These mutants grows on nitrate due to a second, bispecific NAD(P)H-nitrate reductase enzyme Two allelic mutants, A317 and A334, retain less than 6% wild-type level of nitrate reductase activity
Miiller and Mendel (1982); Miiller (1983)
The mutant 305 may be allele of nitA
Nichols and Syrett (1978); Sosa et al. (1978) Braaksma (1982)
Nicotiana tobacum
nid
ENUO mutagenesis or spontaneous resistance to KCIO, in cell suspensions
Nicotiana plumbaginifolia
NA
Either spontaneous, 6OCo irradiation, NEUU mutagenesis in protoplasts, and screening for resistance to KCIO,
Hordeum vulgare
nar-l
NaN, mutagenesis (Ml) of seeds screening M2 progeny for low nitrate reductase activity
Pisum sativum
nar-1
Chlamydomonas reinhardii
nitA
NaN, mutagenesis (Ml) of seeds screening M2 progeny for low nitrate reductase activity UV irradiation, screening for KCIO3 resistance
Arabidopsis thaliana
chl-2 chl-3
aENU and NEU, N-ethyl-N-nitrosourea
EMS mutagenesis (Ml) M2 progenies were screened for KC103
The mutants B21 (chl-2 locus) and B29 (chl-3) locus) retain 30 and lo%, respectively, of the fully induced wild-type level of nitrate reductase
Marton et al. (1982)
Dailey et al. (1982a,b); Warner et al. (1977); Kleinhofs et al. (1983); Kuo et al. (1984) Warner et a / . (1982)
TABLE VII MOLYBDENUM COPACTOR MUTANTSIN PLANTS
Plant species Nicotiana tabacum
w
Nicotiana plumbaginifolia
Hyoscyamus muticus
Datura innoxia
Locus or mutant number
Xanthine Molyb- In vitro dehydate compleNitrate drogen- KC103 repair mention reductase ase ac- resistant/ of de- of nit-1 Method of isolation activity tivity sensitive fect mutant
EMUa mutagenesis of cell suspension and then selection for chiorate resistance EMU mutagenesis CnxB of cell suspension and then selection for chlorate resistance NXl (cnxA) Spontaneous, 6OCo irradiation or NX21 (cnxC) NEU" mutaNX24 (cnxB) genesis of protoplasts then selection for KC103 resistance MNNG mutagenesis I,D 12 of protoplasts, VIC2 then screening for XIVE9 nitrate nonutilizaMA-2 tion Spontaneous reNR 1 sistance to KC103 in cell suspension cnxA
0
0
R
+
+
0
0
R
-
+
0 0 0
0 0 0
R R R
+
+
-
? ?
0 0 0 0
0 0
R R R R
0
7
0 0
R
-
+ -
+ ?
? ? 0
Comments
Reference
Four allelic mutants, Muller and Grafe constituitive (1978); Mendel and NADH-CR, and Muller ( 1979); derepressed MoCo Muller and Mendel synthesis ( 1982) Four allelic mutants, Buchanan and Wray inducible NADH( 1982) CR and MoCo syn- Mendel et al. (1984) thesis NXl similar to cmA. Marton et al. (1982) NXI, NX21, and Mendel et al. (1982a) NX24 are nonallelic mutations, somatic hybrids prototrophic 12D12 and MA-2 are similar to cnxA and NXI. At least two different loci Lacks FMNH2-NR, MVH-NR activity. Has NADH-CR activity
Gebhardt er al. (1 982) Strauss et al. (1981) Lazar et al. (1983) Fankhauser et al. ( 1984) King and Khanna ( 1980)
Physcomirrella paiens
CitX
Arabidopsis thaliana
Cn*
rgn Pisum sativum
Hordeum vulgare
nar-2
Cn*
W
nar-2
nar-3
Chlamydomonas reinhardii
nit-C
NTG mutagenesis of spores and then selection for KC103 resistance NTG mutagenesis (Ml) of seeds, then screening M2-segregating population for NaR activity NaN3 mutagenesis (Ml), screening M2 progeny for NaR activity NaN3 mutagenesis (MI), screening M2 progeny for NaR activity NaN3 mutagenesis (Ml), screening M2 progeny for NaR activity
EMS mutagenesis screening M2 progeny for chlorate resistance UV irradiation
aNEU and EMU, N-ethyl-N-nitrosourea.
0
0
R
1 of 3
?
Three cornplementation groups
Ashton and Cove (1977); Dunlop et al. (1982)
10%
10-
R
+
?
1%
15% 10%
R
-
?
Braaksma and Feenstra (1982)
20%
0
?
?
?
Warner et al. (1982)
1%
0
R
?
?
Bright ef al. (1983)
13%
0
S
?
?
10%
0
R
?
?
0
0
R
?
?
Formerly A ~ 3 4
Warner and Kleinhofs (1981); Kuo et al. (1983; Dailey et al. (1982); Kleinhofs el al. (1983); Tokarev and Shumny (1977) Formerly Xno 18, not Narayanan et al. allelic to nar-2 (1984)
Lacks MVH-NR activity, has NADHCR
Nichols and Syrett (1978)
34
NIGEL S. DUNN-COLEMAN ET AL
retained only 5% of the wild-type level of nitrate reductase and was inherited as a monogenic trait. Feenstra et al. (1982) also studied the effect of nitrate on symbiotic nitrogen fixation in both wild-type (c.v. Rhondo) and E l . Evidently, supply of nitrate to leguminous plants actively fixing nitrogen decreases the rate of N, fixation. Nitrogen fixation, as judged by acetylene reduction, was only inhibited 19% in the mutant E l compared with 47% in the wild-type when grown in the presence of nitrate, indicating that nitrate has to be reduced by the plant before its inhibitory effect on nitrogen fixation can take place. The nitrate reductase mutants isolated in barley by Kleinhofs and Warner’s group appear very similar to those they isolated in pea. Two unlinked loci have been identified. The nar-1 locus appears to be the structural gene for the nitrate reductase apoprotein, and the nar-2 locus, a molybdenum cofactor gene. Genetic analysis of both loci indicates that they are each inherited as a single nuclear gene. Mutants at nar-1 locus are codominant and nar-2 mutants are recessive. Tokarev and Shumny (1977) have also isolated three barley nitrate reductase mutants by EMS mutagenesis and chlorate resistance screening. The mutant Xno 29 is an allele of nar- 1. The other two mutants, Xno 18 and Xno 19, are allelic to one another but are not allelic to either nar- I or nar-2. Kleinhofs et al. proposed that a new locus, nar-3, is defined by the Xno 18 and Xno 19 mutants. These nar-3 mutants lack xanthine dehydrogenase activity and therefore appear to be molybdenum cofactor mutants (Somers et ul., 1983). Narayanan et al. (1983) have recently examined the nature of cytochrome c reductase (CR) in wild-type and nitrate reductase mutants of barley. They found that nur mutants with significant CR activity fall into two classes, one had a 8 S enzyme species, the other had only the 4 S CR form. Barley nitrate reductase is apparently a homodimeric protein with 100 kd subunits. Although the barley nar-1 mutants retain less than 10% of both the in vivo and in vitro nitrate reductase activity of the control, they are capable of substantial nitrate reduction and produce as much dry weight and protein content as control plants. One explanation for these results is that there is a bispecific NAD(P)H-nitrate reductase activity in barley. Dailey et al. (1982b) have looked for such an activity in nar-la mutants. Using Affi-Gel blue affinity chromatography which does not bind wild-type NADH-nitrate reductase, they purified a bispecific NAD(P)H-nitrate reductase from the mutant nar- la. This bispecific nitrate reductase probably accounts for the residual nitrate reductase activity seen in nar-1 mutants, and consequently, their ability to utilize nitrate as a nitrogen source. Kleinhofs et al. (1983) have attempted to select for mutants lacking the NAD(P)H-nitrate reductase. They mutagenized seed from the nar- l a mutant and screened for M2 seedlings lacking nitrate reductase activity. Three mutants were isolated which grew poorly on nitrate and were deficient in xanthine dehydrogenase activity also. They are likely to be molybdenum cofactor mutants. A similar mutant has been isolated by Bright et al. (1983), using NaN, mutagenesis of barley (cv. Mario Mink). This mutant (R9401) is the result of a single recessive mutation and has 1% of the wild-type in vitro nitrate
NITRATE ASSIMILATION IN EUKARYOTIC CELLS
35
reductase activity. R9401 is chlorate resistant, unlike the mutants isolated by Warner and Kleinhofs, and lacks xanthine dehydrogenase activity. It is unable to grow on nitrate as sole nitrogen source either in soil or sterile culture and appears to be a conditional lethal c m type mutant. At present, it is not known if this mutant or the mutants isolated by Kleinhofs et al. define a new locus or a more extreme allele of the nar-2 locus. In both barley and pea, the nar-1 and nar-2 mutants had nitrate reductase specific activities inducible by nitrate at higher than maximal levels in wild-type. Notton et al. (1983) have recently shown that the mutant nar-la contains both the nitrate reductase apoprotein and molybdenum cofactor. The nitrate reductase deficiency in these mutants appears to be due to an inability to maintain nitrate reductase in an active aggregated form. Similar results have been reported by Asan et al. (1983). 4. Arabidopsis thaliana
Arabidopsis thulianu has also been used as a model system to study nitrate assimilation. The advantages of this plant include a short generation time, haploid chromosome number of 5 , self-fertility , and homozygous “wild-type” material (Braaksma, 1982). Chlorate-resistant mutants were generated by treating seeds with either the mutagen EMS or NTG (MI generation). The M1 generation plants were selfed and M2 seedlings grown in the presence of chlorate. Plants which showed resistance or little damage from chlorate were characterized further (Braaksma and Feenstra, 1979, 1982; Braaksma, 1982). A total of 61 chlorate-resistant mutants were isolated, 10 of which had lowered nitrate reductase activity. The remaining 51 mutants were allelic (chl-1 locus) and were impaired in the uptake of chlorate. The mutants which were impaired in their nitrate reductase activity were divided into 7 different complementation groups. Two mutants, B25 and B73, also had low xanthine dehydrogenase activity, and the lesion in B73 was partially repaired by exogenous molybdate. The mutant B73 therefore appears similar to cmA of N . tobacum, NX 1 of N . plumbaginifolia, and cnxE of A . nidulans. A second mutant B25 (rgn locus), which had very low levels of both the NADH-nitrate reductase and FMNH,-nitrate reductase activity, also appears to be defective in molybdenum cofactor biosynthesis. None of the nitrate reductase mutants had constitutive expression of nitrite reductase. Two other chlorate-resistant complementation groups, chl-2 and chl-3, had normal levels of xanthine dehydrogenase activity but lowered levels of nitrate reductase activity. Either or both of these loci could encode the nitrate reductase apoprotein. The functions of the other chlorate-resistant mutants are unknown. 5. Soybean Recently, soybean mutants lacking constitutive nitrate reductase activity have been isolated (Nelson et al., 1983; Ryan et al., 1983). Approximately 12,000 seedlings isolated from mutagenized seed were screened for chlorate resistance
36
NIGEL S. DUNN-COLEMAN ET AL.
and low nitrate reductase activity. Three mutants were isolated (NR-2, NR-3, and NR-4) and shown to be allelic. Nitrate reductase activities were determined in seedlings grown with either nitrate or urea as nitrogen source. When the mutants were grown on nitrate, they had approximately 50% of the activity found in nitrate-grown wild-type seedlings. However, urea-grown mutants had no detectable nitrate reductase activity. Nelson et al. (1983) concluded that the nitrate reductase activity observed in leaves of nitrate-grown soybean plants consists of both inducible and constitutively expressed activities. They also observed that the mutants lacked an associated NO,,, evolution normally found in young leaves of wild-type plants.
6. Physcomitrella patens Ashton and Cove (1977) isolated three chlorate-resistant mutants in the moss Physcomitrella patens, designated nat- 1, nat-2, and nat-3. All three mutants grew poorly when either nitrate or nitrite was the sole nitrogen source. Unlike nat-3, the nat-1 and nat-2 mutants grew like the wild-type on uric acid or hypoxanthine as a nitrogen source. However, as Ashton and Cove pointed out, P. patens utilizes purines very poorly. More recently, Cove’s group has isolated 34 more nitrate nonutilizing mutants in P. patens (Dunlop et al., 1982; Dunlop, 1982) of which 27 have been partially characterized. Eight of these show leaky growth while 19 do not grow at all on nitrate. Of these 19, 12 have been characterized as putative cnx-type mutants since they lack both nitrate reductase and xanthine dehydrogenase activity. Further, all 12 have constitutive expression of nitrite reductase activity. Somatic hybrids produced by protoplast fusion have so far identified at least three complementation groups among this dozen cmtype mutants. One appears molybdate repairable. No nitrate reductase structural gene mutants have been isolated and Dunlop et al. (1982) speculated that the nitrate reductase gene may be duplicated in P. patens. Their finding that the cnxtype mutants demonstrate constitutive expression of nitrite reductase is consistent with the findings of Muller and Mendel (1982) in N . tobacum, Tomsett and Garrett (1981) in N . crassa, and Cove (reviewed in 1979) in A. nidulans. That is, mutations which affect the structural integrity of the nitrate reductase enzyme, such as cm or apoprotein mutants, may produce nitrate reductase and nitrite reductase constitutively (see Section II,B above). 7. Algae Nitrate assimilation mutants have also been isolated in green algae. Huskey et al. (1979) have identified three genes, nit-A, nit-B, and nit-C, in Volvox carteri which are affected in nitrate assimilation. The nit-A and nit-C mutants lack both NADH-nitrate reductase and NADH-cytochrome c reductase activities and may be defective in the formation of nitrate reductase (either the apoprotein or molybdenum cofactor) or in the regulation of nitrate reductase. The nit-B3 mutant,
37
NITRATE ASSIMILATION IN EUKARYOTIC CELLS
unlike nit-A or nit-C types, is not resistant to chlorate and retains some nitrate reductase activity. Levels of nitrate reductase activity in nit-B3 are a function of the nitrate concentration in the media, suggesting that the nit-B gene may play a role in nitrate uptake. Nichols and Syrett (1978) and Nichols et al. (1978) UV irridated Chlamydomonas rienhardii cells and selected for chlorate resistance. Three loci were identified, mutations in which resulted in failure to utilize nitrate as the sole nitrogen source. One, nit-C, when mutated was also unable to use hypoxanthine as a nitrogen source and thus appears to be a molybdenum cofactor mutant. The nit-A gene mutant lacked both NADH-nitrate reductase and NADH-cytochrome c reductase activities but retained BVH-nitrate reductase activity. It is probably a mutant in the nitrate reductase apoprotein structural gene. The nit-B mutant had reduced transport of nitrate. Nichols et al. (1978) concluded that nit-B could be a regulatory locus for nitrate assimilation. Similar mutants to those reported by Nichols and Syrett (1978) have been described by Sosa et al. (1978) and biochemically characterized by Fernandez and Cardenas TABLE VIII POSSIBLECANDIDATES FOR NITRATE UPTAKEMUTANTS Organism
Mutant/locus/cell line
Characteristics
Aspergillus nidulans
crnA
Volvox carteri
nitB
Nicotiana tobacum
0 41
Arabidopsis thaliana
chl- I
KCIO3 resistant. Has nitrate reductase and nitrite reductase activity KCIO, resistant. Level of nitrate reductase activity is related to the external nitrate concentration KCIO, resistant. Initially unable to grow normally on nitrate, eventually both nitrate reductase and nitrite reductase levels higher than the w i Id-type KC103 resistant. Reduced uptake of chlorate. Normal levels of nitrate reductase and nitrite reductdse
Reference Tomsett and Cove (1979); Brownlee and Arst (1983) Huskey et a / . (1979)
Qureshi et a / . (1982)
Braaksma (1982)
38
NlGEL S. DUNN-COLEMAN ET AL
(1981a,b). In addition, Sosa et al. (1978) described a fourth class of mutants having normal levels of nitrate reductase enzyme activities but unable to grow on nitrate. These could possibly be nitrate transport mutants (see below).
8. Transport Mutants Several possible candidates for nitrate transport mutants have been isolated in plants and fungi (see Table VIII). They are all typified by their resistance to chlorate and retention of nitrate reductase activity. The chlorate-resistant cell line (041) of N . tobacum (Quershi et al., 1982) exhibited a lag of 15 days in cell culture before exponential growth on nitrate was achieved. Maximal levels of both nitrate reductase and nitrite reductase were found after 34 days, whereas wild-type cells exhibited expotential growth and maximal enzyme levels after only 4 days. The results in N . tobacum and also in Volvox carteri (Huskey et al., 1979) indicate that diffusion of nitrate is sufficient to enable survival of nitrate transport mutant cells. Recently, Brownlee and Arst ( 1983) have biochemically characterized nitrate uptake in A. niduluns. The crnAl mutation reduced nitrate uptake dramatically in conidiophores and young mycelia but did not influence uptake in older mycelia. The crnA locus is part of the niaD niiA nitrate assimilation gene cluster and is controlled by the areA gene product but not the nirA gene product. Differential control suggests that the crnA niiA niaD gene cluster is not expressed as a tricistronic messenger. B . CONCLUSIONS Two important classes of mutants have yet to be isolated or well characterized in plants. First, no nitrite reductase structural mutants have been isolated. Such mutants have been isolated only in fungi, based upon their failure to grow on either nitrate or nitrite as a nitrogen source. These mutants are not chlorate resistant (because they still have nitrate reductase activity) but can be recognized by replica plating from a reduced nitrogen source such as ammonium onto either nitrate or nitrite. Replica plating is not possible yet in plant tissue culture and this may explain why such mutants have not been isolated. One approach to isolating nitrite reductase mutants in plants would be to screen plants for high in vivo accumulation of nitrite, but high concentrations of nitrite may be very toxic and such plants may die at an early stage of growth. An alternative approach deserving consideration is the protocol of King et nl. (1982). They used a mutant screening procedure with haploid mesophyll protoplasts from H . muticus. Protoplasts surviving mutagenesis were allowed to proliferate in complete medium and then were transferred as clones to minimal medium. Clones which failed to grow on minimal medium were characterized further. A total of 14 clones with specific stable defects were identified.
NITRATE ASSIMILATION IN EUKARYOTIC CELLS
39
Although regulatory mutants affecting the enzymes of nitrate assimilation have been found in several fungi (e.g., the nit-2 and nit-415 loci in N . crassa), no such mutants have been isolated in higher plants. The work of Filner (1966), Muller and Mendel (1982), and others has shown that the nitrate assimilation pathway is both inducible by nitrate and repressible by reduced nitrogen. Also, protein synthesis inhibitors such as cycloheximide prevent an increase in nitrate reductase activity. These results indicate that the nitrate assimilation pathway is under strict regulation, probably at the level of transcription. Tomsett and Garrett (1980) isolated 67 nitrate assimilation mutants in N . crassa, either on the basis of chlorate resistance or failure to grow on nitrate as the sole nitrogen source. Six mutants were isolated at either the nit-2 or nit-415 regulatory loci. Miiller and Mendel (1982) have, to date, isolated 40 nitrate reductase mutants in N . tobacum. However, the mutants are of only two types: niaA apoprotein mutants (36 mutants) or cnxA (4 mutants). No putative regulatory mutants were isolated. These results may reflect problems in isolating recessive mutants in amphihaploid cell culture. Maliga’s group (Marton et al., 1982) have identified four different kinds of mutants (1 nia and 3 cnx-type mutants) in haploid N . plumbaginifolia protoplasts in 9 characterized clones. The use of haploid material as a source material appears to increase the probability of identifying a greater range of mutants. So far, the molybdenum cofactor mutants and apoprotein mutants appear to be analogous to similar mutants found in fungi. It can, therefore, only be a matter of time before the existence of regulatory mutants affecting nitrate assimilation in higher plants is determined.
V. Cell and Molecular Biological Advances in Nitrate Assimilation A. SOMATICHYBRIDIZATION STUDIESIN NITRATEASSIMILATION Nason et al. (1970) initially demonstrated the in vitro complementation of nitrate nonutilizing mutants by the restoration of NADPH-nitrate reductase activity upon cohomogenization of mycelia from a nit- 1 (molybdenum cofactor) mutant and a nit-3 (apoprotein) mutant of N . crassa (Section 11,F). This in vitro complementation was then achieved with higher plants by Mendel and Miiller (1979) who in vitro reconstituted NADH-nitrate reductase activity of N . tobacum using mutant cell lines of the niaA (apoprotein) and cnxA (molybdenum cofactor) mutants. These successes led to in vivo complementation by somatic hybridization in higher plants. Glimelius et al. (1978) selected for somatic hybrids using PEG-mediated fusion of protoplasts of cnxA and niaA mutants of N . tobacum. The hybrid cells had restored nitrate reductase activity, the result of genetic complementation, and were able to utilize nitrate as the sole nitrogen source. However, plating niaAlcnxA fused protoplasts directly onto nitrate was
40
NIGEL S. DUNN-COLEMAN ET AL
unsuccessful in isolating somatic hybrids. The fused protoplasts had to be first plated onto nonselective medium (containing amino acids) for several days until cell walls had regenerated and small clusters of cells had developed. These clusters of cells (mini-calli) were then plated onto nitrate-containing media and the proliferating calli were purified. Shoots could be regenerated from these hybrid cell lines, unlike the parental cell lines which were unable to regenerate shoots. Marton et al. (1982) isolated somatic hybrids between nine nitrate reductase mutants of N . plumbaginifolia. The five apoprotein-defective (NA) cell lines were found to be noncomplementing and are therefore probably allelic. Somatic hybrids were obtained from the fusion of NA and NX (molybdenum cofactor) mutants and they showed restoration of nitrate reductase activity. These results were therefore identical to those of Glimelius et al. (1978) for N . tobacum. Marton et al. (1982) also found that the four NX cell lines belonged to three complementation groups (NXl and NX9, NX21, NX24). Somatic hybrids which had restored nitrate reductase activity also had regained the ability to regenerate shoots. Dunlop (1982) has also used somatic hybridization to isolate diploid hybrids for complementation studies in the moss P hyscomitrella patens and has identified at least three complementation groups among their putative cnx mutants (see Section IV,A,6). Lazar et al. (1983) have used protoplast fusion to isolated intergeneric somatic hybrids between a Hyoscyamus muticus nitrate reductase variant (MA-2) and the niaA( 115) and cnxA(68) mutants of N . tobacum. Only the MA-2/niaA( 115) combination yielded prototrophic intergeneric hybrids capable of growth on nitrate. Lazar et al. (1983) therefore concluded that the MA-2 variant had a defective molybdenum cofactor. When N . tobacum is grown in the absence of nitrate, nitrate reductase activities are only 20-30% of the fully induced, nitrategrown level. H . muticus however has undetectabie nitrate reductase activities when grown in the absence of nitrate. The intergeneric hybrids Lazar et al. (1983) isolated displayed the H . muticus pattern of nitrate reductase activity regulation. Gupta et al. (1983) have reported the correction of the cnxA(68) defect in protoplast derived-intergeneric gene transfers between the cnxA(68) mutant and either Datura innoxia or Physalis minima. Wild-type niesophyll protoplasts of D . innoxia and P . minima were X-ray irradiated, resulting in their becoming mitotically inactive and consequently unable to grow on nitrate. These protoplasts were then used with cnx(68) protoplasts and initially grown under nonselective conditions for several weeks and then plated onto nitrate media. A total of 45 nitrate-utilizing colonies were isolated from the nitrate medium. Since cnxA(68) is a nonrevertable mutant, Gupta et al. (1983) concluded that these colonies must have arisen from heteroplasmic fusions, and the frequency of successful transformation of the cmA(68) mutant was approximately 3%. The nitrate reductase activities of the cnxA(68): P . minima or D. innoxia transformants were 3-5% of the wild-type level in P. minima or D. innoxia. Gupta et al.
NITRATE ASSIMILATION IN EUKARYOTIC CELLS
41
(1983) suggested that inefficient expression of the D. innoxia or P . minima genes in the cnxA(68) cell line might account for the low levels of nitrate reductase activity. The results of Gupta et al. (1983) and Lazar et al. (1983) indicate that intergeneric complementation of nitrate reductase mutants is possible. It is a relatively straightforward technique to identify the hybrids of transformants by selecting for growth on nitrate after allowing such cells to undergo several rounds of cell division on non-selective medium. This step probably allows the hybrid or transformed cells to reach a sufficient population density so that there are enough viable cells to maintain growth when plated on selective media. Pental et al. (1982) used the N . tobacum mutant niaA( 130) to demonstrate the feasibility of identifying rare protophic clones among a background of nitrate nonutilizing niaA( 130) mutants. In the reconstruction experiments they conducted, a few wild-type protoplast-derived colonies were mixed with 4-5 X lo4 niaA( 130) colonies. Depending on the plant hormone composition of the medium used, nearly a 100% throughput of wild-type colonies could be identified. This work showed that it may be possible to select for plant transformation via functional complementation of mutant protoplasts using plant nitrate reductase mutants such as niaA(130). One can envisage using wild-type genomic cosmid libraries introduced into niaA( 130) protoplasts, via liposomes for example, and selecting for transformants capable of utilizing nitrate. This would provide one method of cloning plant nitrate assimilation genes. B. CLONING OF A MOLYBDENUM COFACTOR GENEFROM E . coli Taylor et al. (1982) have reported cloning the chlA molybdenum cofactor gene of E . coli. The presence of molybdenum cofactor in wild-type and chlorateresistant (chl)mutants of E . coli was determined by in vitro complementation of the nitrate reductase mutants of N . tobacum [cnxA(68)] and H . muticus (MA-2). Only the chlA mutant failed to complement either of the two plant nitrate reductase mutants. Taylor el al. (1982) and Kleinhofs et al. (1983) initially cloned chlA from a wild-type cosmid library which had been used to transform a chlA mutant. The cosmid clone (pJT1) was subcloned by HaeII deletion to a 10.8 kb plasmid designated pJ 13. Interestingly, this pJ13 clone would only transform the chlA mutant SA493 to wild phenotype; chlA mutant JP382 was not transformed by pJ13. By subcloning aHindIII fragment from pJTl into pBR322, Kleinhofs et al. were able to transform chlA SA493 to wild-type. Kleinhofset al. concluded that the chlA locus could be divided into two components. They proposed that the chlA locus should be changed to chlM (represented by the SA493 mutant) and chlN (represented by the JP382 mutant). Kleinhofs et al. also showed that the chlM gene encodes a polypeptide of 15,000 MW. Cell-free extracts of the chlA transformants were able to restore nitrate reductase activity to extracts of
42
NIGEL S. DUNN-COLEMAN ET AL
either N . tobacum cnxA(68) or H. muticus MA-2. With the development of a selectable plant transformation system for dicotyledonous plants (see Section V,D, below), it should be possible to insert the cloned chlA gene onto an appropriate vector to see if the chlA gene can complement mutants cnxA(68) and MA-2 in vivo. C. FUNGALTRANSFORMATION SYSTEMS
To date, three N . crassa genes and one A. nidulans gene have been cloned based upon their ability to complement equivalent E. coli mutants. In all cases, genomic libraries of these two fungi were constructed in the plasmid pBR322. Vapnek et al. (1977) reported the cloning of the qa-2 gene which encodes the catabolic quinate dehydrogenase (5-dehydroquinate hydro-lyase), due to its ability to complement an aroD E . coli mutant which lacked this activity. Schechtman and Yanofsky (1983) and Keesey and De Moss (1983) reported cloning the trp-1 gene of N . crassa via complementation of a trpC mutant of E. coli. Buxton and Radford (1983) were successful in cloning the orotidine 5'-phosphate carboxylase (pyr-4) gene of N . crassa by its complementation of the pyr F gene of E . coli. In A . nidulans, Kinghorn and Hawkins (1982) have cloned the synthetic dehydrogenase function of the arom cluster of A. nidulans via complementation of a aroD mutant of E. coli. All four of these cloned genes are apparently being expressed in E. coli due to fortuitous expression of an accompanying fungal promoter rather than expression via either of the two bacterial promoters found on the pBR322 plasmid. Only the qa-2 gene from the qa cluster of 4 genes is expressed in E . coli and it seems exceedingly likely that many genes of interest, particularly those of nitrate assimilation such as nitrate reductase and nitrite reductase, will not be able to be cloned via complementation of an equivalent E . coli mutant. A preliminary report citing evidence for the expression of the N . crassa nit-3 gene in E . coli (Smarrelli and Garrett, 1982) was erroneous and was traced to a Pseudomonas maltophilia contaminant. Case et al. (1979) reported the development of a cloning system for N . crassa based upon cotransformation of sequences upon complementation of a qa-2 mutant with the cloned qa-2 gene. Subsequent to this, Schweizer et al. (1981) reported using this transformation system to clone the remaining three genes of the 4a cluster (4a-1, 4a-3, and qa-4). N . crassa is believed to have no nuclear plasmids equivalent to the yeast 2 p DNA, which has been extensively used in yeast transformation experiments. The 2 p DNA plasmid is capable of autonomous replication and chimeric plasmids have been developed as yeast-E. coli shuttle vectors for cloning genes into either of the two. Selecting for stable N . crassa transformants using a genomic library in a plasmid which cannot replicate autonomously results in the selection of transformants with integrated vector DNA. This integration of the vector presents novel problems in reisolating the
NITRATE ASSIMILATION IN EUKARYOTIC CELLS
43
gene of interest from the transformant. Recently, Stohl and Lambowitz (1983) reported the construction of a shuttle vector of N . crassa using one of the three mitochondria1plasmids of N . crassa. The report by Hughes et al. (1983) describing a chimeric plasmid which can replicate in E . coli and N . crassa was found to be erroneous. Development of a shuttle vector would greatly facilitate the cloning of N . crassa nuclear genes such as those concerned with nitrate assimilation. D. PLANTTRANSFORMATION Many species of dicotyledous plants are susceptible to crown gall disease, which is the result of infection of wound sites by the bacterium Agrobacterium tumefaciens. Recently, it has been shown that bacterial plasmids called Ti (tumor-inducing) are responsible for crown gall development and that a portion of the Ti plasmid, T-DNA, integrates into the plant nuclear genome. The genetic transformation of the plant cell by T-DNA results in the rapid proliferation of these transformed cells to form tumors (see Schell and Van Montagu, 1983, for a review). A consequence of this genetic transformation is the production by the plant cell of opines which are used by free-living A . tumefaciens. The genes coding for octopine and nopaline synthase are encoded in T-DNA. To overcome barriers to the expression of foreign genes in plants, a number of laboratories have constructed various chimeric genes consisting of promoter sequences derived from the nopaline synthase gene and various coding sequences such as the chloramphenicol acetyltransferase gene of pBR325, the niomycin phosphotransferase gene from Tn5, or the methothrexate resistance-conferring dihydrofolate reductase gene of the plasmid R67. These chimeric constructions have very recently been shown to be expressed in plant cells (Schell and Van Montagu, 1983; Bevan et al., 1983; Farley et al., 1983). With the development of a selectable system of plant transformation this should greatly facilitate the study of cloned nitrate assimilation genes reintroduced into plants. E. CLONING NITRATEASSIMILATION GENES-CONCLUDING
REMARKS
Several possible means exist to clone nitrate assimilation genes. First, the chlM molybdenum cofactor gene has already been cloned (Taylor et al., 1982), and it seems likely that additional E . coli genes for nitrate assimilation will be cloned shortly. In view of the demonstrated conserved nature of the molybdenum cofactor, it seems likely that with the appropriate regulatory sequences, molybdenum cofactor genes will be expressed in fungal and plant cells, thus providing a selectable means for plant transformation due to genetic complementation. With the development of a transformation system for N . crussa and the attendant use of shuttle vectors, it seems likely that many nitrate assimilation genes in N . crassa will be rapidly cloned. An extensive range of auxotrophic mutants for
44
NIGEL S . DUNN-COLEMAN ET AL
nitrate assimilation genes is readily available. A similar case can be made for cloning A. nidulans nitrate assimilation genes. Tilburn et al. (1983) have reported the development of transformation of A. nidulans using the previously cloned Aspergillus acetamidase gene (see below), but transformation efficiencies were low. Hynes et al. (1983) have recently reported cloning the acetamidase structural gene (amdS) from A. nidulans. Using in vitro translation, Hynes et al. showed that amdS368 deletion mutant produced no amdS polypeptide, and it seemed likely that this mutant would produce no amdS mRNA. cDNA libraries were constructed from poly(A) mRNA isolated from the amdS368 deletion mutant and from a strain specially constructed to produce high levels of acetamidase. These libraries were used as probes to screen an A. nidulans library constructed in h phage (Charon 30). Three plaques showing strong hybridization to the cDNA from the enhanced amdS strain but not the amdS deletion mutant were isolated. Hynes et al. went on to map the amdS clones and showed that the controlling region of the gene is located at its 5’ end. Similar deletion mutants exist for the nitrate reductase and nitrite reductase genes (niaD and niiA) in A. nidulans. Tomsett and Cove undertook an elegant and extensive analysis of the niiA niaD gene region and deletion mapped closely linked niiA and niaD genes. An approach similar to that used to clone the amdS gene could be used to clone both nitrate reductase and nitrite reductase using these deletion mutants characterized by Tomsett and Cove (1 979). It is possible that cloned fungal nitrate assimilation genes may show enough DNA homology to plant genes to allow their use as probes for specific gene identification and isolation. A second approach would be to purify nitrate assimilation enzymes to homogeneity in quantities sufficient for microsequencing of the protein to be carried out. Knowledge of the amino acid sequence would allow the synthesis of oligonucleotide cDNA probes, assuming the protein sequenced does not have too many amino acids which have degenerate codon usage. Such cDNA probes could be used to screen genomic libraries of plants either in phage or cosmid constructions. This synthetic oligonucleotide approach has been successful in isolating the am gene (encoding NADP-GDH) from N . crassa (Kinnaird and Fincham, 1982). Alternatively, plant cosmid libraries could be shotgunned into plant protoplasts via liposomes with the hope that a gene of interest could be identified via complementation of an appropriate auxotrophic mutant. The development of transformation systems for Volvox and Chlamydomonas would allow the cloning of nitrate assimilation genes via complementation of existing mutants. Over the next few years, it seems exceedingly likely that the nitrate assimilation genes will be cloned and these achievements should provide great insights into the study of nitrate assimilation and its regulation.
NITRATE ASSIMILATION IN EUKARYOTIC CELLS
45
REFERENCES Alberg, B. (1947). Kungf. Luntbrukhogskolans Annul. 45, 37-107. Amy, N. K . , and Garrett, R . H. (1979). Anal. Biochem. 95, 97-107. Amy, N. K., and Garrett, R. H. (1980). J . Bacteriol. 144, 232-237. Amy, N. K., Garrett, R. H., and Anderson, B. M. (1977). Biochim. Biophys. Actu 480, 83-95. Arst, H. N., Jr., and Cove, D. J. (1969). J . Bacteriof. 98, 1284-1293. Arst, H. N., Jr., MacDonald, D. W., and Cove, D. J. (1970). Mof. Gen. Genet. 108, 129-145. Ant, H. N., Jr., Rand, K. N., and Bailey, C. R. (1979). Mol. Gen. Genet. 174, 89-100. Arst, H. N., Jr., Tollervey, D. W., andSealy-Lewis, H . M. (1982a). J . Gen. Microbiol. 128, 10831093. Ant, H. N., Jr., Brownlee, A. G . , and Cousen, S. A. (1982b). Curr. Genet. 6, 245-257. Asan, S., Sawhney, S . K., Kumar, S., Mohanti, P., Sinha, S . K., and Niak, M. S. (1983). Plant Sci. Lett. 30, 17-23. Ashton, N. W., and Cove, D. J. (1977). Mof. Gen. Genet. 154, 87-95. Bahns, M . , and Garrett, R. H. (1980). J . Biol. Chem. 255, 690-693. Beevers, L., and Hageman, R. H. (1980). In “The Biochemistry of Plants” (B. J. Miflin, ed.), Vol. 5 , pp. 115- 168. Academic Press, New York. Bevan, M. W., Flavell, R. B., and Chilton, M.-D (1983). Nature (London) 304, 184-187. Birkett, J. A,, and Rowlands, R. T. (1981). J . Gen. Microsc. 123, 281-285. Braaksma, F. J. (1982). Ph.D. thesis, University of Groningen, Haven. Braaksma, F. J., and Feenstra, W. J . (1979). Mufaf. Res. 19, 175-185. Braaksma, F. J., and Feenstra, W. J. (1982). Physiof. Plant. 54, 351-360. Bright, S. W. J., Norburg, P. B., Franklin, J., Kirk, D. W., and Wray, J. L. (1983). Mol. Gen. Genet. 189, 240-243. Brownlee, A. G., and Arst, H. N., Jr. (1983). J . Bacteriol. 155, 1138-1146. Buchanan, R. J., and Wray, J. L. (1982). Mol. Gen. Genet. 188, 228-234. Bums, R. C., and Hardy, R. W. F. (1975). “Nitrogen Fixation in Bacteria and Higher Plants.” Springer-Verlag, Berlin and New York. Buxton, F. P., and Radford, A. (1983). Mof. Gen. Genet. 190, 403-405. Campbell, W. H. (1978). Z . P’anzenphysiol. 88, 357-361. Campbell, W. H., and Smarrelli, J. Jr. (1978). Plant Physiof. 61, 611-616. Campbell, W. H., and Smarrelli, J. Jr. (1983). In “Biochemical Basis of Plant Breeding” (C. A. Neyra, ed.) CRC Press, Boca Raton, Florida, in press. Case, M. E., Schweizer, M., Kushner, S. R., and Giles, N. H. (1979). Proc. Nutf. Acad. Sci. U.S.A. 76, 5259-5263. Coddington, A. (1976). Mof. Gen. Genet. 145, 195-206. Cove, D. J . (1966). Biochim. Biophys. Acta 113, 51-56. Cove, D. J. (1969). Nature (London) 224, 272-273. Cove, D. J. (1970). Proc. R . SOC. London Ser. B 176, 267-275. Cove, D. J . (1976a). Mol. Gen. Genet. 146, 147-159. Cove, D. J. (1976b). Heredity 36, 191-193. Cove, D. J. (1979). Biol. Rev. 54, 291-227. Cove, D. J., and Pateman, J. A. (1963). Nature (London) 198, 262-268. Cove, D. J . , and Paternan, J. A. (1969). J . Bacreriol. 97, 1374-1378. Dailey, F. A., Kuo, T., and Warner, R. L. (1982a). Plant Physiol. 69, 1196-1 199. Dailey, F. A., Warner, R. L., Somers, D. A., and Kleinhofs, A. (1982b). Plantfhysiof. 69, 12001204. Dantzig, A. H., Zurowski, W. K., Ball, T. M., and Nason, A. (1978). J . Bacteriol. 133, 671-679.
46
NIGEL S . DUNN-COLEMAN ET AL
Davila, G. F., Sanchez, F., Pacacios, R., and Mora, J. (1978). J . Bacteriol. 134, 693-698. Davila, G. F., Lara, M., Guzman, J., and Mora, J. (1980). Biochem. Biophys. Res. Commun. 92, 134-140. De la Rosa, M. A , , Vega, J. M., and Zumft, W. G. (1981). J . B i d . Chem. 256, 5814-5819. Downey, R. J., and Steiner, F. X . (1979). J . Bacteriol. 137, 105-1 14. Dunlop, M. K. (1982). Ph.D. dissertation, University of Leeds. Dunlop, M. K., Cove, D. J., Miklin, B. J., and Lia, P. J. (1982). Proc. Int. Symp. Nitrate Assirnilat., Gatersleben (Abstr.). Dunn-Coleman, N. S . , and Garrett, R. H. (1980). Mol. Gen. Genet. 179, 25-32. Dunn-Coleman, N. S., and Pateman, J. A. (1977). Mol. Gen. Genet. 152, 285-293. Dunn-Coleman, N. S . , and Pateman, J. A. (1979). Mol. Gen. Genet. 171, 69-73. Dunn-Coleman, N. S., Tomsett, A. B., and Garrett, R. H. (1979). J . Bacteriof. 139, 697-700. Dunn-Coleman, N. S . , Tomsett, A. B., and Garrett, R. H. (1981a). Mol. Gen. Genet. 182, 234239. Dunn-Coleman, N. S., Robey, E. A,, Tomsett, A. B., and Garrett, R. H. (1981b). Mol. Cell. Biol. 1, 158-164. Dunn-Coleman, N . S . , Nassiff, M. D., and Garrett, R. H. (1984). Curr. Genet., in press. Evola, S . V. (1983). Mol. Gen. Genet. 189, 447-454. Fankhauser, H., Bucher, F., and King, P. J. (1984). Planta 160, 415-421. Farley, R. T., Rogers, S. G . , Horsch, R. B., Sanders, P. R., Flick, J. S., Adams, S. P., Bittner, M. L., Brand, L. A., Fink, C. L., Fry, J. S., Galluppi, G. R., Goldberg, S. B., Hoffman, N. L., and Woo, S. G. (1983). Proc. Natl. Acad. Sci. U.S.A. 80, 4803-4807. Feenstra, W. J., and Jacobsen, E. (1980). Theor. Appl. Genet. 58, 39-42. Feenstra, W . J., Jacobsen, E., Van-Swaay, A. C. P. M., and DeVisser, A. J . C. (1982). Z . Pflanzenphysiol. 105, 471-474. Femandez, E., and Cardenas, J. (1981a). Biochim. Biophys. Acta 657, 1-12. Femandez, E., and Cardenas, J. (1981b). Planta 153, 254-257. Filner, P. (1966). Biochim. Biophys. Acta 118, 299-310. Funkhouser, E. A., and Ramadoss, C. S . (1980). Plant Physiol. 65, 944-948. Garrett, R. H. (1978). Microbiology pp. 324-329. Garrett, R. H., and Amy, N. K. (1978). Adv. Microb. Physiol. 18, 1-65. Garrett, R. H., and Cove, D. J. (1976). Mol. Gen. Genet. 149, 179-186. Garrett, R. H., and Nason, A. (1967). Proc. Natf. Acad. Sci. U.S.A. 58, 1603-1610. Garrett, R. H., and Nason, A. (1969). J . Biol. Chem. 244, 2870-2882. Gebhardt, C:, Frankhauser, H., and King, P. J. (1982). “Plant Tissue Culture,” pp. 463-464. by Fujiwara, Tokyo. Giri, L., and Ramadoss, C. S . (1979). J . Biol. Chem. 254, I 1703-1 1712. Glimelius, K., Eriksson, T . , Grafe, R., and Muller, A. J. (1978). Physiol. Plant. 44, 273-277. Gonzalez, A., Tenorio, M., Vaca, G., and Mora, J. (1983). J . Bacteriol. 155, 1-7. Grafe, R., and Miiller, A. J. (1983). Theor. Appl. Genet. 66, 127-130. Greenbaum, P., Prodouz, K . N., and Garrett, R . H. (1978). Biochim. Biophys. Acta 526, 52-64. Grove, G., and Marzluf, G. A. (1981). J . Biol. Chem. 256, 463-470. Guerrero, M. G . , and Gutierrez, M. (1977). Biochim. Biophys. Aria 482, 272-285. Guerrero, M. G., Jetschman, K., and Volker, W. (1977). Biochim. Biophys. Acta 482, 19-26. Guerrero, M. G., Vega, J. M., and Losada, M. (1981). Annu. Rev. Plant Physiol. 32, 169-204. Gupta, D. P., Gupta, M., and Schieder, 0. (1983). Mol. Gen. Genet. 188, 378-383. Hageman, R. H., and Reed, A. J. (1979). In “Methods in Enzymology” (Moldave, Kivie, and Grossman, eds.), Vol. 59, pp. 270-280. Academic Press, New York. Hankinson, 0. (1974). J . Bacteriof. 117, 1121-1 130. Hankinson, O., and Cove, D. J. (1974). J . Biol. Chem. 249, 2344-2353.
NITRATE ASSIMILATION IN EUKARYOTIC CELLS
47
Hewitt, E. J., and Notton, B. A. (1980). I n “Molybdenum and Molybdenum-containing Enzymes’’ (M. Coughlan, ed.), pp. 273-325. Pergamon, Oxford. Homer, R. D. (1983). Biochim. Biophys. Acta 744, 7-15. Howard, W. D., and Solomonson, L. P. (1982). J . B i d . Chem. 257, 10243-10250. Hughes, K., Case, M. E., Geever, R., Vapnek, D . , and Giles, N. H. (1983). Proc. Natl. Acad. Sci. U . S . A . 80, 1053-1057. Hummelt, G., and Mora, J. (1980). Biochem. Biophys. Res. Commun. 92, 127-133. Huskey, R. J., Semenkovich, C. F., Griffin, B. E., Cecil, P. O., Callahan, A. M., Chace, K. V., and Kirk, D. L. (1979). Mol. Gen. Genet. 169, 157-161. Hynes, M. J. (1973). J . Gen. Microbiol. 79, 155-157. Hynes, M. J., Corrick, C. M., and King, J. A. (1983). Mol. Cell. Biol. 3, 1430-1439. Jacob, G. S., and Orme-Johnson, W. H. (1980). In “Molybdenum and Molybdenum-containing Enzymes” (M. Coughlan, ed.), p. 327. Pergamon, Oxford. Johnson, J. L., and Rajagopalan, K. V. (1982). Proc. Natl. Acad. Sci. U.S.A. 79, 6856-6860. Johnson, J. L., Hainline, B. E., and Rajagopalan, K. V. (1980). J . Biol. Chem. 255, 1783-1786. Jolly, S. O., and Tolbert, N. E. (1978). Plant Physiol. 62, 197-203. Jolly, S . O., Campbell, W. H., and Tolbert, N. E. (1976). Arch. Biochem. Biophys. 174,429-439. Keesey, J. K., and De Moss, J. A. (1982). J . Bacteriol. 152, 954-958. Ketchum, P. A,, Cambier, H. Y . , Frazier, W. A,, Madansky, C. H., and Nason, A . (1970). Proc. Nafl. Acad. Sci. U . S . A . 66, 1016-1023. Ketchum, P., Zeeb, D. D., and Owens, M. S . (1977). J . Bacteriol. 131, 884-890. King, P. J., and Khanna, V. (1980). Plant Physiol. 66, 632-636. King, P. J . , Fankhouser, H., Gebhardt, C . , Shiamato, K., and Strauss, A. (1982). In “Plant Tissue Culture.” pp. 447-448. Fujiwara, Tokyo. Kinghorn, J. R . , and Hawkins, A. R. (1982). Mol. Gen. Gener. 186, 145-162. Kinnaird, J. A,, Keighren, M. A , , Kinsey, J. A., Eaton, M., and Fincham, J. R. S . (1982). Gene 20, 387-396. Kleinhofs. A , , Warner, R. L., Muehldauer, F. J . , and Nilan, R. A . (1978). Muta?. Res. 51, 29-35. Kleinhofs, A., Kuo, T., and Warner, R. L. (1980). Mol. Gen. Genet. 177, 421-425. Kleinhofs, A., Taylor, J., Kuo, T. M . , Somers, D. A., and Warner, R. L. (1983). In “Genetic Engineering in Eukaryotes” (P, F. Lurquin and A. Kleinhofs, eds.), pp. 215-231. Plenum, New York. Kuo, T. M.. Somers, D. A., Kleinhofs, A., and Warner, R. L. (1982). Biochim. Biophys. Acta 708, 75-8 1. Kuo, T., Kleinhofs, A., Somers, D., and Warner, R . L. (1981). Mol. Gen. Genet. 181, 20-23. Kuo, T. M., Kleinhofs, A , , Somers, D. A,, and Warner, R. L. (1984). Phytochemistry 23, 229232. Lafferty, M . A., and Garrett, R. H. (1974). J . Biol. Chem. 249, 7555-7567. Lancaster, J. R . , Vega, J. M., Kamin, H., Orme-Johnson, N. R . , Orme-Johnson, W. H . , Kreuger, R. J . , and Siegel, L. M. (1979). J . Biol. Chem. 254, 1268-1272. Lazar, G. B., Fankhaauser, H., and Potrykis, I . (1983). Mol. Gen. Genet. 189, 359-364. Lewis, C. M., and Fincham, J. R. S. (1970). J . Bucteriol. 103, 55-61. Lorimer, G. H . , Gewitz, H.-S., Volker, W., Solomonson, L. P . , and Vennesland, B . (1974). J . Biol. Chem. 249, 6074-6079. Losada, M. (1974). I n “Metabolic Interconversion of Enzymes” (E. H. Fischer, E. G. Frebs, H. Neurath, and E. R . Stadtman, eds.), pp. 257-270. Springer-Verlag, Berlin and New York. Losada, M., Guerrero, M. G., and Vega, J. M. (1981). In “Biology of Inorganic Nitrogen and Sulfur” (H. Bothe and A . Trebst, eds.), pp. 30-63. Springer-Verlag, Berlin and New York. MacDonald, D. W. (1982). Curr. Genet. 6 , 203-208. MacDonald, D. W., and Cove, D. J. (1974). Eur. J . Biochem. 47, 107- I 10.
48
NIGEL S. DUNN-COLEMAN ET AL.
MacDonald, D. W., Cove, D. J., and Coddington, A. (1974). Mol. Gen. Genet. 128, 187-189. Marton, L., Dung, T. M., Mendel, R. R., and Maliga, P. (1982). Mol. Gen. Genet. 182,301-304. Marzluf, G. A. (1981). Microbiol. Rev. 45, 457-461. Mendel, R. R., and Muller, A. J. (1979). Mol. Gen. Genet. 177, 145-153. Mendel, R. R., and Muller, A. J. (1980). Plant Sci. Lett. 18, 277-288. Mendel, R. R., Alikulov, Z. A,, Lvov, N. P., and Muller, A. J. (1981). Mol. Gen. Genet. 181,395400. Mendel, R. R., Marton, L., and Wray, J. L. (1982a). Proc. Int. Symp. Nitrate Assimilat., Gatersleben (Abstr.). Mendel, R. R., Alikulov, 2. A., and Miiller, A. J. (1982b). Plant Sci. Lett. 25, 67-72. Mendel, R. R., Alikulov, Z. A., and Miiller, A. J. (1982~).Plant Sci. Lett. 27, 95-101. Mendel, R. R., Buchanan, R. J., and Wray, J. L. (1984). Molec. Gen. Gent. 195, 186-189. Minagawa, N., and Yoshimoto, A. (1982). J. Biochem. 91, 761-774. Miiller, A. J. (1983). Mol. Gen. Genet. 192, 275-281. Miiller, A. J., and Grafe, R. (1978). Mol. Gen. Genet. 161, 67-76. Miiller, A. J., and Mendel, R. R. (1982). “Plant Tissue Culture,” pp. 233-234. Fujiwara, Tokyo. Murphy, M. J., Siegel, L. M., Tove, S. R., and Kamin, H. (1974). Proc. Nut/. Acad. Sci. U.S.A. 71, 612-616. Narayanan, K. R., Somers, D. A,, Kleinhofs, A., and Warner, R. L. (1983). Mol. Gen. Genet. 190, 222-226. Narayan, N. R., Miiller, A,, Kleinhofs, A,, and Warner, R. (1984). Plant Physiol. Suppl. Abstr. 75. Nason, A,, and Evans, H. J. (1953). J. Biol. Chem. 202, 655-673. Nason, A , , Antoine, A , , Ketchum, P. A., Frazier, W., and Lee, D. (1970). Pror. Nut/. Acad. Sci. U.S.A. 65, 137-144. Nason, A., Lee, K.-Y., Pan, S.-S., Ketchum, P. A,, Lamberti, A., and DeVries, J. (1971). Proc. Natl. Acad. Sci. U.S.A. 68, 3242-3246. Nelson, R. A,, Ryan, S. A,, and Harper, J. E. (1983). Plant Physiol. 72, 503-509. Nichols, G. L., and Syrett, P. J. (1978). J . Gen. Microbiol. 108, 71-78. Nichols, G. L., Shehata, S. A. M., and Syrett, P. J. (1978). J . Gen. Microbiol. 108, 79-88. Notton, B. A., and Hewitt, E. J. (1979). In “Nitrogen Assimilation of Plants” (E. J . Hewitt and C. V. Cutting, eds.), pp. 227-244. Academic Press, New York. Notton, B. A., and Hewitt, E. J. (1983). Plant Sci. Lett. 29, 107-1 13. Notton, 8. A,, Fido, R. J., and Hewitt, E. J. (1977). Plant Sci. Lett. 8, 165-170. Notton, 8. A. , Fido, R. J., and Hewitt, E. J. (1983). Plant Sci. Lett. 29, 107-113. Orihel-Iranzo, B . , and Campbell, W. H. (1980). Plant Physiol. 65, 595-599. Pan, S.-S., and Nason, A. (1978). Biochim. Bioph-ys. Acta 523, 297-313. Pateman, J. A,, and Cove, D. J. (1967). Nature (London) 215, 1234-1237. Pateman, J. A , , and Kinghorn, J. R. (1977). I n “Genetics and Physiology of Aspergillus” (J. E. Smith and J. A. Pateman, eds.), pp. 202-241. Academic Press, New York. Pateman, J. A,, Cove, D. J., Rever, B. M . , and Roberts, D. B. (1964). Nature (London) 201, 5860. Pental, D., Cooper-Bland, S., Harding, J., Cocking, E. L., and Miillcr, A. J. (1982). 2. Pflanzenphysiol. 105, 219-227. Prabhakara, D., Choudary, P. V., and Rao, G. R. (1976). Biochem. Biophys. Res. Commun. 7 2 , 598-602. Premakumar, R., Sorger, G. J., and Gooden, D. (1978). Biochim. Biophys. Acta 519, 275-278. Premakumar, R., Sorger, G. J., and Gooden, D. (1979). J . Bacteriol. 137, 1 119-1 126. Premakumar, R., Sorger, G. J., and Gooden, D. (1980). J . Bacteriol. 144, 542-551. Prodouz, K. N., and Garrett, R. H. (1981). J . Biol. Chem. 256, 971 1-9717.
NITRATE ASSIMILATION IN EUKARYOTIC CELLS
49
Quershi, J. A., Buchanan, R. J . , and Wray, J. L. (1982). Proc. Int. Symp. Nitrate Assimilat., Gatersleben (Abstr. ), Redinbaugh, M. G . , and Campbell, W. H. (1981). Plant Physiol. 68, 115-120. Redinbaugh, M. G . , and Campbell, W. H. (1983). Plant Physiol. 71, 205-207. Renosto, F., Omitz, D. M . , Peterson, D., and Segel, I. H. (1981). J . Biol. Chem. 256, 8616-8625. Rivas, J., Guerrero, M. G . , Paneque, A., and Losada, M. (1973). Plant Sci. Lett. 1, 105-113. Rossman, M. G . , Moras, D., and Olsen, K. W. (1974). Nature (London) 250, 194-199. Ryan, S. A . , Nelson, R. S . , and Harper, J. E. (1983). Plant Physiol. 72, 510-514. Rucklidge, G . , Notton, B. A., and Hewitt, E. J. (1976). Biochem. SOC. Trans. 4, 77-80. Sanchez, F., Calva, E., Compomanes, M . , Blanco, L., Guzman, J., Saboris, J. C., and Palacios, R. (1980). J. Biol. Chem. 255, 2231-2234. Schechtman, M., and Yanofsky, C. (1983). Neurosporu Newslett. 30, p. 19. Schell, J. and Van Montagu, M. (1983). Biorechnology 1, 175-180. Schrader, L. E., Ritenour, G. L., Eilrich, G. L., and Hageman, R. H. (1968). Plant Physiol. 43, 930-940. Schweizer, M., Case, M. E., Dykstra, C. L., Giles, N. H., and Kushner, S . R. (1981). Proc. Nutl. Acud. Sci. U.S.A. 78, 5086-5090. Scott, A. I., Irwin, A. J., Siegel, L. M . , and Schoolery, J. A. (1978). J. Am. Chem. Sac. 100,316318. Shen, T. C., Funkhouser, E. A,, and Guerrero, M. G . (1976). Plant Physiol. 58, 292-297. Sherrard, J. H., and Dalling, M. J. (1979). Plant Physiol. 63, 346-353. Sherrard, J. H., Kennedy, J. A,, and Dalling, M. J. (1979). Plant Physiol. 64, 640-645. Siegel, L. M., Davis, P. S . , and Kamin, H. (1974). J. Biol. Chem. 249, 1572-1586. Smarrelli, J . , Jr., and Campbell, W. H. (1981). Planr Physiol. 68, 1226-1230. Smarrelli, J., Jr., and Campbell, W. H. (1983). Biochim. Biophys. Acra 743, 435-445. Smarrelli, J., Jr., and Garrett, R. H. (1982). Fed. Proc. Fed. Am. Soc. Exp. Biol. 41, 756. Solomonson, L. P. (1974). Biochim. Biophys. Actu 334, 297-308. Solomonson, L. P. (1975). Biochim. Physiol. 56, 853-855. Solomonson, L. P., and Spehar, A. M. (1977). Nature (London) 265, 373-375. Solomonson, L. P., Lorimer, G . H., Hall, R. L., Borchers, R., and Bailey, J. L. (1975). J. Biol. Chem. 250,4120-4127. Somers, D. A., Kuo, T. M . , Kleinhofs, A., and Warner, R. L. (1983). Ptanrfhysiol. 71, 145-149. Sorger, G . J., and Davies, I. (1973). Biochem. J. 134, 673-685. Sorger, G. J., Premakumar, R., and Gooden, D. (1978). Biochim. Biophys. Actu 540, 33-47. Sosa, F. M., Ortega, T., and Barea, J . L. (1978). Plant Sci. Lett. 11, 51-58. Stewart, G. R., and Orebamjo, T. 0. (1979). New Phytol. 83, 311-319. Stohl, L. L., and Lambowitz, A. M. (1983). Proc. Narl. Acad. Sci. U.S.A. 80, 1058-1062. Strauss, A., Bucher, F., and King, P. J. (1981). Plunta 153, 75-80. Tachiki, T., and Nason, A. (1983). Biochim. Biophys. Acta 744, 16-22. Taylor, J. L., Kleinhofs, A., Bedbrook, J . R . , and Grant, F. I. (1982). Genetic Engineering: Applications to Agriculture. Abstract No. 41. Taylor, J. L., Bedbrook, J. R., Grant, F. J., and Kleinhofs, A . (1983). J . Molec. Appl. Genet. 2, 261 - 17I . Tilbum, J., Scazzocchio, C., Taylor, G., Lockington, A., Zabicki-Sissman, J. 0.. and Davis, R. W.(1983). Gene 26, 205-221. Tokarev, B. I . , and Shurnny, V. K. (1977). Genetika (Moscow) 13, 2097-2103. Tomsett, A. B., and Cove, D. J. (1979). Genet. Res. Camb. 34, 19-32. Tomsett, A. B., and Garrett, R. H. (1980). Genetics 95, 649-660. Tomsett, A. B., and Garrett, R. H. (1981). Mol. Gen. Genet. 184, 183-190.
50
NIGEL S. DUNN-COLEMAN ET AL.
Tomsett, A. D., Dunn-Coleman, N. S., andGarrett, R . H. (1981). Mol. Gen. Genet. 182,229-233. Vapnek, D., Hautala, J . A., Jacobson, J . W., Giles, N. H., and Kushner, S. R. (1977). Proc. Natl. Acad. Sci. U.S.A. 74, 3508-3512. Vega, J. M. (1976). Arch. Microbiol. 109, 237-242. Vega, J. M., and Kamin, H. (1977). J . Biol. Chem. 252, 896-909. Vega, J . M., Garrett, R. H., and Siegel, L. M. (1975). J . Biol. Chem. 250, 7980-7989. Vennesland, B., and Guerrero, M. G. (1979). Encycl. Plant Physiol. New Ser. 6, 425-444. Wallace, W. (1974). Biochim. Biophys. Acta 341, 265. Wallace, W. (1975). Biochim. Biophys. Acta 377, 239-250. Walls, S., Sorger, G. J . , Gooden, D., and Klein, V. (1978). Biochim. Biophys. Acta 540, 23-32. Warner, R. L., and Kleinhofs, A. (1981). Plant Physiol. 67, 740-743. Warner, R. L., Lin, C. J., and Kleinhofs, A . (1977). Nurure (London) 269, 406-407. Warner, R. L., Kleinhofs, A , , and Muehlbaver, F. J. (1982). Crop Sci. 22, 389-392. Wilkerson, J. O., Janick, P. A , , and Siegel, L. M. (1983). Fed. Proc. Fed. Am. Soc. Exp. Biol. 42, 2060. Wrey, J. L., and Kirk, D. W. (1981). Plant Sci. Lett. 23, 207-213. Yamaya, T., Oaks, A., and Boesel, I. L. (1980). Plant Physiol. 65, 141-145. Xaun, L. T., Grafe, R . , and Muller, A. J. (1983). In?. Protoplast Symp. 6th, Basel, Switzerland (Abstr.). Zauner, E., and Dellweg, H. (1983). Eur. J . Microbiol. Biotech. 17, 90-95.
INTERNATIONAL REVIEW OF CYTOLOGY, VOL. 92
Endocytosis and Exocytosis: Current Concepts of Vesicle Traffic in Animal Cells MARKC. WILLINGHAM AND IRAPASTAN Laboratory of Molecular Biology, National Cancer Institute, National Institutes of Health, Bethesda, Maryland I. Introduction ......... ........................... 11. Events at the Plasma Membrane.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . A . Receptors in the Plasma Membrane . . . . . . . . . . . . . . . . . . . . . . . B. Clustering in Coated Pits.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. The Structure of Coated Pits: Clathrin ..................... D. Vesicle Formation in Endocytosis from Coated Pits.. . . . . . . . . 111. Endocytic Vesicles: Characteristics. . A. Receptosomes ......................................... B. Multivesicular Bodies ............. D.
...........................
Tubular Elements
F. Macro- and Micropinocytosis . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV. Kinetic Classes of Endocytosis.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V. Compartmentalization in the Golgi . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Endocytosis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . _.._.. B. Exocytosis: Materials Destined for the Cell Surface . . . . . . . . . . C. Lysosomal Delivery D. Materials from the Endoplasmic Reticulum.. . . . . . VI. Functions of Endocytic and Exocytic Membrane Traffic A. Recycling of Receptors and Ligands, Degradation . . . . . . . . . . . B. Traffic Control of Synthesized Materials . . . . . . . . . . . . . . . . . . . C. Pathologic Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D. Hormonal Delivery?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E. Other Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VII. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
51
53 53 56 58 59 66 67
70 70 72 72 72 76
79 80 81 83 84 85 85
I. Introduction In recent years, a large amount of research in cell biology has concentrated on the events surrounding the binding, internalization, processing, and degradation of macromolecules that enter cells from their external environment (reviewed in Farquhar, 1983; Farquhar and Palade, 1981; Goldstein et al., 1979; Pastan and Willingham, 1981a,b, 1983; Pearse and Bretscher, 1981; Silverstein et al., 51 Copyright 0 1984 by Academic Press, Inc All rights of reproduction in any form reserved ISBN 0-12-364492-5
52
MARK C. WILLINGHAM A N D IRA PASTAN
1976; Steinman et al., 1983). This work has demonstrated that cells can selectively remove materials from their environment, and route these materials to different intracellular sites in a carefully controlled manner. Much of the data on the nature of the pathways taken by materials coming from the outside, as well as those originating from the inside, has been descriptive, but recently some of the molecular details of these processes have been uncovered. While there are still many unknown factors that control the shuttling of molecules between various membrane compartments, enough new information is available to justify a review of this subject and some speculation on possible mechanisms that might mediate these and related processes. The major thrust of many recent studies has been to understand how a cell can select from its surroundings a certain type of macromolecule, concentrate it on the cell surface, internalize it, and send it to a specific intracellular destination, where it may be degraded or returned intact to the cell surface. A considerable effort has also been devoted to the elucidation of how molecules synthesized within the cell are directed to the proper location. In some cases the same pathways appear to be involved. Much of this review will concentrate on work in our own laboratory as it relates to data in other systems. A schematic drawing summarizing the pathways to be discussed in this review is shown in Fig. 1.
P Me
Granule [Only in specialized
9 ‘ Cis
Nucleus
FIG. 1. Diagrammatic summary of the major pathways of endocytosis and exocytosis in animal cells. The receptors shown are examples of two of the different pathways possible: the solid receptor with its appropriate ligand (open square) represents the EGF receptor: the open receptor and its ligand (open triangle) represents the transferrin receptor.
ENDOCYTOSlS AND EXOCYTOSIS
53
11. Events at the Plasma Membrane
A. RECEPTORS IN THE PLASMA MEMBRANE A wide variety of receptors are found on the surface of animal cells. Receptors have been identified for hormones and growth factors (Ackerman and Wolken, 1980, 1981; Ackerman et al., 1983; Ahmed et al., 1981; Bergeron et al., 1979; Carpenter and Cohen, 1976; Carpentier et al., 1978, 1979a,b, 1982; Cheng et al., 1980; Conn et al., 1978; Goldfine et al., 1977; Gorden et al., 1978a,b; Haigler et al., 1978, 1980a,b; Hillman and Schlessinger, 1982; Horiuchi et al., 1982, 1983; Kahn, 1976; King and Cuatrecasas, 1981; Levi et al., 1980; Maxfield et al., 1978, 1981a,b, 1979b; Orci e t a l . , 1978; Pastan rtal., 1981; Posner, et al., 1981; Varga et al., 1976), plasma proteins (Anderson et al., 1976, 1977, 1982; Ashwell and Morell, 1974; Bleil and Bretscher, 1982; Brown et al., 1982, 1983; Courtoy et al., 1982; Dautry-Varsat et al., 1983; Dickson et al., 1981a-c, 1982a-d, 1983; Fan et al., 1983; Geuze et al., 1983; Goldstein et al., 1979; Hanover et al., 1983, 1984; Iacopetta et al., 1983; Kaplan and Nielson, 1979; Kaplan et al., 1977; Klausner et al.. 1983; Maxfield et al., 1978, 1979a,b, 1981a,b; Neufeld and Ashwell, 1979; Neufeld et al., 1975; Pastan et al., 1977, 1981; Pastan and Willingham, 1981a,b; Pricer and Ashwell, 1971; Sando and Neufeld, 1977; Sly et al., 1981; Van Leuven et al., 1978; Via et al., 1982; Wall et al., 1980; Willingham et al., 1979, 1981a-c, 1983a-c; Willingham and Pastan, 1978, 1980, 1981a,b, 1982, 1983c,d), toxins (Donovan et al., 1981; FitzGerald et al., 1980, 1983a,b; Iglewski et al., 1977; Pappenheimer, 1977; Sandvig and Olsnes, 1980; Yamaizumi et al., 1978), viruses (Chardonnet and Dales, 1970; Dales, 1973, 1978; Dales and Choppin, 1962; Dales and Hanafusa, 1972; Dickson et al., 198Ic; Helenius et al., 1980; Helenius and Marsh, 1982; Marsh and Helenius, 1980; M a t h et al., 1982; Schlegel et al., 1982a,b, 1983; Simons et al., 1982; Simpson et al., 1969; Wehland et al., 1982a; Willingham et al., 1981a), and other substances found in the blood or other body fluids, as well as artificial materials (Abrahamson et al., 1979; Abrahamson and Rodewald, 1981; Adams et al., 1982; Bessis, 1963; Bessis and Breton-Gorius, 1962; Bretscher, 1983; Bretscher and Thomson, 1983; Bretscher et al., 1980; Fan et al., 1982; Farquhar, 1978, 1981, 1983; Fawcett, 1964; FitzGerald e t a l . , 1983b; Gonatas et al., 1977, 1980; Herzog and Farquhar, 1977; Herzog and Reggio, 1980; Joseph et al., 1978, 1979; Nagura et al., 1979; Nicolson, 1974; Ottosen et al., 1980; Petersen and van Deurs, 1983; Policard and Bessis; 1958; Rodewald, 1973; Rodewald and Abrahamson, 1980; Roth et al., 1976; Roth and Porter, 1962; Ryser et al., 1982; Salisbury et al., 1980; Stome, 1979; Willingham et al., 1979. 1981d; Woodward and Roth, 1978; Zoon et al., 1983). These receptors are generally, but not always, intrinsic membrane proteins that have a high affinity for a specific ligand.
54
MARK C. WILLINGHAM AND IRA PASTAN
Much of our own work has concentrated on studying the binding and internalization of the plasma protein a,-macroglobulin (aZM) and its receptor. a2M is a tetrameric glycoprotein present in plasma which has the ability to complex tightly with a wide variety of proteases (Heimburger, 1979; van Leuven et al., 1978). On complexing with a protease or upon chemical reaction with amines, a,M becomes modified to a form that is recognized by a specific receptor on the surface of many mesenchymal cells, including many lines of tissue culture fibroblasts. These cultured cells have up to 5 X lo5 receptors per cell (Dickson et al., 1981b; Hanover et al., 1984). About 10-20,000 have a very high affinity; the rest have somewhat lower affinity. a,M is conveniently made from outdated human plasma. The human form of the protein binds well to the receptor on rodent cell lines (Pastan ef al., 1977). The large size of this protein (750,000 daltons) makes chemical derivitization of this molecule relatively easy. Another ligand that has been extensively studied is low-density lipoprotein (LDL) for which there are specific receptors on cultured fibroblasts (Anderson et al., 1977; Goldstein et al., 1979; Handley et al., 1981; Vermeer et al., 1980; Via et al., 1982). It was in the early morphologic studies with labeled LDL that selective localization of this ligand was seen in the coated pits of the plasma membrane of fibroblasts (Anderson et al., 1976, 1977). Accumulation in coated pits has been seen for many ligands bound to their receptors (Abrahamson and Rodewald, 1981; Ackerman et al., 1983; Bleil and Bretscher, 1982; Bretscher and Thomson, 1983; Brown et al., 1983; Carpentier et al., 1982; Courtoy et al., 1982; FitzGerald er al., 1983a,b; Iacopetta et al., 1983; McGookey et al., 1983; Pastan and Willingham, 1981a,b, 1983; Rodewald and Abrahamson, 1980; Roth et al., 1976; Salisbury et al., 1980; Via er al., 1982; Wall et al., 1980; Willingham et al., 1979, 1981a,c, 1983b; Willingham and Pastan, 1980, 1981a,b, 1982, 1983c,d; Zoon et al., 1983). For some ligands, the initial distribution of unoccupied receptors appears to be diffuse over the entire plasma membrane. The lateral mobility of some of these receptors when labeled with bound ligands has been measured by the fluorescence photobleaching recovery technique (Axelrod et al., 1976; Edidin et al., 1976; Hillman and Schlessinger, 1982; Jacobson and Poste, 1976; Maxfield et al., 1981a,b; Schlessinger et al., 1978). These studies show that most of these receptors move randomly and rapidly in the plane of the plasma membrane. These data indicate that at 37°C each receptor will move through one of the 1000 or so coated pits found on most types of cultured cells in about three seconds. Thus, the distribution of receptors can change rapidly. Evidence for the LDL system has suggested that many of the receptors for LDL are concentrated in coated pits in the absence of added ligand (Anderson et al., 1982). The receptor for transferrin and the phosphomannosyl receptor also are found clustered in coated pits in the absence of exogenously added ligand (Willingham et al., 1983b, 1984). Presumably this clustering is related to the fact that these receptors continuously recycle in order to bring metabolically
ENDOCYTOSIS AND EXOCYTOSIS
55
important substances such as cholesterol (LDL), iron (transferrin), or newly synthesized lysosomal components into the cell. On the other hand, the epidermal growth factor (EGF) receptor of KB cells has been shown to cluster in coated pits only after the addition of ligand. In addition, this clustering is temperature dependent. It is unable to do so at 4°C in spite of the fact that it is mobile in the membrane at this temperature (Hillman and Schlessinger, 1982; Willingham and Pastan, 1982). When the binding characteristics for a,M were analyzed using radiolabeled material, the binding affinities indicate that there are two classes of receptors (Dickson et al., 1981b; Hanover et al., 1984). The high-affinity class of receptors is detected under conditions in which the ligand-receptor complexes are found clustered in coated pits on the cell surface. This finding has led to the suggestion that there may be some conformational change in a,M receptors that results in both a higher affinity for a2M and accumulation in coated pits. In order for different types of receptors to become clustered in the same coated pit, receptors should share some common structural feature that enables a recognition system in the pit to selectively concentrate them prior to internalization. No protein structural data to indicate the nature of this common feature is available at this time. The number of receptors on the plasma membrane is often, but not always, regulated by occupancy. For example, the number of receptors for a,M and transferrin are relatively constant under various culture conditions. On the other hand, the number of the receptors for LDL, EGF, and insulin are lowered or down-regulated by the presence of LDL, EGF, or insulin, respectively. Since serum contains LDL, exposure of cells to LDL-deficient medium for a few days is necessary to increase the number of LDL receptors (Goldstein et al., 1979). Receptors are not always uniformly distributed over the cell surface. In hepatocytes, the cells are separated by junctional elements such that only one side of each cell has receptors for the asialoglycoprotein receptor (Wall et al., 1980). In a similar fashion, the upper and lower cell surfaces of cell lines like MDCK, which have junctional complexes, have been found to contain different components (Rodriquez-Boulan and Pendergast, 1980). Thus, cells can regulate the total number of receptors on their surface, and the portion of the surface on which they are located. Some cell types have domains of their membranes which have higher concentrations of receptors than adjacent areas in which junctional complexes have no role. For example, microvilli have been reported to have higher receptor numbers than adjacent areas of membrane (Weller, 1974). When HCG is added to granulosa cells, the ligand bound to its receptor appears to have an extremely long residence time on the cell surface (8 hours) (Ahmed et al., 1981). Because many components of the cell surface are internalized every few hours, this unusual receptor system probably may have a special mechanism for keeping it on the cell surface.
56
MARK C. WILLINGHAM AND IRA PASTAN
B. CLUSTERING IN COATED PITS The entry of ligand-receptor complexes into the specific endocytic pathway from the cell surface is mediated by a unique structure termed a coated pit which is present on most animal cells. In many cultured mammalian cells there are 500-1500 coated pits on the surface of each cell. These structures serve as foci in which receptor-ligand complexes are concentrated just prior to internalization. In this manner a cell can selectively concentrate ligands from other parts of the cell surface and internalize them, without internalizing other portions of the plasma membrane. Coated pits at the plasma membrane are relatively uniform in size in any one cell type (100-150 nm across). They derive their name from the fuzzy bristle coat that can be seen by transmission electron microscopy on the cytoplasmic face of the plasma membrane in these regions. They have been postulated to have a specific association with actin-containing microfilament bundles in cultured cells because they have been noted to be aligned along stress fibers in flat cells (Anderson et al., 1980; Salisbury et al., 1980). However, this association with stress fibers is more likely related to overall cell organization than to specific interactions of coated pits with microfilament bundles (Willingham et al., 1979, 1981e). Coated pits involved in endocytosis were probably first observed in erythroblastic bone marrow cells by Bessis alid co-workers (Bessis, 1963; Bessis and Breton-Gorius, 1962; Policard and Bessis, 1958), but the nature of the coat of these pits was not recognized. It was subsequently demonstrated by Fawcett (1964) that the pits described by Bessis were actually “coated” pits. In 1961, Gray published images of the coat structures in neuronal cells (Gray, 1961), although he later questioned their nature (Gray, 1972). Between 1962 and 1964, a number of reports demonstrated the existence of coated structures in diverse cell types such as liver (Ashford and Porter, 1962; Rouiller and Jezequel, 1963), insect oocytes (Anderson, 1964; Roth and Porter, 1962, 1964), other insect cells (Bowers, 1964), kidney (Wissig, 1962), and neuronal cells (Brightman, 1962; Brightman and Palay, 1963). The possible role of these structures in endocytic activity was repeatedly discussed. Roth and Porter (1962, 1964) coined the term “bristle-coated pits”; this terminology has been widely used. The concentration of a specific ligand in coated pits was first demonstrated using labeled LDL on cultured fibroblasts (Anderson et al., 1976). Subsequently, many ligands have been found to be concentrated in coated pits in studies on a wide variety of cultured cells and in intact animals. The process of clustering in coated pits varies in detail from receptor to receptor. Some receptors, such as those for EGF, are diffusely distributed when unoccupied, and cluster in coated pits only after addition of ligand (Willingham and Pastan, 1982). LDL receptors, as mentioned above, appear to be able to cluster even in the absence of ligand (Anderson et al., 1982). Clustering of the
ENDOCYTOSIS AND EXOCYTOSIS
57
EGF-EGF receptor complex in coated pits does not occur at 4°C; higher temperatures are required for clustering to occur (Willingham and Pastan, 1982). This inhibition of clustering does not appear to be due to restricted lateral mobility of the EGF receptor (Hillman and Schlessinger, 1982). A number of chemical substances have been found to inhibit clustering in coated pits (Cheng ef al., 1980; Davies et al., 1980; Dickson et al., 1981b, 1982b,c; FitzGerald et al., 1980; Haigler et al., 1980a,b; Hanover et al., 1983; Levitzki et al., 1980; Maxfield et al., 1979a; Schlegel et al., 1982a; Via et al., 1982). One of the most potent of these inhibitors is dansylcadaverine. Because dansylcadaverine is a potent inhibitor of the enzyme transglutaminase, it was suggested that this enzyme might have a role in the clustering process (Davies et al., 1980). Recently, dansylcadaverine has also been found to be a potent inhibitor of phosphatidylcholine biosynthesis (Mato et al., 1983) and to stimulate phosphotidylinositol synthesis. It is hoped that studies with compounds that block clustering in coated pits may help clarify the biochemical mechanism that regulates the clustering process. One area that has been overlooked in considerations of the coated pit pathway is its contribution to nonconcentrative endocytosis, both adsorptive and fluidphase. Some of the confusion has come from experiments in which a marker, such as horseradish peroxidase (HRP), that had been used in one cell type (such as macrophages) to identify the major pathway of fluid-phase endocytosis, was used in a different cell type (such as cultured fibroblasts) and labeled a different organelle. Macrophages and fibroblasts differ significantly in their surface ruffle and macropinocytotic activities. When cultured fibroblasts are incubated at 4°C with HRP, there is no specific binding of HRP to the plasma membrane; thus, it is not an adsorptive marker. However, when the incubation is conducted at 37"C, a large amount of HRP is rapidly found in the same endocytic vesicles that contain ligands that, in double-label experiments, are found bound to the cell surface and concentrated in coated pits prior to entry (Ryser et al., 1982; Willingham and Pastan, unpublished data). Thus, we can conclude that in such cells HRP enters via coated pits, whereas in macrophages it may mainly enter in macropinosomes. The question that often arises is how much of the observed uptake is fluid-phase through the coated pit system, and how much is adsorptive through the coated pit system. One striking fact that is clear from quantitative studies on ligand uptake is that internalization by the coated pit system can be very rapid. Estimates from quantitative ligand uptake suggest that each coated pit may be able to cluster ligands and produce an endocytic vesicle every 20 seconds at 37°C (Pastan and Willingham, 1981b). The large amount of fluid taken up rapidly by this pathway indicates the slower processes of macropinosomal and caveolae vesicle formation are probably much less important in fluid uptake in these types of cultured cells than previously thought.
58
MARK C. WILLINGHAM AND IRA PASTAN
C. THE STRUCTURE OF COATED PITS:CLATHRIN Following the morphologic description of coated pits as structures in fixed cells by electron microscopy, coated vesicles were isolated from cell homogenates and subjected to biochemical analysis. Pearse (1975, 1976, 1978) identified the major protein component of coated vesicle preparations as a protein with a M, of 180,000, which formed a basket-like coat around the vesicle. She named this protein “clathrin,” from the Latin “clathra-” meaning “lattice.” Subsequent work showed that this protein could be purified and induced to form basket-like coat structures in the absence of membranes (Pearse, 1978; Fine et al., 1978; Keen et al., 1979; Woodward and Roth, 1978). The structure of the clathrin coat is rather unique in that it has the appearance of a latticework arranged as hexagons and pentagons similar to the surface of a soccer ball (Crowther and Pearse, 1981; Crowther et al., 1976; Kanaseki and Kadota, 1969; Kirchhausen and Harrison, 1981; Schook et al., 1979; Ungewickell and Branton, 1981). These basket structures have been shown to contain proteins in addition to clathrin and these other proteins may also be required for the integrity of the basket structures seen in the living cell (Kirchhausen and Harrison, 1981; Ungewickell and Branton, 1981; Pearse, 1978; Crowther and Pearse, 1981). Antibodies to clathrin have been produced and used to analyze the location of clathrin in intact cells (e.g., Willingham et al., 1981b). Many, if not all, of these antibodies are directed against one or both of the low-molecular-weight proteins that associate with the 180 kilodalton (kd) clathrin molecule (Brodsky and Parham, 1982; Keen et d., 1981; N. D. Richert, M. C. Willingham and I. Pastan, unpublished data). The components of coated vesicles have been referred to as the clathrin heavy chain (1 80 kd) and clathrin light chains (LC-A and LC-B, 36 and 33 kd) (Kirchhausen and Harrison, 1981; Pearse, 1978). In cultured fibroblasts the clathrin distribution determined by immunocytochemistry corresponds exactly to the distribution of morphologically visible coated structures seen by transmission electron microscopy. Immunocytochemical studies at the ultrastmctural level have also suggested that very little, if any, clathrin is present in other sites (Willingham el al., 1981b). In cultured fibroblasts and epithelial cells there are two populations of clathrincoated structures: one group of coated pits (approx. 150 nm diameter) is associated with the plasma membrane (about 1000 per cell). The other group of coated pits is smaller (approx. 70 nm in diameter) and found in the transreticular elements of the Golgi system (about 200 per cell). Both of these clathrin-coated structures react with antibodies made against clathrin prepared from mammalian brain (rat, bovine, or human). When tissues are homogenized, they yield two populations of closed vesicles which differ in their size. The predominant class obtained from homogenization of brain is the smaller (70 nm) size class. When the coat lattice is investigated by
ENDOCYTOSIS AND EXOCYTOSIS
59
transmission electron microscopy, it appears in sections as a periodic, spikelike array covering the cytoplasmic face of these structures (Fig. 2 ) . When isolated baskets are studied by negative staining or shadowing methods or observed in tangential section in intact cells, one can see the hexagon-pentagon arrangement of the basket. These hexagon-pentagon arrangements are shown in very clear images using cell “fracture” and surface replica techniques (Heuser and Evans, 1980). D. VESICLEFORMATION IN ENDOCYTOSIS FROM COATEDPITS
1 . Coated Vesicles? While coated pits serve as foci of concentration of receptor-ligand complexes on the cell surface, there is disagreement about whether the coated structures themselves ever leave or pinch off from the plasma membrane during the endocytic process. Endocytosis via coated pits occurs very frequently (about every 20 seconds in fibroblasts) and the endocytic event (the transfer of ligand from coated pits to uncoated vesicles within the cell) probably occurs within a few seconds or less at 37°C. It has been assumed that during endocytosis the coated pit pinches off from the plasma membrane to form an isolated coated vesicle, but direct evidence that this occurs in living cells is lacking. The evidence which suggested that these coated pits do pinch off to form coated vesicles is mainly based on two types of experiments. The first is that one can retrieve from cell homogenates intact isolated vesicles bearing a clathrin coat (coated vesicles) (Pearse, 1976). The second is that in electron microscopic images one can see cross sections of vesicular structures bearing a clathrin coat (Roth and Porter, 1964; Friend and Farquhar, 1967; Franke et al., 1976; Anderson et al., 1976). In addition, biochemical characteristics of clathrin indicate that it is able to assemble and disassemble in a test tube with sufficient ease that an assembly process seemed possible in the living cell (Keen et al., 1979). There are some difficulties, however, with these lines of evidence. One problem is that the existence of isolated vesicles after homogenization does not necessarily indicate that the structures existed as isolated vesicles in the living cell. The vesiculation of membrane structures by homogenization is well known for microsomes derived from endoplasmic reticulum or for small plasma membrane vesicles derived from intact plasma membrane. Further, observation of isolated vesicle cross-sections by electron microscopy in section does not prove that these structures are actually isolated (Fig. 2 ) . A glancing section of a structure with a tortuous shape can yield this same image without the structure being isolated from other membranous connections. It has been shown previously that most of the structures that appear to be isolated vesicles in section are, in fact, connected to the cell surface, because they can be heavily labeled by impermeant
60
MARK C. WILLINGHAM AND IRA PASTAN
FIG.2. Morphologic appearance of clathrin-coated pits at the plasma membrane. K B human carcinoma cells (A-K) or Swiss 3T3 mouse fibroblast cells (L-0) were fixed and processed using glutaraldehyde and routine Os04 (F,G,H,I,J,K,), ferrocyanide-reduced Os04 in the OTO sequence (Willingham and Pastan, 1983a) (L,M,N,O), or routine Os04 in the OTO sequence (Willingham and Rutherford, 1984). The typical appearance of clathrin-coated pits is shown in (A-C). (E) shows two different shapes of the coated regions (arrowheads). (B,F) show the appearance of “cryptic” pits
ENDOCYTOSIS AND EXOCYTOSIS
61
markers applied to the cell exterior after cell fixation (Willingham et al., 1981d). This type of study clearly shows that the majority of the plasma membrane coated structures are not isolated but are connected to the cell surface. Other experiments with impermeant markers have shown that the communication of these structures with the cell surface is a temperature-dependent process. Virtually all of these large surface-related coated structures can be shown to communicate with the cell surface at 4°C (Willingham et al., 1981d). However, at 37°C a significant percentage (about 30%) of plasma membrane-type coated structures label poorly with impermeant markers after fixation (Wall et al., 1980; Willingham et al., 1981d). These studies also showed that one could find images of coated pits with narrow necks whose membranes were continuous with the cell surface membrane at 37”C, but failed to admit the impermeant marker (cryptic coated pits; Fig. 2). This has led to the question of whether these colloidal impermeant markers, such as ruthenium red (Wall et al., 1980; Willingham er al., 1981d), lathanum hydroxide (Shaklai and Tavassoli, 1982), lathanum carbonate (M. C. Willingham, unpublished data), tannic acid (Blanchette-Mackie and Scow, 1982; M. C. Willingham, unpublished data), or horseradish peroxidase (Willingham et al., 198 Id) can unequivocally prove the lack of anatomical communication of one structure with another structure. If coated pits pinch off to form coated vesicles, the clathrin must return to the cell surface. This presumably would happen by the coat disassembling and the “soluble” disassembled clathrin returning to the surface. Using antibodies to clathrin, immunocytochemical methods have been employed to determine the (Willingham et al.. 1981d) which exclude impermeant extracellular marker (Ruthenium Red), in spite of their continuity with the plasma membrane. (G-K) are cells incubated with lathanum chloride after fixation in glutaraldehyde, followed by rapid washing in sodium carbonate to produce an impermeant precipitate of lathanum carbonate. (G) demonstrates an image showing the narrow neck connection of a cryptic coated pit (arrowhead) with the cell surface; note the sudden decrease in accessibility to the impermeant marker at the upper edge of the coated region, which admits only small amounts of the marker beyond the restriction (arrows). (H-K) show varying planes of section of pits that show communication with the surface by their content of lathanum label (arrows). (H) shows membrane connection to the surface, which become more and more tangential to the surface connection in (ILK), until in (K) the vesicular profile has the appearance of an isolated vesicle. The presence of lathanum precipitate in (K), however, shows that this structure has a connection to the cell surface that is still intact, since the lathanum was added only after fixation with glutaraldehyde. The lack of labeling with such impermeant markers does not prove that the vesicular structure is not connected to the cell surface, as shown for the cryptic pits shown in (B,F,G). (L and M) are adjacent serial sections of the same coated pit labeled with concanavalin A-horseradish peroxidase (con AHRP). The vesicular image in (L) (arrowhead) can be seen to be, in reality, a narrow-necked pit in connection with the surface in (M). (N and 0) show similar narrow-necked structures during the endocytosis of Con A-HRP. In these images, the narrow necks (arrows) connect the coated pits (arrowheads) to the cell surface. (A,B,C, x 140,000; D, x240,OOO; E,F,H,I,J,K,N,O, x90,OOO; G , X 138,000; L,M, X 180,000; bars = 0. I pm; lead citrate counterstain.)
62
MARK C. WILLINGHAM AND IRA PASTAN
location of clathrin. The only structures in the cell that showed significant amounts of clathrin morphologically were coated pits (Willingham et al., 1981b). A significant pool of diffusely distributed “soluble” clathrin could not be found (Willingham et al., 1981b; Wehland et al., 1981). A series of direct experiments have also been performed to investigate how uncoated vesicles form from coated pits. Using microinjection to introduce antibodies against clathrin into living cells, it could be shown that the antibody gained access to all of the cytosolic regions of the cell, and that it reacted with the exposed clathrin on the cytosol face of all pits (Wehland et al., 1981). However, precipitation of clathrin aggregates or isolated coated vesicles in the cytosol was not seen even after the antibody was present for many hours. In addition, the process of endocytosis from these coated pits proceeded normally in spite of their decoration by anticlathrin antibody. This antibody was active in the sense that it could precipitate soluble clathrin or isolated vesicles from an exogenous source when these were injected at a later time into cells previously injected with anticlathrin antibody. More recently, studies have appeared which utilized seriaf sections of surface coated pits in intact cells to investigate if images of coated vesicles are connected to the cell surface (Fan et al., 1982; Peterson and van Deurs, 1983). These authors have reported that from 10 to 50% of coated vesicular profiles in different cell types seemed to be isolated coated vesicles when adjacent serial sections were examined. One difficulty with such studies is that the density of membranes is very close to that of the surrounding cytoplasm, particularly when these membranes are seen in tangential section in very small tubular structures (<20 nm in diameter). This difficulty produces images that may not be unequivocally interpreted. There is no clear precedent that such small, tortuous structures can be visualized when analyzed at this high level of resolution by serial sections. The real question is whether the interpretation of these images is as clear cut as these studies suggest. We have undertaken similar studies, but have developed and employed a new high contrast preservation method for membranes. The results of those analyses show that all coated vesicular profiles in cultured fibroblasts can be shown to be connected to the cell surface (Willingham and Pastan, 1983a). These connections are often quite narrow (17 nm in diameter) and can be very long (>700 nm in length) and have tortuous shapes. We have compared our method with the preservation-contrast techniques used in previous studies and find that these other methods fail to preserve the narrow necks. We feel that this new morphologic study, taken with the previous evidence from immunocytochemical, impermeant marker, and microinjection studies, strongly argue that coated pits do not pinch off to form isolated coated vesicles during the endocytic event. How we think this step in the endocytosis process might occur will be discussed below.
ENDOCYTOSIS AND EXOCYTOSIS
63
2. Receptosomes When a,M is labeled with rhodamine, its uptake into living cells can be observed by fluorescence image intensification microscopy (Pastan et al., 1977; Willingham and Pastan, 1978, 1980, 1983e). The clustering of ligands on the surface of cells is difficult to see clearly because of the large amount of diffuse binding over the entire surface of the cell, and the quenching of the rhodamine signal that occurs when this directly labeled ligand is clustered in coated pits on the cell surface (Willingham et al., 1981a). However, after only a few minutes at 37°C the pattern of fluorescence changes, and small rapidly moving punctate dots appear within the cell (Pastan and Willingham, 1981b). These structures are vesicles which move by saltatory motion, a characteristic form of motion generally restricted to intracellular organelles such as mitochondria and lysosomes (Freed and Lebowitz, 1970; Rebhun, 1972; Wehland and Willingham, 1983; Willingham et al., 1979; Willingham and Yamada, 1978). These fluorescent structures are not visible by phase contrast microscopy, which indicates that they are either very small and/or they are not very dense. Electron microscopy shows that both of these predictions are correct (Willingham and Pastan, 1980). These organelles show the unusual property, in contrast to macropinosomal endocytic vesicles (Lewis, 1931; Straus, 1964), of not fusing with lysosomes when they approach them as they move along microtubule tracks used for saltatory motion. These structures are, therefore, special nonlysosomal vesicles which move by saltatory motion and which contain ligands that have entered cells via coated pits. They were named “receptosomes” to emphasize their role in the uptake of receptor-mediated ligands (Willingham and Pastan, 1980). Typical examples of the appearance of these vesicles are shown by electron microscopy in Fig. 3; their characteristics are listed in Table I. They have appeared in electron micrographs for many years, but it was not always known where they came from or what their destination might be. They have been variously identified as condensing vacuoles near the Golgi, tubulo-vesicles, endosomes, intermediate vacuoles, pinocytic vesicles, multivesicular bodies (presumed to be lysosomal), CURL, mature lysosomes, macropinosomes, phagocytic vacuoles, or simply as lucent vesicles or vacuoles (Abrahamson et al., 1979; Abrahamson and Rodewald, 1981; Ackerman and Wolken, 1980, 1981; Ackerman et al., 1983; Anderson et al., 1976, 1977, 1982; Bergeron et al., 1979; Bessis, 1963; Bleil and Bretscher, 1982; Bretscher et al., 1980; Brown et al., 1982, 1983; Carpentier et al., 1978, 1979a,b, 1982; Courtoy et al., 1982; Dales, 1973, 1978; DeDuve and Wattiaux, 1966; Fan et al., 1983; Farquhar, 1978, 1981, 1983; Franke et al., 1976; Friend and Farquhar, 1967; Geuze et al., 1983; Gonatas et al., 1977, 1980; Gorden et al., 1978a,b; Haigler et al.. 1978, 1979; Handley et al., 1981; Helenius et al., 1980; Helenius and Marsh, 1982; Herzog and Farquhar, 1977; Herzog and Reggio, 1980; Iacopetta et al., 1983;
64
MARK C. WILLINGHAM AND IRA PASTAN
FIG.3 . Morphologic appearance of receptosomes. Swiss 3T3 cells (A,B) or K B human carcinoma cells (C-Q) were fixed in glutaraldehyde and processed using ferrocyanide-reduced OTO (A,B), routine OsO, (C), or routine OsO, OTO (Willingham and Rutherford, 1984) (D-Q). Receptosomes were labeled with Con A-HRP (B), epidermal growth factor-horseradish peroxidase (C), or unlabeled (A,D-Q). The appearance of receptosomes initially after formation is shown in (A-E), in which they characteristically appear empty, and often contain a single intraluminal vesicular profile (small arrows), as well as a fuzzy decoration on a single straightened edge of their cytoplasmic
ENDOCYTOSIS AND EXOCYTOSIS
65
TABLE I PROPERTIES OF RECEPTOSOMtS Empty appearance by electron microscopy Move by saltatory motion Contain ligand derived from plasma membrane coated pits Selectively fuse with trans-reticular Golgi Do not directly fuse with lysosomes Do not contain lysosomal hydrolases Have an acidic pH (approx. pH 5.0) Have a smooth, continuous membrane in routine EM processing (lysosomes have a fragmented membrane) Often 10% of membrane is straightened and contains a fuzzy fibrillar material on its cytosolic face Often contain a single intraluminal vesicular structure (forms when receptosomes form) Fuse with other receptosomes creating multivesicular receptosomes The isolated vesicle has an intermediate bouyant density between plasma membrane and lysosomes Often contain ligands and receptors
Joseph et al., 1978, 1979; Marsh and Helenius, 1980; M a t h et al., 1982; McGookey et al., 1983; Michaels and Leblond, 1976; Nagura et al., 1979; Nicolson, 1974; Novikoff et al., 1980; Orci et al., 1978; Ottosen et al., 1980; Posner et al., 1981; Quintart et al., 1983; Rodewald, 1973; Rodewald and Abrahamson, 1980; Ryser et al., 1982; Salisbury et al., 1980; Silverstein et al., 1976; Simons et al., 1982; Steinman et al., 1983; Straus, 1964; Wall et al., 1980). In many cases the relationship of these structures to endocytosis was not recognized. In any random section of cells fixed from 37"C, the frequency with which images of receptosomes appear is low; this probably reflects their short life span in cells (often less than 10 minutes). This, plus their relatively nondescript appearance, probably led to the failure to describe them as specific organelles. surface (arrowheads). This f u z z y decoration has a substructure which appears lamellar and periodic, shown at higher magnification in (G,H). With OTO fixation, one can also see the densities of lipoprotein particles contained within these receptosomes (small arrowheads-F,I,K,M,O), presumably representing low-density lipoproteins taken up from the culture medium. At later times, receptosomes fuse with each other, generating larger receptosomes that contain multiple intraluminal vesicular profiles (small arrows-F,I,J,K) and multiple fuzzy edges (large arrows-H,I,K,M). A commonly observed fixation artifact in these vesicles (caused by glutaraldehyde) is the inward blistering on the membrane, producing a C-shaped profile (large arrows-L-Q). The artifactual nature of these blisters is evident from the empty appearance of the central regions of the blisters, indicating a lack of cytosol matrix material and, therefore, formation after the cytoplasmic matrix had been stabilized by glutaraldehyde fixation. One of these blisters is shown in cross section in (N), making it appear to be a separate vesicle contained within the receptosome. (A-F, 1-Q, X90,OOO: G,H, X 180,000; bars = 0.1 pm; lead citrate counterstain.)
66
MARK C. WILLINGHAM AND IRA PASTAN
Ligands which bind to specific receptors can usually be detected concentrated in coated pits prior to entry into receptosomes, but as discussed above other substances which do not concentrate in pits also enter the cell through this common endocytic pathway. For example, horseradish peroxidase does not bind to specific receptors on the cell surface of fibroblasts and is not found concentrated in open coated pits. Yet 10 minutes after cells are incubated with HRP the ligand is found in typical receptosomes (Fig. 3) (Ryser et al., 1982; Willingham and Pastan, unpublished data). Because, as far as is known, receptosomes only form from coated pits, this experiment indicates HRP enters cells through coated pits, but in an unconcentrated form.
111. Endocytic Vesicles: Characteristics
A. RECEPTOSOMES The general properties of receptosomes and their identification as vesicles derived from coated pits at the plasma membrane were described in the previous section. In this section, the morphologic characteristics of endocytic organelles will be discussed. Receptosomes can be described as clear vesicles, since they generally have very little electron-dense material in their interior (Fig. 3). Their small size and low density make them invisible by light microscopic methods that depend on refractile resolution, such as phase contrast microscopy. They generally contain a very small single intralumenal vesicle when they first form from coated pits. It is possible that some of these intralumenal vesicles represent images of infoldings of the receptosomal membrane seen in tangential section. Receptosomes have been observed to fuse with each other in living cells when observed by video intensification microscopy (Willingham and Pastan, 1983b). As a result, at later stages they can become multivesicular (Fig. 3 ) and represent one class of vesicles that have been termed “multivesicular bodies.” (However, the term multivesicular body is a “catch-all” phrase used to describe many different organelles.) In some cultured cells, such as very flat fibroblasts, there is a considerable separation of the peripheral cytoplasmic surface from the cell center where most of the elements of the Golgi system are located. In such cells, receptosomes that form from coated pits in the peripheral parts of the cell may travel 10-100 p m to fuse with elements of the Golgi. It is in such cells (such as Swiss 3T3 cells) that receptosomes are easiest to visualize and study, since their life span can be from 15 to 60 minutes depending on the part of the cell from which they originate. On the other hand, in more compact or rounded cells, such as KB cells or CHO cells, there is often a very short distance (<1 pm) from the coated pits at the cell surface to the Golgi reticular elements. In fact, in electron microscopic studies both the plasma membrane coated pits (150 nm) and the Golgi coated pits (70 nm) are often found in the same microscopic field (Fig. 4).
ENDOCYTOSIS AND EXOCYTOSIS
67
In these cells, the life span of the receptosome appears to be much shorter, with virtually all of the internalized ligand appearing in Golgi and/or lysosomal elements within 15 minutes. Many epithelial cell types in vivo have a compact or rounded shape. Thus, if such cells are examined 15-20 minutes after ligand internalization, one would completely miss the coated pit-receptosome stages of internalization. Some investigators have referred to receptosomes as “endosomes” (Marsh and Helenius, 1980; Simons er al., 1982). Other investigators have referred to images of receptosomes, particularly those fused with trans-reticular Golgi structures, as “CURL” (Geuze et al., 1983) to denote them as a compartment where uncoupling of receptor and ligand occurs. The structures most often shown as “CURL” are probably trans-reticular Golgi elements. It seems more likely that dissociation of ligands from receptors occurs in the low pH environment of the receptosome (Tycko and Maxfield, 1982), not the Golgi trans elements. Receptosomes often show a straightened portion of membrane at one edge or, after fusion with other receptosomes, at more than one edge (Fig. 3). These straight membrane segments are decorated on their cytoplasmic face by a fuzzy, lamellar material which, by immunocytochemistry, reacts poorly with antibodies to actin, clathrin, myosin, tubulin, or fodrin (Willingham and Pastan, 1980; M. C. Willingham, C. B. Klee, and I. H. Pastan, unpublished data). The exact nature of this fuzzy material is not yet known, but its presence has led some investigators to describe these structures as “partially coated” vesicles. In fact, two different structures are encompassed by this term. One structure is a tangential image of what may be a transition structure generated when a receptosome is formed from a coated pit. In this case, the coat material is clathrin. The other structure is the “dense plaque” regions on a receptosome or on a lysosome (Fig. 3). It is not clear if this dense plaque material is composed of clathrin in a modified or relatively inaccessible form, but the periodic nature of these structures might suggest a relationship to clathrin. The frequency of these dense plaques is also not clear. It is possible that these regions occur frequently and are related to surface clathrin; it is also possible that they are infrequent and are related to clathrin derived from the Golgi system. Receptosomes also show other regions of “fuzzy” material on their cytoplasmic surface that are morphologically different from these dense plaques. An understanding of the exact nature of these fuzzy regions clearly will require further study. B. MULTIVESICULAR BODIES The characteristic morphology of receptosomes was not recognized in earlier studies. Often, structures resembling receptosomes were identified as some other cell organelle, most often as lysosomes. A special form of lysosome has been described in some cell types which has a lower density matrix than typical dense lysosomes, and usually contains many small vesicular structures (Friend and
68
MARK C. WILLINGHAM AND IRA PASTAN
FIG.4. Morphologic appearance of the coated pits of the Golgi in comparison with those at the plasma membrane. KR human carcinoma cells were incubated with epidermal growth factor conjugated to horseradish peroxidase (EGF-HRP) and warmed to 37°C for 10-15 minutes prior to fixation. At this time, EGF-HRP has progressed by endocytosis from the plasma membrane, through the coated pits of the cell surface, to receptosomes, and into the trans-reticular Golgi elements that
ENDOCYTOSIS AND EXOCYTOSIS
69
Farquhar, 1967). These have been called “multivesicular bodies.” Cytochemically, these structures often contain lysosomal enzymes. However, these types of multivesicular bodies can morphologically resemble receptosomes that have fused and contain more than one small vesicle. An important point is that receptosomes do not contain lysosomal enzymes and, therefore, ligands contained within them are not degraded until they are delivered to lysosomes. Receptosomes are not lysosomal or degradative organelles !
C . LYSOSOMESAND PRELYSOSOMES As pointed out above, the multivesicular type of lysosome can easily be confused on morphologic grounds with receptosomes in some cell types. Since it was recognized that lysosomes could have varied appearances, the term “prelysosome’’ was used to describe a vesicle (receptosome) that would somehow develop into a lysosome or directly deliver its contents to lysosomes. There are a number of problems with the prelysosomal concept. One is that direct visualization of receptosomes in living cells fails to show them fusing with lysosomes. Another is that the ligand contained in receptosomes is transferred from them to the tubules of the trans-reticular Golgi and only later is found in lysosomes. A third difficulty with the prelysosomal concept is that not all ligands that enter receptosomes get transferred to lysosomes. The best example of such a ligand is transferrin, which enters cells by coated pits and receptosomes (Willingham et al., 1984), but leaves the cell at a postreceptosomal level within a very short time (4-8 minutes); it does not accumulate in lysosomes, or undergo significant degradation prior to exocytosis (Klausner et al., 1983). Further, many viruses and toxins enter via the coated pit-receptosome system, but their functionally important destination is the cytosol rather than lysosomes. Such studies show that not every ligand that enters cells necessarily ends up in lysosomes, and points out the difficulty of calling a receptosome a “prelysosomal organelle.” As mentioned above, experiments with large macromolecular ligands, such as colloidal gold and virus particles, occasionally lead to the generation of large receptosome-derived organelles that have retained some of the fuzzy cytoplasmic material associated with receptosomes, but also appear to contain lysosomal material. We think that it is possible that these structures may be induced to form by the presence of these unusually large bodies (colloidal gold or viruses), and are not normally found in most cells. It is not clear that degradation takes place in contain the smaller sized coated pits of the Golgi. This ligand concentrates in these structures, serving as a marker for these small coated pits. In (A), note the proximity of the plasma membrane coated pits (arrowhead) to the Golgi coated pits (small arrow) adjacent to a receptosome (R). (B) also shows this close relationship of plasma membrane coated pits (arrowheads) to the smaller Golgi pits that contain ligand at this time (small arrows). (C and D) show images of these labeled Golgi coated pits (small arrows) near a labeled receptosome (R) in (C) and at higher magnification in (D). (A-C, x90,OOO; D, x 180,000; bars = 0. I pm; lead citrate counterstain.)
70
MARK C. WILLINGHAM AND IRA PASTAN
such organelles, but on morphologic grounds they could be called “lysosomes.” In this unusual case, receptosomes may b e in some sense “prelysosomal” organelles.
D. TUBULARELEMENTS In some studies of endocytosis, a major population of tubular or tubulovesicular elements have been seen to contain ligand shortly after ligand internalization (Ackerman and Wolken, 1980, 1981; Ackerman et al., 1983; Geuze et al., 1983; Willingham et al., 1984). These endocytic structures are seen most often in cuboidal cells in which the Golgi system is close to the cell surface. Most of these structures probably represent the tubular-reticular elements of the Golgi, the site in the Golgi system that contains the Golgi population of coated pits (see below) and receives ligands from receptosomes. It is also possible that some of these images represent the formation process of receptosomes or the elongated necks of plasma membrane coated pits sometimes seen during receptosome formation. In addition, in certain regions of very flattened cells, receptosomes lying between large microfilament bundles sometimes appear quite elongated or tubular. Based on the evidence so far available there seems to be no compelling reason to assume that these images represent different structures from those seen in cells where receptosomes are more easily seen.
E. C-SHAPEDVESICLES Frequently, glutaraldehyde fixation produces an unusual alteration in receptosomes which resembles the blistering seen during glutaraldehyde fixation at the plasma membrane (Shelton and Mowczko, 1977, 1978, 1979). This phenomenon can lead to an inward blistering of the membrane of a receptosome into its empty center (Fig. 3). The shape of the membrane of this rounded organelle is transformed into a C-shaped structure (Fig. 3). A clear indication of the artifactual nature of this phenomenon is the dense proteinaceous matrix of the cytosol which is preserved in the first few seconds of glutaraldehyde fixation. The space formed between the inwardly curved membrane of the blister and the cytosol matrix is empty, indicating that it formed after the cytosol was initially fixed into a matrix by glutaraldehyde. By analogy with the surface blisters, the blister into the center of the receptosome probably takes a number of minutes to form, and would probably be minimized by shorter fixation times.
F. MACRO-AND MICROPINOCYTOSIS The process of surface ruffling and macropinosome formation was first described by Lewis in 1931 and called “pinocytosis” (Lewis, 1931). This is a
ENDOCYTOSIS AND EXOCYTOSIS
71
prominent activity in some cell types such as activated macrophages, but it is usually of minor significance in flattened cells in culture. Epidermal growth factor (EGF) induces ruffling in A431 cells and this effect has complicated the analysis of EGF uptake in A431 cells. In this cell type EGF induces a single, slow wave of ruffling that causes the cell to take in some of its surface membrane by macropinocytosis (Chinkers et al., 1979; Haigler et al., 1979; Willingham et al., 1983a). Some EGF is internalized by this process. However, these cells also have coated pits, and most EGF is probably internalized by coated pits and receptosomes in a normal fashion (Willingham et al., 1983a). Since the ruffling membrane is not necessarily selective as to what part of the surface it takes in, some coated pits themselves may be incorporated into macropinosomes. A similar phenomenon might explain the appearance of coated regions in phagocytic membrane (Aggeler and Werb, 1982), although some specific coated pit-associated ligands may participate in this process. Thus, intracellular vesicles derived from other forms of endocytosis may, in a nonspecific way, have clathrin-coated portions of membrane. A major difference between receptosomes derived from single coated pits and these large phagocytic vesicles often found in phagocytic cells is the destination of the material within. Receptosomes deliver their contents directly to Golgi elements and not to lysosomes, whereas the other types of vesicles can be seen directly fusing with mature lysosomes in the cytoplasm (Willingham and Yamada, 1978). Another nonspecific form of endocytosis, micropinocytosis, has been postulated to occur from small flask-shaped invaginations of the plasma membrane called caveolae (Bruns and Palade, 1968; Palade et al., 1979; Simionescu et al., 1975). Experiments using generalized colloidal impermeant markers of the cell surface such as ruthenium red have shown that most of these small structures, even those tangentially sectioned and therefore appearing to be isolated vesicles near the membrane, are in continuity with the cell surface (Bungaard, 1983; Chien et al., 1982; Frokjaer-Jensen, 1980; M. C. Willingham, unpublished data). When soluble markers such as horseradish peroxidase are introduced into the incubation medium, very little appears to enter the cell through these structures within the short time span when large amounts are found in receptosomes. This suggests that the caveolae, if they participate in endocytosis in these types of cultured cells, do so at a very slow rate. Both macropinocytosis (Lewis’s pinocytosis) and micropinocytosis would be expected to be able to internalize markers that are strictly in the fluid phase of the medium with no binding at all on the membrane itself. Fluid phase uptake does not appear to be a quantitatively important pathway in cultured fibroblastic or epithelial cells. As discussed previously, some caution must be exercised in interpretation of pathways of uptake of so-called “fluid-phase’’ markers, since peroxidase, a prototype of this class, is also taken up by the coated pit-receptosome system, albeit in a nonconcentrative fashion. It has already been suggested that endocytosis via coated pits
72
MARK C. WILLINGHAM AND IRA PASTAN
may account for a large percentage of the uptake of fluid phase markers (Marsh and Helenius, 1980; Steinman et al., 1983).
IV. Kinetic Classes of Endocytosis Phagocytosis, macropinocytosis, and micropinocytosis are endocytic processes which do not involve coated pits on the cell surface. They have varying kinetics depending on the cell type and the stimulus, but in general probably are relatively slow events, usually requiring many minutes to take in materials into endocytic vesicles. Coated pits mediate the endocytosis of some receptor-bound ligands that is very rapid (seconds), but also can be responsible for endocytosis with slower kinetics: (1) Materials that bind to receptors that cluster and concentrate in coated pits have internalization rates with a half time of 1-3 minutes at low enough receptor occupancy to allow one cycle of endocytosis to internalize the surface-bound material (Willingham and Pastan, 1982). (2) Some materials may bind to the cell surface on specific receptors but not to receptors that participate in the clustering process in coated pits. These materials will be internalized through coated pits, but much more slowly, perhaps requiring 30-60 minutes to completely clear the cell surface. Examples of these materials are adenovirus (FitzGerald et al., I983a) and interferon (Zoon et al., 1983). The rate of endocytosis of these molecules probably reflects the turnover of the entire plasma membrane through coated pits. (3) Some ligands bind to specific receptors which appear to be prevented from entering coated pits. An example of such receptors is the hCG (LH) receptor which remains on the cell surface for many hours (Ahmed et al., 1981). Thus, it is important to realize that while the coated pit system mediates a concentrative form of endocytosis, it also can mediate nonconcentrative absorptive events. In this way, the coated pit can serve as the entrance site for the entire plasma membrane and probably accounts for the overall turnover of membrane.
V. Compartmentalization in the Golgi A . ENDOCYTOSIS
1 . Endocytic Ligands Destined for Lysosomes Most of the work defining the coated pit-receptosome pathway has involved the use of ligands that eventually were destined for lysosomes, such as LDL, a,M, EGF, asialoglycoproteins, and lysosomal enzymes. All of these ligands enter via coated pits, move to receptosomes, and then are transferred to the transreticular elements of the Golgi probably by fusion of receptosomes with this
ENDOCYTOSIS AND EXOCYTOSIS
73
structure. In some cells these Golgi elements are located in the perinuclear Golgi region, but in other cells they are found extending throughout the cytoplasm, with only a slightly higher concentration near the Golgi stacks. These transreticular elements are possibly the structures originally described by Golgi using light microscopy (Golgi, 1898); the cisternal stacks were only identified many years later by electron microscopy (Dalton and Felix, 1954). The fact that the Golgi has tubules as well as stacks is usually not recognized because the typical introduction to Golgi morphology emphasizes the stack structures. Further, much more is known about the passage of secretory proteins through the stacks than about the functions of the trans-reticular elements. Even SO, the transreticular elements may represent the majority of the membranes of the Golgi system. This portion of the Golgi was highlighted by Novikoff and co-workers (Novikoff and Novikoff, 1977; Novikoff et al., 1971, 1980) as a region of the Golgi which had high levels of cytochemically detectable acid phosphatase reactivity (Goldfischer, 1982). Their studies were carried out in cells that contained relatively large amounts of an acid phosphatase activity that could be detected using a substrate such as cytidine monophosphate to produce an insoluble lead phosphate reaction product in the presence of lead nitrate. The exact biochemical nature of the enzyme(s) responsible for this activity is still not clear. The part of the Golgi containing this type of acid phosphatase activity was given the name “GERL,” since this type of acid phosphatase was also found in the endoplasmic reticulum and in the lysosomes of the same cell. The presumption was that this represented a normal lysosomal enzyme that could be detected at its site of synthesis (in the endoplasmic reticulum) and on its way through the Golgi (in the GERL) to lysosomes. For some reason, many cultured cells contain very little, if any, of this cytochemical enzyme marker. One of the hallmark morphologic features of the trans-reticular (TR) portion of the Golgi is the presence of small clathrin-coated pits (Friend and Farquhar, 1967; Novikoff etal., 1980). These structures are often seen in cross section and are called, therefore, Golgi-coated vesicles, but whether these structures are really isolated vesicles or instead are like the large coated pits of the surface connected to the tubules of the trans-Golgi is still under investigation. The function of these small (70 nm) coated pits is currently under study. One group of studies based on fractionation of cell homogenates suggested that a secretory protein, the G protein of vesicular stomatitis virus, might selectively be associated with the small type of Golgi coated pits during its secretion (Rothman et al., 1980). In previous morphologic studies, secretory proteins were detected in coated structures at the cell surface, but these were large structures (Franke et al., 1976; Rodewald, 1973). These images were interpreted to be coated vesicles involved in secretion. However, another interpretation of these images is that after the secretory material was released into the extracellular space in high concentration, some became bound to cell surface coated pits.
74
MARK C. WILLINGHAM AND IRA PASTAN
When these pits were sectioned tangentially and their connection to the surface was masked, the images gave the appearance of isolated coated vesicles containing secretory material. In recent studies in which the coated regions of the transGolgi were directly examined in situ by electron microscopic immunocytochemistry for the presence of G protein, none was found (Wehland et al., 1982a,b). Instead, large amounts were present in adjacent trans-reticular elements (Wehland et al., 1982a; Levine et al., 1981). A different role for the coated pits of the Golgi has emerged from studies on endocytosis. Gonatas and co-workers (Gonatas et a l . , 1977, 1980; Joseph et al., 1978, 1979) have studied the endocytosis of lectins and found them in the TR Golgi and in lysosomes, suggesting that the TR Golgi may be involved in transfer of endocytosed materials to lysosomes. Novikoff and co-workers (Novikoff et al., 1980) identified acid phosphatase in these same elements (GERL), indicateing they are part of a system that delivers lysosomal enzymes to lysosomes. The role of the TR Golgi, and particularly the coated pits of the TR Golgi, became clearer when kinetic studies on EGF uptake were performed (Willingham and Pastan, 1982; Willingham et d . , 1983a). High concentrations of recently endocytosed epidermal growth factor was found to accumulate in this type of coated pit 12 minutes after cellular entry and just prior to delivery to lysosomes (Willingham and Pastan, 1982; Willingham et al., 1983a). These findings suggested that the coated pits of the Golgi might mirror in function those at the cell surface by accumulating or clustering ligands in some sort of receptorspecific process in the Golgi coated pits. Rather than delivering the ligand to receptosomes, such as occurs at the cell surface, the ligand is delivered to newly forming lysosomes. Such a mechanism could explain the fact that some ligands do not get delivered to lysosomes, in spite of the fact that they are present in the trans-reticular tubules. For example, EGF is found in the trans-reticular Golgi tubules and concentrated in the coated pits of the Golgi, whereas transfenin is also found in the trans-reticular Golgi tubules, but is not concentrated in coated pits of the Golgi (Willingham et al., 1984). EGF is delivered to lysosomes, whereas transferrin is returned back to the cell surface (Klausner et al., 1983). Further, the phosphomannosyl receptor that mediates the compartmentalization of lysosomal enzymes through the common mannose 6-phosphate recognition marker on these enzymes was localized to Golgi coated pits, as well as other parts of the trans-reticular Golgi (Willingham et al., 1983~).Thus, the Golgi coated pits may be a morphologic structure responsible for the physical compartmentalization of materials destined for lysosomes from the Golgi. The trans-reticular elements of the Golgi with their coated pits can extend far out into the cytoplasm (Fig. 4) and often lie close to the inner surface of the microfilament mat under the plasma membrane. The large difference in size between plasma membrane and Golgi coated pits is most evident in these areas, but it can also lead to confusion as to the type of coated structure being exam-
ENDOCYTOSIS AND EXOCYTOSIS
75
ined. In some cell types, the surface coated pits can be 100-120 nm in diameter. This proximity of Golgi and plasma membrane pits might suggest that there is transfer of one type of coated structure to another (perhaps by fission of the larger coat to form two smaller coats). However, to date there is no direct evidence to suggest that the two coated pit systems have any rapid interchange. 2. Ligands Destined for Exocytosis
SOfar, the major example discussed of a ligand that enters by coated pits and recycles back to the cell surface rapidly is transferrin. However, data suggest that a somewhat similar mechanism may exist as part of the secretion of polymeric IgA in hepatocytes (Courtoy et al., 1982; Quintart et al., 1983). This ligand is endocytosed by coated pits at the plasma membrane of the apical surface of liver cells in the same vesicle as asialoglycoproteins, but is segregated inside the cell, probably in the reticular Golgi elements, and secreted at the lateral cell surface. Another interesting example is the endocytosis and secretion of maternal IgG by neonatal gut cells (Abrahamson et al., 1979; Abrahamson and Rodewald, 1981; Rodewald, 1973; Rodewald and Abrahamson, 1980). This process of “transcytosis” results in the delivery of intact proteins from one side of a cell to another cell surface in these polarized epithelial cells. In the type of cultured cells used for the transferrin and EGF uptake studies, there is no apparent polarization, and the transferrin is simply delivered out of the cell back into the medium. 3. Receptor Recycling The amount of ligand taken in by cells is very large. In quantitative experiments, fibroblasts in culture have been found to internalize over 2 x lo6 a,M molecules per hour (Dickson et al., 1981b). The number of receptors on the cell to accomplish this task is usually only a few hundred thousand. The cell does not have the capacity to resynthesize new receptors at a fast enough rate to replace those internalized from the cell surface. Thus, the receptors brought in from the cell surface must be recycled to the surface free of ligand to be used again. Such recycling is commonly seen in many ligand systems and must be a general ability of most cells (reviewed in Brown el al., 1983). Since various receptors have now been shown to be internalized with the ligand into receptosomes, separation of receptor from ligand must occur in the cell prior to introducing the receptor into an organelle that can physically carry it back to the surface. In the case of transferrin both the ligand and receptor recycle to the surface. Receptosomes have been shown to have a pH around 5.0, similar to lysosomes (Tycko and Maxfield, 1982). The dissociation of many ligands from receptors is known to be facilitated by low pH. Thus, ligand separation from receptor probably occurs in receptosomes prior to their fusion with the trans-reticular Golgi elements, where the pH may rise to allow other types of associations to occur for recompartmentalization of ligands. Interestingly, transferrin does not separate from its receptor
76
MARK C . WILLINGHAM AND IRA PASTAN
at low pH, and this probably explains its different behavior from ligands destined for lysosomes (Klausner e f al., 1983; Dautry-Varsat et al., 1983). One might imagine that the same pathway used to transport transferrin to the cell surface might be the pathway that receptors use to recycle (Willingham et al., 1984). As yet, there is no clear evidence as to what this pathway might be, but it is likely that it is the same as the normal exocytic pathway used by other materials destined for exocytosis from the Golgi.
B. EXOCYTOSIS: MATERIALS DESTINEDFOR
THE
CELLSURFACE
1. Exocytosis in Undifferentiated Cells Autoradiographic studies of newly synthesized proteins that are destined for secretion show them to be concentrated in the endoplasmic reticulum, followed by elements of the Golgi stacked cisternae. Usually a larger amount is seen in the trans-most elements, suggesting a delay at this stage (Palade, 1975, 1982). The route by which these materials then travel depends on the cell type and the type of material. In cultured fibroblastic and epithelial cells, the exocytic pathway has not yet been clarified. These cells do not form or contain the classical secretory granules seen in specialized cells of the exocrine and endocrine systems. A protein in the lumen of the Golgi, such as fibronectin, appears to be secreted continuously out of the cell, with no obvious intermediate concentrative granule stage (Hedman, 1980; Yamada et al., 1980). The same is true for integral membrane proteins such as the G protein of vesicular stomatitis virus. In addition to the trans-most cisternae of the stacks, G protein is found in the trans-reticular system (Wehland et al., 1982a). Some of these trans-reticular elements (Goldfischer, 1982) have been referred to by other names such as GERL (Novikoff et al., 1980) or CURL (Geuze et al., 1983). Since ligands such as tranferrin, which are rapidly exocytosed after entry in the trans-reticular Golgi, appear to be in this same compartment and cannot be found in the Golgi stacks during this transit (Willingham et al., 1984), one can speculate that constitutive exocytic materials such as fibronectin might also be exocytosed from the trans-reticular elements. This would place the trans-reticular Golgi as a major traffic control point for the shuttling of macromolecules. It receives materials from both endocytosis and the cistemal stacks, and directs these components either to lysosomes (many endocytosed materials and newly synthesized lysosomal enzymes arriving from the endoplasmic reticulum via the Golgi stacks) or to the exocytic pathway (some newly internalized receptors, ligands such as transferrin, and newly synthesized secretory or membrane proteins). Physical separation might occur in the coated pits of the Golgi by some type of receptor clustering process. Transfer to the lysosomes would require trapping or clustering in these pits; transfer to the cell surface would not. Cytochemical experiments have suggested a mechanism for sorting. It will be important to have biochemical confirmation of these processes.
ENDOCYTOSIS AND EXOCYTOSIS
77
2. Exocytosis by Granule Secretion The process of exocytosis of synthesized materials was originally visualized in cells that form specialized secretory granules, usually secretion that is induced by a specific signal, such as a hormone response (reviewed in Farquhar and Palade, 1981; Jamieson and Palade, 1977; Palade, 1975). These endocrine and exocrine tissues are specialized cell types. Indeed, myeloma and plasma cells that secrete huge quantities of immunoglobulins do not form these types of granules. In spite of the presence of large concentrations of secretory material in their endoplasmic reticulum and Golgi, and despite high rates of synthesis and secretion, the intermediate secretory organelles have remained elusive (Ottosen et al., 1980). Because the frequency of observing images of such organelles depends in large part on the life span of the organelle, it is likely that these secretory events are extremely rapid, much more rapid than the processing steps within the endoplasmic reticulum and Golgi. The acinar cells of the pancreas or the parotid gland produce large granules containing high concentrations of secretory enzymes, contents that they release in response to hormonal stimulation in rapid exocytic events. Similarly, the pancreatic beta cell stores insulin for rapid, quanta1 release in response to changes in blood glucose levels. The endocrine cells of the anterior pituitary are classic examples of long-term storage organelles (granules) which can rapidly release their contents in response to specific releasing hormones. But most secretory processes in other less highly specialized cells are of a continuous, constitutive nature. In addition, it is possible that even these specialized granule-producing cells have a second form of exocytosis for continuous membrane renewal and metabolic recycling. The morphological identification of this major exocytic pathway is of great interest. DELIVERY C. LYSOSOMAL Two classes of materials that are known to enter lysosomes are materials derived from endocytosis (as discussed above) and materials delivered to the lysosome directly from the Golgi; these include lysosomal enzymes and matrix proteins. In addition, materials such as proteins from the cytosol may get into lysosomes for degradation. The mechanism of degradation of cytosol proteins is not clear. Some evidence suggests that they are degraded in lysosomes, while other data suggest degradation occurs in the cytosol (Bigelow et al., 1981; Hershko and Ciechanover, 1982; Stacy and Alfrey, 1977). Although some lysosomal enzymes may be secreted and reabsorbed by cells, for the most part, newly synthesized lysosomal enzymes are delivered to lysosomes from the Golgi without undergoing exocytosis. Since these materials are synthesized within the endoplasmic reticulum and processed through the Golgi, the Golgi must have a way of directing these lysosomal constituents directly to lysosomes without external secretion. In morphologic studies of the uptake of lysosomal enzymes,
78
MARK C. WILLINGHAM A N D IRA PASTAN
P-galactosidase was found to accumulate in high concentrations in the coated pits of the trans-reticular Golgi (Willingham et al., 1981~).This is also the region where increased acid phosphatase reactivity is found (Novikoff er al., 1980). These data, as well as other studies discussed above, suggested that these coated structures might in some way be involved in directing proteins to lysosomes. The mechanism of compartmentation probably involves receptor binding and targeting, since most of these systems have specific receptors with which they interact, such as the phosphomannosyl receptor system for lysosomal enzymes which contain mannose 6-phosphate moieties (Kaplan et al., 1977; Neufeld and Ashwell, 1979; Sando and Neufeld, 1977). Thus, the trans-reticular portion of the Golgi system has the capacity to direct recently endocytosed materials or newly synthesized materials into lysosomes. It also has the ability to direct other materials to the exocytic pathway such as has been shown for the delivery to the cell surface of VSV G protein and transferrin.
D. MATERIALS FROM
THE
ENDOPLASMIC RETICULUM
It is likely that in most cells proteins made in the endoplasmic reticulum are always delivered to the Golgi system. Once in the Golgi (the cis face), these materials pass across the Golgi stacked cisternae, in many cases gaining carbohydrate moieties, and are efficiently delivered to the trans face (reviewed in Rothman, 1981). There the trans-reticular elements, which appear to comprise the bulk of the Golgi membranes, somehow direct the flow of these materials to either exocytic pathways or to lysosomes. As pointed out above, some of the steps in the control of intracellular traffic involve specific receptors. Usually, targeting is very efficient. For example, in the case of G protein of VSV, virtually all of this trans-membrane protein is delivered to the cell surface; very little is delivered to the lysosomal system (Wehland et al., 1982a; Bergmann ef al., 1981; Levine et al., 1981). One interesting abnormality in the targeting of lysosomal enzymes occurs in I-cell disease. In this genetic disorder cells fail to process lysosomal enzymes properly so that mannose-6-P04 is not available for interaction with the phosphomannosyl receptor and the abnormal enzymes are secreted instead of being delivered to lysosomes (Neufeld and Ashwell, 1979; Neufeld et al., 1975; Sly et al., 1981; Sly and Stahl, 1978). We have shown that the phosphomannosyl receptor is concentrated in coated pits of the Golgi and suggested that lysosomal enzymes are bound to this receptor and concentrated in this portion of the Golgi just prior to their delivery to lysosomes. Substances not bound to specific receptors should not accumulate in Golgi coated pits, but instead somehow be directed to the cell surface. If the protein is embedded in a membrane, it will become a component of the plasma membrane; if it is a soluble protein in the lumen of the Golgi, it will be secreted.
ENDOCYTOSIS AND EXOCYTOSIS
79
VI. Functions of Endocytic and Exocytic Membrane Traffic OF RECEPTORS AND LIGANDS,DEC' rRADATION A. RECYCLING
If receptors are recycled, the internalization of ligands bound to these receptors demands that the internal cellular membrane machinery somehow separate these components so that the receptor can be reutilized and the ligand degraded. One mechanism available to the cell to perform this function is a change in pH. One special feature of receptosomes is that they have a low pH (Tycko and Maxfield, 1982). When measured directly in a living cell, the pH of receptosomes approaches the acid pH of lysosomes (pH 5.0). The pH in the transreticular Golgi elements is probably closer to neutrality (preliminary experiments directly measuring this compartment using transferrin have suggested a pH near 6.5 or higher, F. R. Maxfield, personal communication). Many ligand-receptor interactions are sensitive to low pH, and in many of the cases studied (e.g., asialoglycoproteins) the ligands are delivered to lysosomes, while the receptor appears to be recycled to the cell surface. Presumably, these ligands dissociate from their receptors in the receptosome while being carried from the cell surface to the Golgi. In the Golgi, sorting is completed and the ligand is sent to lysosomes and the receptor back to the surface. An exception is the transferrin system, in which the receptor-ligand interaction is not destabilized by low pH; in this system the ligand and receptor are both recycled back to the surface without ligand degradation (Dautry-Varsat et al., 1983; Klausner et al., 1983; Willingham et al., 1984). Since the transferrin system delivers iron to the cell which is necessary for cell growth (iron which is released in the low pH environment of receptosomes), this property which avoids degradation of transferrin allows the protein and the receptor to be reutilized. The degradation of internalized materials could serve some specific function, such as creation of biologically active proteolytic fragments which have some specific intracellular role. Degradation also has a nutritional role such as occurs in the release of cholesterol from LDL. Some endocytosed materials, such as the protease inhibitor a,-macroglobulin, could have a regulatory effect on the lysosomal system itself by limiting the activity of lysosomal proteases. The internalization of asialoglycoproteins is the most striking demonstration of internalization for the purpose of degradation, since the delivery of this material from the circulation to liver cells is almost certainly for degradative purposes (Ashwell and Morell, 1974; Courtoy et al., 1982; Geuze et al., 1983; Quintarat et al., 1983; Wall et al., 1980). However, in the very same hepatocytes, the uptake of IgA follows the same initial pathway but is diverted to be secreted into the bile (Courtoy et al., 1982; Orlans et al., 1979; Quintart et al., 1983). The uptake of IgG from neonatal gut also occurs through coated pits and receptosomes, but is
80
MARK C. WILLINGHAM AND IRA PASTAN
not delivered to lysosomes; it is secreted intact (Rodewald and Abrahamson, 1980). In the case of IgA and IgG, the cells involved are polar and their polarity results in a transfer (transcytosis) of the materials from one side of the cell to another. The ability of polarized epithelia to perform such a function has also been studied in cells synthesizing viral glycoproteins; different viral products are delivered to different surfaces of a polarized cultured cell type depending on the nature of the virus (Rodriquez-Boulan and Pendergast, 1980). B. TRAFFICCONTROL
OF SYNTHESIZED MATERIALS
In '1898, Camillo Golgi described an elaborate array of tubular elements visible with special light microscopic procedures (Golgi, 1898). It was not until the advent of electron microscopy that it was fully appreciated that this system was involved in the transport and processing of proteins synthesized in the endoplasmic reticulum (reviewed in Palade, 1975). The clearest demonstration of the existence of such a pathway was in those cell types in which the synthesized material was morphologically visible, particularly those systems in which large secretory granules were formed. Recent work on endocytosis has shown that the Golgi also has a role in the processing and handling of materials derived from endocytosis. The part of the Golgi that deals with endocytosed materials is the large trans-reticular network that possibly corresponds to the structure originally described by Camillo Golgi. Granule forms of secretion are limited to special cell types that store materials and rapidly release them in response to specific stimuli. Most cells also continuously secrete (exocytose) material without the use of concentrative secretory granules. The Golgi system, composed in large part of reticular elements, plays a key role in the process. It received these materials from two sources: endocytosis and internal synthesis in the endoplasmic reticulum. It delivers materials to two destinations: lysosomes or the cell exterior. With the exception of certain unusual conditions, these pathways appear to be the only way known for substances to enter and leave the Golgi system. The endoplasmic reticulum and Golgi are functionally in communication and the nuclear envelope is part of the endoplasmic reticulum. Since components of the plasma membrane are continually derived from and delivered to the Golgi system, and lysosomes are similarly related to the Golgi, it follows that all these membrane systems of the cell interact together, and this occurs as their components move through a traffic control center in the Gotgi. The one membrane system in animal cells that is an exception is the mitochondrion and this may be because that organelle was originally derived by a parasitic mechanism involving incorporation of bacteria. The description of the pathways of endocytosis and exocytosis is based on experiments in which large amounts of ligand have been traced sequentially from
ENDOCYTOSIS AND EXOCYTOSIS
81
one compartment to another. Because these compartments are functionally related, small amounts of trace substances can be found in locations that do not reflect major pathways. A number of interesting questions about how sorting in the Golgi occurs are currently under active study. Some of these are listed below: How does the Golgi stack system sequester and concentrate enzymes involved in glycosylation? How do the Golgi membranes get reorganized into stacks and reticular tubules after each mitosis? What are the signals in the trans-Golgi that direct ligands and proteins to lysosomes or exocytosis? What are the structures that mediate constitutive (nonsecretory granule) exocytosis and how are they directed to the cell surface? What is the role of the small coated pits of the transreticular Golgi, and why are they smaller than those at the plasma membrane? (Do they have a different function?) How do lysosomes avoid incorporating material destined for exycotosis? C. PATHOLOGIC PROCESSES
The ubiquitous distribution of the coated pit endocytic pathway in most animal cells [and some evidence indicates its existence even in higher plants (Mersey et al., 1983; Van der Valk and Fowke, 1981)] suggests both that the pathway was developed early in evolution and that, having been in existence for so long, pathogenic systems have had ample opportunity to take advantage of it. Two notable pathologic processes have already been described which utilize this pathway: intoxication by bacterial or plant toxins and infection by viruses. The existence of toxin receptors and their specific interactions with target cells has been clearly established (Donovan et al., 1981; FitzGerald et al., 1980; Iglewski et al., 1977; Neville and Chang, 1978; Nicolson, 1974; Pappenheimer, 1977; Sandvig and Olsnes, 1980; Yamaizumi et al., 1978). Toxins that have been actively studied include diphtheria toxin, tetanus toxin, ricin toxin, and Pseudomonas exotoxin. These toxins bind to specific receptors on appropriate cell types and somehow enter to produce their effects on target cells Pseudomonas exotoxin has been shown to enter cells through the coated pit pathway, and inhibition of this entry prevents toxicity (FitzGerald et al., 1980). Diphtheria toxin and Pseudornonas exotoxin act by inactivation of elongation factor 2 . This leads to inhibition of protein synthesis and eventual cell death. Only a single toxin molecule is theoretically necessary to kill each cell because of the catalytic nature of the reaction (Yamaizumi et al., 1978). One problem these toxins have is that the target they interact with is located in the cell cytosol. Therefore, they must cross a membrane to reach their site of action. Some toxin can cross membranes efficiently when the pH is lowered (Sandvig and Olsnes, 1980). Since both receptosomes and lysosomes are part of the coated pit endocytic pathway, and since both have a pH below 5.0, the entry of toxins through the
82
MARK C. WILLINGHAM AND IRA PASTAN
coated pit system is a practical mechanism for certain toxins to gain access to an acidic intracellular compartment. By using the receptosome, which contains no significant amounts of hydrolases, the toxins can achieve an appropriate environment for membrane translocation. Since only a few molecules/cell are necessary for toxicity, the entry into receptosome need not necessarily be through the concentrative receptor route, but could occur as the entire plasma membrane is internalized through coated pits as part of overall membrane turnover. Thus, some toxins could, if necessary, bind to ubiquitous carbohydrate or lipid determinants, instead of specific protein receptors which are present in much smaller amounts, and still eventually enter the cell in sufficient amounts to be toxic. Generally, toxins have a cell binding region that is separated from the catalytically active site. In diphtheria toxin the molecule is split by a protease into two subunits, whereas with Pseudomonas exotoxin no processing or nicking occurs (Iglewski et al., 1977). The exact mechanism of translocation across a membrane is not fully understood, but in some cases the low pH environment appears to alter the conformation of the molecule in a way that allows it to penetrate the membrane and reach the cytosol. In some cases, it may be necessary for the toxin to be bound to the proper receptor for this event to occur, but some translocation seems to be possible even if the toxin is brought into the cell on a different receptor. This has been observed with Pseudomonas toxin conjugated to epidermal growth factor (FitzGerald et at. , 1983a,b), in which case the conjugation of the toxin has destroyed its ability to interact with its own receptor, so that entry can occur only through the EGF receptor. Another pathologic process that often involves the coated pit pathway is the entry of viruses into cells; this includes both membrane-limited, as well as naked viruses. Dales and co-workers showed some time ago that it was common to find particles of many different viruses in surface coated pits, and subsequently in intracellular vesicles which we now recognize as receptosomes; later, viruses appear in lysosomes (Chardonnet and Dales, 1970; Dales, 1973; Dales and Choppin, 1962; Dales and Hanafusa, 1972; Simpson et al., 1969). With our present knowledge of the fate of materials that are present in coated pits, we can predict that these particles will enter the cell in receptosomes and, if nothing intervenes, they will eventually be delivered to lysosomes. Helenius and coworkers have examined the entry of Semliki Forest virus and saw it enter through coated pits and appear in vesicles they termed endosomes (receptosomes). Exit from this vesicle occurs by fusion of the SFV membrane with the receptosome with release of the nucleocapsid into the cytosol (Helenius et al., 1980; Helenius and Marsh, 1982; Marsh and Helenius, 1980; M a t h et al., 1982; Simons et al., 1982). In parallel experiments, it was shown that the infection of cells by this and other membrane-limited viruses required a low pH environment; fusion of the viral membrane with host cell membrane was stimulated below pH 6.0. Probably many membrane-limited viruses, including vesicular stomatitis virus, enter the
ENDOCYTOSIS AND EXOCYTOSIS
83
cell cytosol by fusing with the membrane of the receptosome (Dickson et al., 1981c; M a t h et al., 1982; Schlegel et al., 1983). The entry of nonencapsulated viruses into the cytosol probably occurs in a different manner. Adenovirus, for example, enters cells through the coated pitreceptosome system, but since it lacks a membrane, it cannot fuse with cell membranes. Dales and co-workers showed that intact adenovirus particles could be found in the cytosol (Chardonnet and Dales, 1970). Therefore, the virus particles have a mechanism for exit from a membrane-limited organelle into the cytosol without gross alterations in the structure of the virus. Recently, this exit process has been studied in more detail, and it appears that adenovirus has the capacity to lyse the receptosome in which it is contained (FitzGerald et al., 1983a). As a consequence, other materials contained within the receptosome are also released into the cytosol in the process of lysis. If a toxin such as Pseudomonus toxin (or a toxin-EGF conjugate) is added to cells with the virus, the presence of the virus in the same receptosome enhances the release and subsequent toxicity of the toxin up to 10,000-fold (FitzGerald et al., 1983a,b). The release process appears to be dependent on the low pH in the receptosome. It probably is catalyzed by a protein in the capsid because UV-inactivated virus that is unable to replicate is still active (Seth et al., 1983). The exact mechanism by which the virus lyses the vesicle is still unclear. However, this phenomenon provides a way of introducing materials into the cytosol, and may be useful in enhancing the delivery of drugs and toxins to the cytosol of neoplastic cells. D. HORMONAL DELIVERY? Early in the studies of endocytosis, investigators speculated whether such an entry mechanism could be used by cells to deliver hormones to receptors deep within the cell. It is now clear that many polypeptide hormones such as insulin, triiodothyronine, epidermal growth factor, MSH, and NGF enter cells by receptor-mediated endocytosis (Bergeron et al., 1979; Carpentier et al., 1978, 1979a,b, 1982; Cheng et al., 1980; Fan et al., 1983; Gorden et al., 1979a,b; Haigler et a l . , 1978, 1979; Horiuchi et al., 1982, 1983; Levi et al., 1980; Maxfield et al., 1978; Orci et al., 1978; Posner et al., 1981; Varga et al., 1976; Willingham et al., 1983a; Willingham and Pastan, 1982). One still unanswered question is whether this entry is necessary for the trophic actions of these hormones. For most of the systems studied so far, there is no compelling evidence to indicate that hormone internalization by endocytosis is needed for hormone action, but a likely candidate is triiodothyronine (TJ. Although the majority of the cellular receptors for T, lie on the plasma membrane, some are found in the nuclear fraction (Cheng et al., 1980; Horiuchi et al., 1982, 1983) and these ndclear receptors are believed to be necessary for thyroid hormone action (Oppenheimer, 1979; Samuels and Tsai, 1973). The presence of labeled T, or T,
84
MARK C. WILLINGHAM AND IRA PASTAN
receptors in nuclear fractions is thought to represent specific intranuclear binding to chromatin, but could also be due to binding to other structures present in the nuclear fraction. The evidence to date does not lend major support to the concept that the action of polypeptide hormones is mediated by internalization in vesicles. It appears likely that the endocytosis system plays a major role in the regulation of the availability of the receptors for these hormones. Internalized receptors can either be destroyed in lysosomes or recycled to the cell surface. One interesting way in which vesicular transport may play a role in hormone action is in the stimulation of glucose transport by insulin (Cushman and Wardzala, 1980; Lienhard, 1983; Wheeler et al., 1982). Within a few minutes of insulin addition, membrane glucose carriers appear to be transported from an intracellular pool to the cell surface resulting in increased glucose transport. Further, the disappearance of carriers from the surface and the subsequent decrease in glucose transport following insulin removal seems to follow the time course expected for the endocytic pathway through coated pits. Thus, it may be possible that hormone-induced events that involve plasma membrane functions such as nutrient transport are a result of membrane-mediated vesicular traffic events.
E. OTHERFUNCTIONS The cell surface is often exposed to a changing extracellular environment which can contain proteases, lipases, toxins, and other materials that are able to digest or modify surface components. Constant and selective internalization and replacement enable cells to maintain a functionally intact cell surface. This process is undoubtedly vital to survival; for this reason it is unlikely that major mutations in the endocyticlexocytic systems discussed here could be tolerated by cells. Most work on exocytosis has concentrated on specialized cells that exocytose materials in concentrated secretory granules on the application of an external stimulus. It is almost certain that this is not the same mechanism that cells use for constitutive exocytosis and replacement of the cell surface. Continuous cell lines in culture provide a good system to study this process because few of them make secretory granules: they contain the basic undifferentiated structures necessary for cell survival without superimposed special differentiated functions. The organelles present in such cells must carry out all the functions necessary for cell survival and growth. The transfer and mixing of the membranous compartments of the cell are of central importance in cell function. Some proteins move rapidly from one compartment to another; others move much more slowly. The factors that control the rates by which these processes occur are still unclear. Recent advances in the
ENDOCYTOSIS AND EXOCYTOSIS
85
identification of the organelles involved in these processes should make such studies possible.
VII. Summary Animal cells have specific pathways to transport macromolecules from their surrounding environment to their interior, and from internal compartments to the cell surface or other intracellular locations, Many of these movements appear to be receptor-dependent processes in which specific membrane receptors bind macromolecules, segregate them into discrete membrane-limited compartments, and move the molecules to new locations. Such processes include the clustering and internalization of receptor-bound ligands at the cell surface in clathrin-coated pits, the formation of endocytic vesicles (receptosomes) from coated pits, the movement of receptosomes by saltatory motion to the Golgi system, the concentration of materials in the coated pits of the Golgi system that are destined for delivery to lysosomes, and the directed traffic of materials destined for exocytosis out of the Golgi to the cell surface. This review describes some of the experiments which have led to our current understanding of the various organelles involved in this traffic and some of the biochemical mechanisms involved.
ACKNOWLEDGMENTS The authors wish to express their thanks to their many co-workers involved in this field of research over the last few years, particularly Drs. D. J. P. FitzGerald, S.-y. Cheng, N. D. Richert, J. A. Hanover, R. B. Dickson, L. Beguinot, C. R. Schlegel, M. Ruff, J. Wehland, A. Levitzki, F. R. Maxfield, J. Schlessinger, J. H. Keen, P. J. A. Davies, H. T. Haigler, M. M. Gottesman, G. G. Sahagian, E. F. Neufeld, L. C. Smith, D. P. Via, A. LeCam, J.-C. Nicolas, R. Padmanabhan, P. Seth, R. Horiuchi, H. Arnheiter, and K. Zoon. In addition, the authors especially acknowledge the invaluable expert technical assistance of A. V. Rutherford, M. G. Gallo, and S. S. Yamada.
REFERENCES Abrahamson, D. R., and Rodewald, R. (1981). J. Cell Biol. 91, 270. Abrahamson, D. R., Powers, A , , and Rodewald, R. (1979). Science 206,567. Ackerman, G. A , , and Wolken, K. W. (1980). IRCS Med. Sci. 8, 842. Ackerman, G. A,, and Wolken, K. W. (1981). J . Hisrochem. Cytochem. 29, 1137. Ackerman, G. A,, Yang, J., and Wolken, K. W. (1983). J . Hisrochem. Cytochem. 31, 433 Adam, C. J . , Maurey, K. M., and Storrie, B. (1982). J . Cell Biol. 93, 632. Aggeler, J . , and Werb, Z. (1982). J . Cell Biol. 94, 61 3 .
86
MARK C . WILLINGHAM AND IRA PASTAN
Ahmed, C. E., Sawyer, H. R . , and Niswender, G. D. (1981). Endocrinology 109, 1380. Anderson, E. (1964). J . Cell B i d . 20, 131. Anderson, R. G. W . , Brown, M. S . , and Goldstein, J. L. (1976). Proc. Natl. Acad. Sci. U.S.A. 73, 2434. Anderson, R. G . W., Brown, M. S., and Goldstein, J. L. (1977). Cell 10, 351. Anderson, R. G. W., Goldstein, J. L., and Brown, M. S. (1980). J . Receptor Res. 1, 17. Andersun, R. G. W., Brown, M. S . , Beisiegel, U . , and Goldstein, J. L. (1982). J . Cell Biol. 93, 523. Ashford, T. P., and Porter, K. R. (1962). Int. Congr. Electron Microsc., 5th 2, LL4. Ashwell, G., and Morell, A. G. (1974). Adv. Enzymol. 41, 99. Axelrod, D., Koppel, D. E., Schlessinger, J., Elson, E. L., and Webb, W. W. (1976). Biophvs. J . 16, 1055. Bergeron, J. J. M., Sikstrom, R., Hand, A. R., and Posner, B. 1. (1979). J . Cell Biol. 80, 427. Bergmann, J . E., Tokuyasu, K. T., and Singer, S. J. (1981). Proc. Natl. Acad. Sci. U.S.A. 78, 1746. Bessis, M. (1963). Harvey Lect. Ser. 58, 125. Bessis, M. C., and Breton-Gorius, J. (1962). Blood 19, 635. Bigelow, S . , Hough, R., and Rechsteiner, M. (1981). Cell 25, 83. Blanchette-Mackie, E. J., and Scow, R. 0. (1982). Anat. Rec. 203, 205. Bleil, J. D., and Bretscher, M. S. (1982). EMBO J . 1, 351. Bowers, B. (1964). Protoplasma 59, 351. Bretscher, M . S . (1983). Proc. Nut/. Acad. Sci. U.S.A. 80, 454. Bretscher, M. S., and Thomson, J. N. (1983). EMBO J . 2, 599. Bretscher, M. S., Thomson, J. N., and Pearse, B. M. F. (1980). Proc. Natl. Acud. Sci. U.S.A. 77, 4156. Brightman, M. W. (1962). Anat. Rec. 142, 219. Brightman, M. W., and Palay, S. L. (1963). J . Cell Biol. 19, 415. Brodsky, F. M., and Parham, P. (1982). J . Cell B i d . 95, 433a. Brown, M. S., Anderson, R. G. W . , Basu, S. K., and Goldstein, J . L. (1982). Cold Spring Harbor Symp. Quunt. Biol. 46, 713. Brown, M. S., Anderson, R. G. W., and Goldstein, J . L. (1983). Cell 32, 663. Bruns, R. R., and Palade, G . E. (1968). J . Cell Biol. 37, 277. Bungaard, M. (1983). Fed. Proc. Fed. Am. Soc. Exp. Biol. 42, 2425. Carpenter, G., and Cohen, S. (1976). J . Cell Biol. 71, 159. Carpentier, J.-L., Gorden, P., Amherdt, M., van Obberghen, E . , Kahn, C. R . , and Orci, L . (1978). J . Clin.Invest. 61, 1056. Carpentier, J.-L., Gorden, P., Barazzone, P., Freychet, P., LeCam, A., and Orci, L. (1 979a). Pmc. Natl. Acad. Sci. U.S.A. 76, 2803. Carpentier, J.-L., Gorden, P., Freychet, P., and LeCarn, A. (1979b). J . Clin. bwesr. 63, 1249. Carpentier, J.-L., Gorden, P., Anderson, R. G . W., Goldstein, J. L., Brown, M. S., Cohen, S., and Orci, L. (1982). J . Cell B i d . 95, 73. Chardonnet, Y . , and Dales, S. (1970). Virologv 40, 462. Cheng, S.-y., Maxfield, F. R . , Robbins, J . , Willingham, M. C . , and Pastan, 1. H . (1980). Proc. Natl. Acad. Sci. U.S.A. 77, 3425. Chien, S . , Laufer, L., and Handley, D. A. (1982). J . U/rrustruer. Res. 79, 198. Chinkers, M., McKanna, J. A., and Cohen, S. (1979). J . Cell Biol. 83, 260. Conn, P. M., Conti, M., Harwood, J. P., Dufau, M. L., and Catt, K. J. (1978). Nature (London) 274, 598. Courtoy, P. J., Quintart, J., and Baudhuin, P. (1982). J . Cell Biol. 95, 424a. Crowther, R. A , , and Pearse, B. M. F. (1981). J . Cell Biol. 91, 790.
ENDOCYTOSIS AND EXOCYTOSIS
87
Crowther, R. A,, Finch, J. T., and Pearse, B. M. F. (1976). J . Mol. Biol. 103, 785. Cushman, S . W., and Wardzala, L. J. (1980). 1.B i d . Chem. 255, 4758. Dales, S . (1973). Bucteriol. Rev. 37, 103. Dales, S. (1978). In “Transport of Macromolecules in Cellular Systems” (S. C. Silverstein, ed.), pp. 47-68. Dahlem Konferenzen, Berlin. Dales, S . , and Choppin, P. W. (1962). Int. Congr. Electron Microsc., 5th 9. Dales, S . , and Hanafusa, H. (1972). Virology 50, 440. Dalton, A . J., and Felix, M. D. (1954). Am. 1. Anat. 94, 171. Dautry-Varsat, A , , Ciechanover, A . , and Lodish, H. F. (1983). Proc. Nutl. Acud. Sci. U.S.A. 80, 2258. Davies, P. J. A , , Davies, D. R., Levitski, A . , Maxfield, F. R., Milhaud, P., Willingham, M. C., and Pastan, I. H. (1980). Nature (London) 283, 162. DeDuve, C., and Wattiaux, R. (1966). Annu. Rev. Physiol. 28, 435. Dickson, R. B., Nicolas, J.-C., Willingham, M. C., and Pastan, 1. (1981a). Exp. CellRes. 132, 488. Dickson, R. B., Willingham, M. C . , and Pastan, I. (1981b). J . Biol. Chem. 256, 3454. Dickson, R. B., Willingham, M. C., and Pastan, I . (1981~).J . Cell B i d . 89, 29. Dickson, R. B., Schlegel, C. R . , Willingham, M. C., and Pastan, I. (1982a). J. Cell. Physiol. 113, 353. Dickson, R. B., Schlegel, C. R., Willingham, M. C., and Pastan, I. (1982b). Exp. Cell Res. 140, 215. Dickson, R. B., Schlegel, R., Willingham, M. C., and Pastan, I. (1982~).Exp. CellRes. 142, 127. Dickson, R. B., Willingham, M. C., and Pastan, I. H. (1982d). Ann. N.Y. Acud. Sci. 401, 38. Dickson, R. B., Beguinot, L., Hanover, J. A . , Richert, N. D., Willingham, M. C., and Pastan, I. (1983). Proc. Natl. Acad. Sci. U.S.A. 80, 5335. Donovan, J. J., Simon, M. I . , Draper, R. K . , and Montal, M. (1981). Proc. Natl. Acad. Sci. 78, 172. Edidin, M., Zagyansky, Y . , and Lardner, T. J. (1976). Science 191, 466. Fan, J. Y . , Carpentier, J.-L., Gorden, P., van Obberghen, E. V., Blackett, N. M . , Grunfeld, C., and Orci, L. (1982). Proc. Nail. Acud. Sci. U.S.A. 79, 7788. Fan, J. Y., Carpentier, J.-L., van Obberghen, E., Blackett, N. M., Grunfeld, C . , Gorden, P., and Orci, L. (1983). J . Histochem. Cvrochem. 31, 859. Farquhar, M. G . (1978). J. Cell Biol. 77, R35. Farquhar, M. G . (1981). Mefhods Cell Biol. 23, 399. Farquhar, M. G . (1983). Fed. Proc. Fed. Am. Soc. Exp. Biol. 42, 2407. Farquhar, M. G . , and Palade, G . E. (1981). J . Cell Eiol. 91, 77s. Fawcett, D. W. (1964). Anat. Rec. 148, 370. Fine, R. E., Blitz, A. L., and Sack, D. H. (1978). FEBS Lett. 94, 59. FitzGerdld, D. J . P., Morris, R . E., and Saelinger, C. B . (1980). Cell 21, 867. FitzCerald, D. J. P., Padmanabhan, R . , Pastan, I., and Willingham, M. C. (1983a). Cell 32, 607. FitzCerald, D. J. P., Trowbridge, 1. S . , Pastan, I . , and Willingham, M. C. (1983b). Proc. Nut/. Acud. Sci. U.S.A. 80, 4134. Franke, W. W., Luder, M . , Kartenbeck, J., Zerban, H., and Keenan, T. W. (1976). J . CellBiol. 69, 173. Freed, J. J., and Lebowitz, M. M. (1970). J . Cell B i d . 45, 334. Friend, D. S . , and Farquhar, M. G . (1967). J . Cell B i d . 35, 357. Frokjaer-Jensen, J. (1980). J . Ultrustrucr. Res. 73, 9. Geuze, H. J., Slot, J. W . , Strous, G . J . A. M . , Lodish, H . F., and Schwartz, A . L. (1983). Cell 32, 277. Goldfine, 1. D., Smith, G . J., Wong, K. Y., and Jones, A . L. (1977). Pruc. Nut/.Acad. Sci. U.S.A. 74, 1368.
88
MARK C. WILLINGHAM AND IRA PASTAN
Goldfischer, S. (1982). J . Histochem. Cytochem. 30, 717. Goldstein, J. L . , Anderson, R. G . W., and Brown, M. S. (1979). Nature (London) 279, 679. Golgi, C. (1898). Arch. Itul. Biol. 30, 60. Gonatas, J., Steiber, S . , Olsnes, S . . and Gonatas, N. K. (1980). J . Cell Biol. 87, 579. Gonatas, N. K., Kim, S. U., Steiber, A., and Avrameas, S . (1977). J . Cell Biol. 73, 1. Gorden, P., Carpentier, J. L., Cohen, S . , and Orci, L. (1978a). Proc. Natl. Acad. Sci. U.S.A. 75, 5025. Gorden, P., Carpentier, J.-L., Freychet, P. LeCam, A., and Orci, L. (1978b). Science 200, 782. Gray, E. G . (1961). J . Anat. 95, 345. Gray, E. G., (1972). J . Neurocytol. 1, 363. Haigler, H., Ash, J. F., Singer, S . J., and Cohen, S . (1978). Proc. Natl. Acad. Sci. U.S.A. 75, 3317. Haigler, H. T., McKanna, J. A., and Cohen, S . (1979). J . Cell B i d . 81, 382. Haigler, H. T., Maxfield, F. R., Willingham, M. C., and Pastan, I. (1980a). J . B i d . Chem. 255, 1239. Haigler, H. T., Willingham, M. C., and Pastan, I. (1980b). Biochem. Biophvs. Res. Commun. 94, 630. Handley, D. A., Arbeeny, C. M., White, L. D., and Chien, S . (1981). Proc. Nutl. Acad. Sci. U.S.A. 78, 368. Hanover, J. A,, Cheng, S.-y., Willingham, M. C., and Pastan, I. H. (1983). J . Biol. Chem. 258, 370. Hanover, J . A., Willingham, M. C . , and Pastan, I. H. (1984). Ann. N.Y. Acad. Sci. 421, 410. Hedman, K. (1980). J . Histoehem. Cytochem. 28, 1233. Heimburger, N. (1979). In “Proteases and Hormones” (M. K. Agarwall, ed.), pp. 44-65. Elsevier, Amsterdam. Helenius, A,, and Marsh, M. (1982). Ciba Found. Symp. 92, 59. Helenius, A,, Kartenbeck, J., Simons, K., and Fries, E. (1980). J . Cell B i d . 84, 404. Hershko, A,, and Ciechanover, A. (1982). Annu. Rev. Biochem. 51, 335. Herzog, V., and Fdrquhar, M. G . (1977). Proc. Natl. Acad. Sci. U.S.A. 74, 5073. Herzog, V., and Reggio, H. (1980). Eur. J . Cell Biol. 21, 11. Heuser, J., and Evans, L. (1980). J . Cell B i d . 84, 560. Hillman, G . M . , and Schlessinger, J. (1982). Biochemistry 21, 1667. Horiuchi, R., Cheng, S . - y . , Willingham, M. C . , and Pastan, 1. (1982). J . Biol. Chem. 257, 3139. Horiuchi, R., Johnson, M. L., Willingham, M. C., Pastan, I . , and Cheng, S.-y. (1983). Proc. Natl. Acad. Sci. U.S.A. 79, 5527. Iacopetta, B . J., Morgan, E. V., and Yeoh, G. C. T. (1983). J . Hisrochem. Cytochem. 31, 336. Iglewski, B . H . , Liu, P. V . , and Kabat, D. (1977). Infect. frnmun. 15, 138. Jamieson, J. D., and Palade, G . E. (1977). fnt. Cell B i d . pp. 308-317. Jacobson, K., Wu, E., and Poste, G. (1976). Biochim. Biophys. Acta 433, 215. Joseph, K. C., Kim, S. U., Steiber, A., and Gonatas, N. K . (1978). Proc. Natl. Acad. Sci. U.S.A. 15, 2815. Joseph, K. C., Steiber, A., and Gonatas, N . K. (1979). J . Cell Biol. 81, 543. Kahn, C. R. (1976). J . Cell Biol. 70, 261. Kanaseki, T., and Kadota, K. (1969). J . Cell Biol. 42, 202. Kaplan, A., Achord, D. T., and Sly, W. S . (1977). Proc. Natl. Acad. Sci. U.S.A. 74, 2026. Kaplan, J., and Nielsen, M. L. (1979). J . Biol. Chem. 254, 7323. Keen, I. H., Willingham, M. C., and Pastan, I. H. (1979). Cell 16, 303. Keen, J . H . , Willingham, M . C . , and Pastan. I . (1981). J . B i d . Chem. 256, 2538. King, A. C., and Cuatrecasas, P. (1981). N . Engl. J . Med. 305, 77. Kirchhausen, T., and Harrison, S . C. (1981). Cell 23, 755.
ENDOCYTOSIS AND EXOCYTOSIS
89
Klausner, R. D., Ashwell, G., van Renswoude, J., Harford, J. B., and Bridges, K. R. (1983). Proc. Natl. Acad. Sci. U.S.A. 80, 2263. Levi, A,, Shechter, Y., Neufeld, E. J., and Schlessinger, J . (1980). Proc. Natl. Acud. Sci. U.S.A. 71, 3469. Levine, G., Kotwal, G., Hamilton, M. G., Rachubinski, R., Bilan, P., Shore, G. C., Ghosh, H., and Bergeron, J. J. M. (1981). J . Cell Eiol. 91, 413a. Levitski, A,, Willingham, M. C., and Pastan, I. (1980). Proc. Nutl. Acad. Sci. U.S.A. 77, 2706. Lewis, W. H. (1931). Bull. Johns Hopkins Hosp. 49, 17. Lienhard, G. E. (1983). Trends Eiochem. Sci. 8, 125. Marsh, M., and Helenius, A. (1980). J . Mol. Biol. 142, 439. M a t h , K. S., Reggio, H., Helenius, A,, and Simons, K. (1982). J . Mol. Biol. 156, 609. Mato, J. M., Pencev, D., Vasanthakumar, G., Schiffmann, E., and Pastan, I. (1983). Proc. Natl. Acud. Sci. U.S.A. 80, 1929. Maxfield, F. R., Schlessinger, J., Schecter, Y., Pastan, I., and Willingham, M. C. (1978). Cell 14, 805. Maxfield, F. R., Willingham, M. C., Davies, P. J. A,, and Pastan, I. (1979a). Nurure (London) 277, 661. Maxfield, F. R., Willingham, M. C., Schlessinger, J., Davies, P. J. A , , and Pastan, I. (1979b). Cold Spring Harbor Conf. Cell Prolif. 6, 159-166. Maxfield, F. R., Willingham, M. C., Haigler, H. T., Dragsten, P., and Pastan, I. (1981b). Biochemistry 20, 5353. Maxfield, F. R., Cheng, S.-y., Dragsten, P., Willingham, M. C . , and Pastan, I. (1981a). Science 211, 63. McGookey, D. J . , Fagerberg, K . , and Anderson, R. G . W. (1983). J . Cell Eiol. 96, 1273. Mersey, B. G., Fowke, L. C., Constabel, F . , and Newcomb, E. H. (1983). Exp. Cell Res. (in press). Michaels, J. E., and Leblond, C. P. (1976). J . Microsc. Biol. Cell 25, 243. Nagura, H., Nakane, P. K . , and Brown, W. R. (1979). J . immunol. 123, 2359. Neufeld, E. F., Lim, T. W., and Shapiro, L. J. (1975). Annu. Rev. Eiochem. 44, 357. Neufeld, E., and Ashwell, G. (1979). i n “Biochemistry of Glycoproteins and Proteoglycans” (W. Lennarz, ed.), pp. 241-266. Plenum, New York. Neville, D. M., and Chang, T.-M. (1978). Curr. Top. Membr. Trunsp. 10, 65. Nicolson, G. L. (1974). Nature (London) 251, 628. Novikoff, A. B., and Novikoff, P. M. (1977). Hisrochem. J . 9, 525. Novikoff, P. M., Novikoff, A. B., Quintana, N., and Hauw, J. J. (1971). J . Cell Biol. 50, 859. Novikoff, A. B., Novikoff, P. M., Rosen, 0. M., and Rubin, C. S. (1980). J . Cell Eiol. 87, 180. Oppenheimer, J. H. (1979). Science 203, 971. Orci, L., Perrelet, A., and Gorden, P. (1978). Recent f r o g . Horm. Res. 34, 95. Orlans, E., Peppard, J . , Fry, J. F., Hinton, R . H . , and Mullock, B. M. (1979). J . Exp. Med. 150, 1577. Ottosen, P. D., Courtoy, P. J., and Farquhar, M. G. (1980). J . Exp. Med. 152, I . Palade, G. E. (1975). Science 189, 347. Palade, G. E. (1982). Ciba Found. Symp. 92, I . Palade, G. E., Simionescu, M., and Simionescu, N. (1979). Acta Physiol. S c a d . Suppl. 463, 11. Pappenheimer, A. M. (1977). Annu. Rev. Biochem. 46, 60. Pastan, I. H . , and Willingham, M. C. (1981a). Annu. Rev. Physiol. 43, 239. Pastan, I. H., and Willingham, M. C. (1981b). Science 214, 504. Pastan, I. H., and Willingham, M. C. (1983). Trends Eiochem. Sci. 8, 250. Pastan, I., Willingham, M. C., Anderson, W. B., and Gallo, M. (1977). Cell 12, 609. Pastan, I . , Haigler, H. T . , Dickson, R. B., Cheng, S.-y., and Willingham, M. C. (1981). Miami Winter Symp. Ser. 18, 137-148.
90
MARK C. WILLINGHAM AND IRA PASTAN
Pearse, B. M. F. (1975). J . Mol. Biol. 97, 93. Pearse, B. M. F. (1976). Proc. Natl. Acad. Sci. U.S.A. 73, 1255. Pearse, B. M. F. (1978). J. Mu/. B i d . 126, 803. Pease, B. M. F., and Bretscher, M. S. (1981). Annu. Rev. Biuchem. 50, 85. Petersen, 0. W., and van Deurs, B. (1983). J. Cell Biol. 96, 277. Policard, A., and Bessis, M. (1958). C. R. Acad Sci. 246, 3194. Posner, B. I., Bergeron, J. J. M., Josefsberg, Z., Khan, M. N., Khan, R. J . , Patel, B. A , , Sikstrom, A., and Verma, A. K. (1981). Recent frog. Horm. Res. 37, 539. Pricer, W. E., and Ashwell, G. (1971). J . Biol. Chem. 246, 4825. Quintart, J., Courtoy, P. J., Lemet, J . N., and Baudhuin, P. (1983). Eur. J . Biochem. 131, 105. Rebhun, L. I. (1972). In?. Rev. Cytol. 32, 93. Rodewald, R. (1973). J . CellBiul. 58, 189. Rodewald, R., and Abrahamson, D. R. (1980). Eur. J . Cell Biol. 22, 178. Rodriquez-Boulan, E., and Pendergast, M. (1980). Cell 20, 45. Roth, T. F., and Porter, K. R. (1962). Inr. Congr. Elecrron Microsc. 5th 2, LL4. Roth, T. F., and Porter, K. R. (1964). 1.Cell Biol. 20, 313. Roth, T. F., Cutting, J. A., and Atlas, S. B. (1976). J. Sicprarnol. Struct. 4, 527. Rothman, J. E. (1981). Science 213, 1212. Rothman, J . , Bursztyn-Pettigrew, H., and Fine, R. E. (1980). J. Cell B i d . 86, 162. Rouiller, Ch., Jezequel, A.-M. (1963). In “The Liver” (Ch. Rouiller, ed.), Vol. I , pp. 225, 233. Academic Press, New York. Ryser, H. J.-P., Drummond, I . , and Shen, W.-C. (1982). J. Cell. Physiol. 113, 167. Salisbury, J. L., Condeelis, J . S.,and Satir, P. (1980). J . Cell B i d . 87, 132. Samuels, H. H., and Tsai, J. S. (1973). Proc. NatI. Acad. Sci. U.S.A. 70, 3488. Sando, G. N., and Neufeld, E. F. (1977). Cell 12, 619. Sandvig, K., and Olsnes, S. (1980). J. Cell Biol. 87, 828. Schlegel, C. R., Dickson, R. B., Willingham, M. C., and Pastan, I . (1982a). Proc. Natl. Acad. Sci. U.S.A. 79, 2291. Schlegel, R., Willingham, M. C., and Pastan, I. H. (1982b). J. Virol. 43, 871. Schlegel, R., Tralka, T. S., Willingham, M. C., and Pastan, I. (1983). Cell 32, 639. Schlessinger, J . , Schecter, Y . , Cuatrecasas, P., Willingham, M. C., and Pastan, I. (1978). Proc. Natl. Acad. Sci. U.S.A. 75, 5353. Schook, W., Puszkin, S., Bloom, W., Ores, E., and Kochwa, S. (1979). Proc. Natl. Acad. Sci. U.S.A. 76, 116. Seth, P., FitzGerald, D. J. P., Willingham, M. C., and Pastan, I. (1983). J . Cell Biol. 97, I IOa. Shaklai, M . , and Tavassoli, M. (1982). J . Hislochem. Cvtochem. 30, 1325. Shelton, E . , and Mowczko, W. E. (1977). J . Cell B i d . 72, 206a. Shelton, E . , and Mowczko, W. E. (1978). Scanning 1, 166. Shelton, E., and Mowczko, W. €5. (1979). I n “Freeze Fracture: Methods, Artifacts, and Interpretations” (J. E. Rash and C. S. Hudson, eds.), pp. 67-69. Raven, New York. Silverstein, S. C . , Steinman, R. M., and Cohn, Z. A . (1976). Annu. Rev. Biochern. 46, 669. Simionescu, N., Simionescu, M., and Palade, G. E. (1975). J. Cell B i d . 90, 605. Simons, K . , Garoff, H., and Helenius, A. (1982). Sci. Am. 246, 58. Simpson, R. W., Hauser, R. E., and Dales, S. (1969). Virology 37, 285. Sly, W. S., and Stahl, P. (1978). I n “Transport of Molecules in Cellular Systems’’ (S.C. Silverstein, ed,), pp. 229-244. Dahlem Konferenzen, Berlin. Sly, W. S., Natowicz, M., Gonzalez-Noriega, A,, Grubb, J. H., and Fischer, H. D. (1981). I n “Lysosomes and Lysosomal Storage Diseases” (J. W. Callahan and 3 . A. Lowden, eds.), pp. 131-146. Raven, New York. Stacy, D. W., and Alfrey, V. G. (1977). J . Cell B i d . 75, 807.
ENDOCYTOSIS AND EXOCYTOSIS
91
Steinman, R. M., Mellman, I. S . , Muller, W. A., and Cohn, Z. A. (1983). J . Cell Biol. 96, 1. Stome, B. (1979). Exp. CellRes. 118, 135. Straus, W. (1964). J . Cell B i d . 20, 497. Tycko, B., and Maxfield, F. R. (1982). Cell 28, 6 4 3 . Ungewickell, E., and Branton, D. (1981). Nature (London) 239, 420. Van der Valk, P., and Fowke, L. C. (1981). Can. J . Bot. 59, 1307. Van Leuven, F., Cassiman, J. J . , and van den Berghe, H. (1978). Exp. Cell Res. 117, 273. Varga, J. M., Moellmann, G . , Fritsch, P., Godawska, E., and Lerner, A. B. (1976). Proc. Natl. Acad. Sci. U.S.A. 73, 559. Vermeer, B. J., Havekes, L., Wijsman, M. C . , and Emeis, J. J. (1980). E.xp. Cell Res. 129, 201. Via, D. P., Willingham, M. C., Pastan, I., Gotto, A. M., and Smith, L. C . (1982). Exp. Cell Res. 141, 15. Wall, D. A,, Wilson, G., and Hubbard, A. L. (1980). Cell 21, 79. Wehland, J . , and Willingham, M. C. (1983). J . Cell Biol. 97, 1476. Wehland, J., Willingham, M. C., Dickson, R. B . , and Pastan, I. (1981). Cell 25, 105. Wehland, J . , Willingham, M. C., Gallo, M. G., and Pastan, I. (1982a). Cell 28, 831. Wehland, J . , Willingham, M. C., Gallo, M. G., Rutherford, A. V., Rudick, J., Dickson, R. B., and Pastan, I. (1982b). Cold Spring Harbor Symp. Quant. Biol. 46, 743-753. Weller, N. K. (1974). J. Cell Biol. 63, 699. Wheeler, T. J., Simpson, I. A., Sogin, D. C., Hinkle, P. C., and Cushman, S . W. (1982). Biochem. Biophys. Res. Commun. 105, 89. Willingham, M. C., and Pastan, I . (1978). Cell 13, 501. Willingham, M. C., and Pastan, I. (1980). Cell 21, 67. Willingham, M. C., and Pastan, I. (1981a). In “Cellular Controls in Differentiation” (C. W. Lloyd and D. A. Rees, eds.), pp. 59-78. Academic Press, New York. Willingham, M. C., and Pastan, I. H. (1981b). In “Receptor-Mediated Binding and Internalization of Toxins and Hormones” (J. L. Middlebrook and L. D. Kohn, eds.), pp. 135-147. Academic Press, New York. Willingham, M. C., and Pastan, I . H. (1982). J. Cell Biol. 94, 207. Willingham, M. C., and Pastan, I. H. (1983a). Proc. Natl. Acad. Sci. U.S.A. 80, 5617. Willingham, M. C., and Pastan, I. (1983b). In preparation. Willingham, M. C., and Pastan, I. H. (1983~).In “Genes and Proteins in Oncogenesis” (I. B. Weinstein and H. I. Vogel, eds.), pp. 183-195. Academic Press, New York. Willingham, M. C., and Pastan, I. H. (1983d). Recept. Recognition Ser. B 15, 4-17. Willingham, M. C., and Pastan, I. H. (1983e). In “Methods in Enzymology” (S. Fleischer and B . Fleischer, eds.), Vol. 98, pp. 266. Academic Press, New York. Willingham, M. C . , and Rutherford, A. V. (1984). J. Histochem. Cytochem. 32, 455. Willingham, M. C., and Yamada, S . S . (1978). J. Cell Biol. 78, 480. Willingham, M. C., Maxfield, F. R., and Pastan, I. (1979). J. Cell Biol. 82, 614. Willingham, M. C., Haigler, H. T., Dickson, R . B., and Pastan, I. (1981a). Int. CellBiol. p. 613. Willingham, M. C., Keen, J. H . , and Pastan, I . (1981b). Exp. Cell Res. 132, 329. Willingham, M. C., Pastan, I., Sahagian, G . G . , Jourdian, G . W . , andNeufeld, E. F.(1981c). Proc. Natl. Acad. Sci. U.S.A. 78, 6967. Willingham, M. C., Rutherford, A . V., Gallo, M. G., Wehland, J., Dickson, R. B., Schlegel, R . , and Pastan, I. (1981d). J. Hisrochem. Cytochem. 29, 1003. Willingham, M. C., Yamada, S . S . , Davies, P. J. A., Rutherford, A. V., Gallo, M. G . , and Pastan, 1. (1981e). J. Histochem. Cytochem. 29, 17. Willingham, M. C., Haigler, H. T., FitzGerald, D. J. P., Gallo, M. G., Rutherford, A. V., and Pastan, 1. H. (1983a). Exp. CellRes. 146, 163. Willingham, M. C., Pastan, I. H., and Sahagian, G. G . (1983b). J. Hisrochem. Cytochem. 31, 1.
92
MARK C. WILLINGHAM AND IRA PASTAN
Willingham, M. C., Hanover, J. A,, Dickson, R. B . , and Pastan, I. (1984). Proc. Nutl. Acud. Sci. U.S.A. 81, 175. Wissig, S. L. (1962). Anat. Rec. 142, 292. Woodward, M. P., and Roth, T. F. (1978). Proc. Natl. Acad. Sci. U.S.A. 75, 4394. Yamada, S. S . , Yamada, K. M., and Willingham, M. C. (1980). J . Hisrochem. Cyfochem. 28,953. Yamaizumi, M., Mekada, E., Uchida, T., and Okada, Y. (1978). Cell 15, 245. Zoon, K. C., Amheiter, H., Zur Nedden, D., FitzGerald, D. J . P., and Willingham, M. C. (1983). Virology 130, 195.
INTERNATIONAL REVIEW OF CYTOLOGY, VOL. 92
Stability of the Cellular Translation Process T. B. L. KIRKWOOD,R. HOLLIDAY,AND R. F. ROSENBERCER National Institute for Medical Research, The Ridgeway, Mill Hill, London, England I. 11.
111.
IV . V.
VI. VII. VIII.
IX .
93 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 Models of Error Propagation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 A. Protein Synthesis , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 B. Generalized Theory . . . ............................. 103 Evolutionary Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103 A. Origins of Translation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103 B. Selection for Accuracy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 C. Evolution of Aging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 Nature of Proof or Disproof , . . , . . . . . . . . . . . . . . . . , . . . . . . . . . . . . 111 Measurement of Error Rates.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . _ . . . . . _ . . _ . . _111 . A. Protein Errors. . . . . . . . . . . . . . . B. Transcription Errors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 Turnover of Proteins and Messages . . . . . . . . . . . . . . . . . . . . . . . . . . . 118 Evidence from Prokaryotes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119 Evidence from Eukaryotes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121 A. Fungi.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121 B. Mammalian Cells.. . . . . . . . . . . . . . . . . . . . . , . . . . . . . . . . . . . . . 123 Conclusions . . . . . . . . . . . . . . . . . . , . . . . . . . 126 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
I. Introduction Stability of the cellular translation process is easily taken for granted. Life, as we know it on earth, could not continue without it, and the near universality of a single genetic code indicates that translation is among the oldest and most tightly conserved features of cellular organization. Nevertheless, it would be a mistake to assume too lightly that genetic translation has always been stable, or that it is always stable now. First, it seems very likely that the earliest versions of the translation apparatus were comparatively imprecise, so that one of the most challenging problems concerning the origins of life is to explain how an accurate translation system evolved out of initially inaccurate parts (see, for example, Eigen and Schuster, 1979). Second, a major theory to explain aging in contemporary higher organisms is the error theory, which suggests that the stability of translation breaks down in somatic cells, resulting eventually in senescence and death of the entire organism (Orgel, 1963, 1973; Kirkwood, 1977; Kirkwood and Holliday, 1979). 93 Copyright 0 1984 by Academic Press. Inc.
All rights of reproduclion in any form reserved. ISBN 0-12-364492-5
94
T. B. L. KIRKWOOD ET AL.
Genetic translation is indeed a remarkable process. Information encoded in essentially inert form as polynucleotide sequences of DNA is first transcribed into messenger RNA working copies, and the mRNAs are then translated into functionally active form as a set of polypeptides (proteins). Each step of this process is mediated by a highly specific set of information handling molecules, most of which are proteins. The molecules which are most directly involved in the translation process are, for transcription, RNA polymerases, and, for translation, ribosomal proteins, aminoacyl-tRNA synthetases, and the tRNAs themselves. In addition, large numbers of proteins are also involved in regulatory aspects of mRNA processing and of protein synthesis (see, for example, Alberts et al. , 1983), as well as in enzymic editing reactions which are used to detect and eliminate errors as they occur (Fersht, 1981). The net result of this concerted activity is the production of proteins with an astonishingly low level of errors, estimated at approximately one mistake in every 103-104 amino acids (Loftfield and Vanderjagt, 1972; Gallant and Foley, 1980; Gallant and Palmer, 1979; Parker et al., 1980; Rosenberger et al., 1980). The smallness of the error level in protein synthesis means that for many purposes it can be safely ignored. Unless a protein subunit is quite large, greater than say 500 residues, it is likely not to contain any errors at all. Also, since translation cannot work in reverse, any errors which are made will in due course be eliminated from the cell as the defective protein is degraded during the course of normal molecular turnover. The irreversibility of protein synthesis is enshrined in the so-called “central dogma” of molecular biology, and it is in effect guaranteed by the redundant (i.e., several-to-one) nature of the genetic code (Watson, 1976). Orgel (1963) has pointed out, however, that there is a possible major consequence of translational errors which is overlooked by the conventional dogma of a unidirectional flow of information from DNA to RNA to protein; namely, that the errors, while transient in themselves, may initiate a cyclic feedback of mistakes within the translation process. Occasionally, an error made in the synthesis of an information handling enzyme will create a molecule which retains some or all of its activity, but has altered specificity. By taking part in the next round of translation, this erroneous enzyme can, in turn, generate more faults, and in this way errors might be propagated progressively within the translation apparatus until eventually the cell could no longer synthesize proteins with sufficient accuracy to carry out essential functions. At this point the cell would succumb to what Orgel graphically termed an “error catastrophe,” and he speculated that such a mechanism may explain the clonal senescence of certain types of dividing cells, for example, mammalian fibroblasts, and the ultimate loss of function of fixed postmitotic cells, such as neurons (Orgel, 1963, 1973; see also Medvedev, 1962). Over the two decades since the error propagation hypothesis was proposed, it
CELLULAR TRANSLATION PROCESS
95
has generated considerable controversy, both at the level of theory and in the interpretation of experimental results. It has also contributed to an understanding of the minimum stability requirements for the translation process to have started up at all. The idea that error feedback can, and probably does, occur is not in serious dispute. What is at issue is whether feedback occurs at a level which is sufficiently high to have biological significance, especially in the context of aging. The main experimental difficulty lies in the fact that error levels are normally so low that they are extremely hard to measure, and it is not known what size of increase would be lethal to a cell. Theoretical studies therefore play an essential role in refining the formulation of the hypothesis so as to make it easier to interpret experimental data within a unified framework. In this review, we examine the current status both of the theory and of the experimental evidence concerning the potential for error propagation to destabilize the translation process.
11. Models of Error Propagation
A. PROTEINSYNTHESIS To make a full study of the theoretical dynamics of error propagation requires that each level of macromolecular information transfer is considered. These include transcription, translation, and also DNA replication and repair, since errors in the synthesis of DNA polymerases and repair enzymes can give rise to mutations in genes coding for information handling molecules. This latter kind of error will be especially damaging to cell viability and, in the following section, we consider its consequences more closely. For the moment, however, a useful start is made by confining attention to a simplified representation of the translation process, in which only a single feedback loop is considered (Fig. 1). Figure I represents for convenience the translation apparatus as a series of discrete “generations,” each of which is responsible among other things for synthesizing its successor. The definition of a generation is somewhat arbitrary since the components of the translation apparatus are renewed more or less continuously, but it may be regarded roughly as the mean time to replacement of each constituent molecule. This definition has the advantage that it works equally well in nondividing and in dividing cells, since protein and RNA turnever occur in both. The reduction of the translation process to a single feedback loop can be understood either as combining the separate steps of transcription and translation into one, or as confining attention only to the latter. It is reasonable as a first approximation to concentrate primarily on protein synthesis as errors in this step are likely to be statistically more numerous than errors of transcription, even if the basic error rates are about the same.
96
T. B. L. KIRKWOOD ET AL
I
+ ---
G e n e r a t i o n t-1 t r a n s l a t i o n a p p a r a t u s ( a c c u r a c y q t-1 )
+
Generat i o n t t r a n s l a t i o n apparatus (accuracy q ) t
I -,
FIG. I . Simplified feedback loop illustrating the potential for propagation of errors during the translation of genetic information from DNA into protein. The stability of translation is determined by the relationship between 4, and 4,-
,.
The central issue as regards stability of the translation process is how the average accuracy qt (probability of correctly inserting an amino acid) of any given generation t depends on the accuracy 9,- of its predecessor. Obviously, if q,
q r - , the accuracy is increasing. In a stable state, accuracy would be constant with qr =
,
qr-
1
-
qstable.
Orgel (1 963) suggested originally that the error frequency e, (= 1 - 4,) should be expected to multiply exponentially, in which case error catastrophe is inevitable, and some special consideration is required to resolve the paradox of how any life form based on genetic translation can exist. Later, Orgel amended this view and pointed out that if the feedback of errors into the next generation was numerically smaller than the errors already existing, the error level would stabilize at a constant value (Orgel, 1970). By means of this correction, Orgel introduced a distinction that it is useful to draw between ( I ) errors which would arise even if the current translation apparatus contained no faults and (2) errors which are the direct consequence of existing erroneous proteins. He suggested that the error frequency e, could be written as a sum of two terms
e,
=
E
+ ae,-,
where E is the background error frequency of an error-free translation apparatus, and a is a parameter representing feedback from previous errors. Equation (1) predicts two alternative forms of dependence for e, on t, depending on whether cc is less than or greater than one. If (Y < 1, e, will always converge to estable= E / ( 1 - a).If a > I , e, increases without limit and error catastrophe is inevitable. Clearly, in the framework of this model any cell which is not rapidly to become extinct must have a <1, and the question is whether it is ever possible that this rule is violated. A later model of the dynamics of error propagation refined the concept of
CELLULAR TRANSLATION PROCESS
97
translational stability further (Hoffmann, 1974; Kirkwood and Holliday, 1975a). In this model, each generation of the translation apparatus was defined to consist of a set of polypeptide “adaptors” which combine the functions of all components of the protein synthesizing machinery. (Alternatively, the model can be regarded as representing just the aminoacyl-tRNA synthetases on the assumption that errors elsewhere in the translation apparatus can temporarily be ignored.) For each adaptor it was assumed there is a subset of m residues whose correct insertion is vital for activity, and a further nonoverlapping subset of n residues vital for specific recognition of the correct substrate. Incorrect insertion of one or more of the m residues would result in total loss of activity, while incorrect insertion of one or more of the n residues would result in total loss of specificity. (In fact, it is more likely that an error in one of the n residues would alter specificity, rather than abolish it, but in a population of adaptors, if the change in specificity is completely random, the average effect is the same.) Errors in any remaining residues are assumed to be unimportant. In order to model error propagation, attention can be restricted to only those adaptors which contain no errors in the m sites, as inactive adaptors do not contribute to translation. The viable adaptors in generation r consist of two types: a fraction q:- which are normal (i.e., contain no errors in the n sites), and a fraction 1 - 4:- which are erroneous (i.e., have lost their specificity). The model is completed by specifying the different rates of insertion of correct and incorrect amino acids by normal and erroneous adaptors. For a normal adaptor it was supposed that correct insertions are made S times as fast as each of the X - 1 possible incorrect insertions are made, A being the number of different amino acids and S being the specificity of a normal adaptor. For an erroneous adaptor it was supposed that all the A possible insertions (i.e., correct and incorrect) are made at equal rates, but that the total rate of insertion was R times the total rate for a normal adaptor. R may be termed the residual activity of an erroneous adaptor. In the original version of the adaptor model (Hoffmann, 1974) R did not feature as a parameter and it was supposed that for an erroneous adaptor the rate of each possible insertion was the same as the rate of each incorrect insertion for a normal adaptor. This is equivalent to setting R = h/(X + S - l), which for any reasonable value of S amounts to a drastic reduction in activity. Kirkwood and Holliday (1975a) pointed out, however, that single amino acid substitutions are known which alter specificity but do not severely reduce activity, and they suggested R should be set as a flexible parameter of the model. The idea that R need not necessarily be small is also supported by the restriction of attention to only those adaptors which contain no errors in the m activity-determiningsites. It should be noted that R is affected not only by the total rate of activity of an erroneous adaptor while it is operational but also by its relative lifetime in the cell compared to its normal counterpart. If an erroneous adaptor is more labile than a
98
T. B. L. KIRKWOOD ET AL.
normal one, or subject to preferential scavenging by proteolytic enzymes, its average activity measured over time will be reduced, and this will be reflected in the value of R . From these assumptions, the average rates for generation t adaptors of making correct and incorrect insertions may be calculated, and the average accuracy qr determined as the rate of making correct insertions, divided by the total rate of making insertions, both correct and incorrect. This gives the equation
q:-l[AS qr = q:-lA(S
+
- (S A - 1)R] - 1)(1 - R)
+A
+ (S + A
+ A(S + A
- l)R - l)R
(2)
for the relationship between qr and qr- (Kirkwood and Holliday, 1975a). (A minor approximation is made in deriving this discrete time recurrence equation by assuming that accuracy remains constant during each single generation of the translation apparatus, but this is negligible except when q changes rapidly from one generation to the next, as could occur in the final approach to an error catastrophe. ) Despite its apparent complexity, Eq. (2) generates a simple picture of the dependence of qr on qt- from which a number of important conclusions may be drawn. In a plot of qt against qr- I the equation is represented by an S-shaped curve (Fig. 2 ) . If the curve lies entirely below the diagonal line qr = qr- I , on which accuracy is constant, accuracy decreases steadily from generation to generation, and error catastrophe is unavoidable. On the other hand, if the curve crosses into the upper left region of the graph where qr > qr- I , then it must intersect the line qr = qr- I in two points and the higher of these defines a point of stable accuracy, qstable.The lower point of intersection, qthreahold, is also a point of constant accuracy, but it may readily be seen that it is unstable. Provided accuracy does not at any time fall below qthreshold it will always return to qstable after a perturbation, but if accuracy does drop below qthreshold it will thereafter decline irreversibly toward error catastrophe. The increased versatility of the adaptor model compared with Eq. (1) is readily seen if Eq. ( 1 ) is plotted in a form similar to Fig. 2 (see Fig. 3). In this case, the relationship between qr and qt- I is a straight line which either crosses (a< 1) or does not cross (a >1) the line of constant accuracy. Because a is treated as a constant, independent of q, stability is either absolute, or nonexistent. No threshold level of accuracy exists for Eq. ( I ) , so the transition of a cell from stability to instability is either disallowed, or requires a change in the parameter a. The flexibility of the adaptor model is further increased if account is taken of the stochastic nature of errors in protein synthesis. Because the deterministic model is much more readily tractable than its stochastic counterpart, Eq. ( 2 ) has been derived only in the deterministic case. In reality, however, the expression for qr on the right-hand side of the equation represents only some form of statistical average, about which there will at any time be random scatter (see
99
CELLULAR TRANSLATION PROCESS 1 .(
qt
1.o
lfh qt-1
FIG.2. Relationship between q, and q r - , predicted by adaptor model of error propagation (Hoffmann, 1974; Kirkwood and Holliday, 1975). The curve intersects the line q, = q r - , in two points, the higher of which defines a point of stable accuracy, qrt&le. Provided accuracy remains above the lower point of intersection, qthreshold, it will always revert to its stable value. Below qthreshold, accuracy falls progressively resulting in total loss of specificity (9 = ]/A, where A is the number of different amino acids). Separation between qstahle and qthreshold is determined by two parameters, S (specificity of error-free adaptors) and R (residual activity of erroneous adaptors); see text for definitions. Increasing S moves the curve upward, including the right-hand endpoint. Decreasing R moves the curve upward and to the left, leaving both endpoints fixed. It is possible also for the curve to lie entirely below the line q, = 4,- , in which case no point of stable accuracy exists (not shown).
,
Hoffmann, 1974; Kirkwood and Holliday, 1975a). By considering the probabilities of correct and incorrect insertion of the n residues as well as the rates, an approximation to the standard error of qr may be written as
where N is the number of viable (i.e., normal + erroneous) adaptors present (Kirkwood and Holliday , 1975a). If the separation between qstableand qthreshold is not too large relative to SE (q),there will be a possibility of a purely random transition from the stable to the unstable state. One further model of error propagation in protein synthesis was described by
100
T. B. L. KIRKWOOD ET AL.
0
Frc. 3. Relationship between qr and q,-, for a model with linear feedback of errors (Orgel, 1970). Depending on whether the linear feedback coefficient a is less than or greater than one, translation is either always stable or always unstable. No threshold level of accuracy exists for this model.
Goel and Ycas (1975). However, it has been argued elsewhere that this model is fundamentally flawed (Kirkwood, 1980) so we shall not consider it further here.
B. GENERALIZED THEORY Models based on protein synthesis alone demonstrate certain principal features of error propagation in a self-replicating translation apparatus. However, as has been remarked already, a fully realistic theory will have to take account of transcription and DNA replication errors as well. Since proteins are involved in each of these operations, the possibilities for highly complex feedback relationships are many, and to date no detailed mathematical model involving multiple levels of information transfer has been described. Kirkwood (1980) reviewed in outline the primary features which a more general model will require, and these are represented diagrammatically in Fig. 4. More recently, Hasegawa et al. (1984) reconsidered these points in relation to evolution of subcellular organelles and suggested a numerical classification of errors into three types. Type I errors are the errors which arise simply in the
CELLULAR TRANSLATION PROCESS
101
translation step of protein synthesis. Type 2 errors are errors in the transcription of DNA to RNA. Type 3 errors are mistakes in the replication or repair of DNA, or in other words, gene mutations. It is also useful in thinking about the dynamics of error propagation to restrict attention only to those constituents of DNA, RNA, and protein that are themselves involved in some aspect of the translation process. With this restriction, we are interested only in how errors are propagated within the process itself, and we can defer for a moment consideration of how the rest of the cell’s functions are affected. Type 3 errors are qualitatively different from types 1 and 2 in that, once made, they are permanent, whereas types 1 and 2 last individually only as long as the protein or mRNA molecule persists within the cell. In consequence, type 3 errors are very much more serious, and it is unsurprising that they occur at very much lower rates (see below). Type 2 errors are again more serious than type 1 since, on average, a single message may be translated many times and, therefore, an error in transcription will tend to produce a burst of errors in newly synthesized protein. On the other hand, type 1 errors although individually less serious are likely on average to be by far the most numerous and it is therefore reasonable to regard type 1 errors as the primary medium through which error propagation takes place. The differences in frequency and effect of types 1 and 2 errors may be of particular significance in the way a stable translation process could become
I
I
Type3
DNA polymerase
1
I -1,
- DNA
I
I d
Ribosomal proteins, aminoacyl-tRNA synthetases
Protein
FIG. 4. Schematic representation of the main pathways of information transfer between macromolecules. The bold arrows denote transcription of DNA into RNA and translation of RNA into protein. The dashed arrow denotes DNA replication. The remaining arrows indicate pathways for cyclic propagation of errors through mistakes in the synthesis of information-handling proteins.
T. B. L. KIRKWOOD ET AL
I02
I
Error catastrope
Error level
I
Tvoe 2 errors
Time FIG.5 . The error level in the cellular translation process is expected to fluctuate randomly about qatableunless it crosses qthresholdwhen it will increase progressively to error catastrophe. Type 2 (transcription) errors might be particularly significant in destabilizing the translation process since a single mistake will produce many erroneous proteins. In the case illustrated, two type 2 errors are assumed to have occurred in close succession.
destabilized. Periodically, a type 2 error might cause a relatively large increase in the protein synthetic error rate and this might on occasion be sufficient to push the system past the threshold of recovery (Fig. 5). Once destabilization had occurred, the error frequency would continue to rise under the impetus primarily of type 1 errors, until eventually the accuracy of synthesis of RNA and DNA polymerases became so low that types 2 and 3 errors would occur more frequently, and these would then contribute to the final demise of the cell (Orgel, 1973; Lewis and Holliday, 1970; Fulder and Holliday, 1975; Kirkwood, 1980). The scenario in which cells experience an initial build up of protein errors, followed by progressive deleterious changes in RNA, DNA, membranes, organelles, and so on, is the basis of what is termed the general error theory of aging (Holliday and Kirkwood, 1983). [This theory should not be confused with the somatic mutation theory (Szilard, 1959; Curtis, 1966; Burnet, 1974; Holliday and Kirkwood, 198'1; Morley, 1982) which attributes cell aging directly to the gradual accumulation of multiple mutations in the genetic material of cells, and which does not invoke the idea of error propagation (Holliday and Kirkwood, 1983). In contrast, the error theory predicts gene mutations will occur, but does
CELLULAR TRANSLATION PROCESS
103
not require them, and a key point is that even with a mutation-free genome a cell cannot survive if either it inherits a nonviable translation apparatus or the accuracy of its translation breaks down. J
111. Evolutionary Considerations A. ORIGINS OF TRANSLATION The evolutionary origins of genetic translation remain obscure (Crick et af., 1976; Miller and Orgel, 1973; Eigen and Schuster, 1979). Various models have been proposed based on properties of the genetic code (Crick, 1968; Woese, 1967; Hartman, 1975), while others have developed from the assumption that there existed an initially random autocatalytic system comprising polypeptides and polynucleotides (Hoffmann, 1974; Mizutani and Ponnamperuma, 1977). Whatever the true sequence of events may have been, at some point primitive replicators acquired the ability to separate the responsibilities of storing and implementing genetic information among two different sets of molecules and thereby gained a greatly enhanced adaptational flexibility. For the purpose of this review, we ignore the steps leading to the first instance of genetic translation and consider only the selective pressures which operated subsequently. B . SELECTIONFOR ACCURACY Since the translation process is fundamental to every aspect of life, it is obvious that natural selection acts as readily on the accuracy of macromolecular synthesis as on any other trait. The force of this selection may be judged from the impressive array of mechanisms which cooperate to produce the high fidelity of synthesis of DNA, RNA, and proteins that is found in present-day species. The primitive translation process presumably lacked anything but the most basic chemical or mineral catalysts and it would have depended primarily on simple differences in binding energies and stereochemistry for its specificity of translation. Gradually, there must have evolved enzymes, which were themselves the products of translation, that had greater and greater specificity until at some point the entire replication and translation machinery became fully autocatalytic. In the evolution of the translation process from its primitive state to the complex system we see today, natural selection would have favored two separate, but related characteristics: accuracy and stability. The selection pressure continually to improve these characteristics may be understood in terms of the deleterious consequences of errors in RNA or protein synthesis. These errors have both a general consequence and a specific one. The general consequence is a lowering of metabolic efficiency: it is a waste using energy and resources to
104
T. B . L. KIRKWOOD ET AL
synthesize an RNA or protein molecule which cannot perform its normal function. For this reason alone, we can expect selection to produce a high degree of translational accuracy. The specific consequence relates only to the proteins required for information transfer between macromolecules: it is the possibility of error propagation which could lead ultimately to cell death. As discussed earlier, error propagation occurs initially in proteins, but in due course it will also affect the fidelity of DNA replication itself. Therefore to protect its DNA, the organism must achieve a considerable degree of accuracy in transcription and translation, and it must also ensure that the translation process as a whole remains stable. The evolution toward greater translational accuracy and stability can be described most conveniently in the framework of the Hoffmann-Kirkwood-Holliday (HKH) model, described in the previous section (see Fig. 2). By selecting for greater specificity, S , the curve in the plot of q, against 4,- is raised. This results in an increase in the average accuracy, q\table.By selecting for lower residual activity, R , the curve is raised upward and to the left in such a way that the endpoints remain fixed. This also results in an increase in qctable,but less directly. Both these changes are likely also to increase the separation between qstable and qthre+,ld and therefore to decrease the risk of precipitating the cell onto the path to error catastrophe. Evolution toward reduced R is most readily seen as occurring through acquisition of scavenging mechanisms for preferentially degrading erroneous protein, although in the early evolution of enzyme structures there may also have been selection for information handling molecules which have stringent requirements for structural stability and which are therefore likely to be highly labile if an error is made in their synthesis. The benefits of raising the accuracy and stability of translation are obvious, but it is also clear that there are costs. The cost may either be a direct cost, in the sense of an increased energy requirement, or it may be an indirect trade-off against, for example, the rate of protein synthesis. Generalized schemes for proofreading the synthesis of macromolecules have been discussed by Hopfield (1974) and by Ninio (1975). In Hopfield’s kinetic proofreading scheme, the accuracy of synthesis is raised by means of one or more intermediate reaction steps in which the enzyme-substrate complex is driven into a high-energy transitional state from which there is an increased probability of dissociation of an incorrect match. In Ninio’s “delayed reaction” scheme, each step of synthesis is subjected to a time delay, during which the differential likelihood of dissociating correct and incorrect bonds results in an increased level of discrimination. This causes a slowing of the total rate of synthesis, and it also increases the chance that correct bonds may dissociate, requiring reinitiation of the entire step. Other mechanisms, such as “double-sieve’’ editing (Fersht and Dingwall, 1979) require that, on average, more than one unit of substrate (amino or nucleic acid) is inserted and excised during each completed step, and these consume both energy and time. Similarly, a variety of costs can be identified in the development of
,
CELLULAR TRANSLATION PROCESS
105
protein scavenging systems, where preferential elimination of abnormal molecules is likely to involve a greater turnover of normal protein as well, and there is experimental evidence that a cell’s ability to hydrolyse abnormal protein does indeed require energy (Goldberg, 1972; Bukhari and Zipser, 1973; Waxman and Goldberg, 1982). Theoretical analysis by Savageau and Freter ( 1 979) of the energy cost of proofreading the aminoacylation of tRNAs concluded that this alone accounts for approximately 2% of the energy required to synthesize a bacterial cell. In rapidly growing bacteria, it has been shown that ribosomes account for up to 50% of the RNA and 30% of the protein which is synthesized (Maaloe, 1979), and since evolution of the ribosome is likely to have been directly related to the evolution of accurate translation, this too represents a major investment in accuracy. Therefore, although detailed calculation of the full metabolic investment in accuracy is not yet possible, it appears highly probable that it is a very significant fraction of a cell’s total energy budget. The upshot of these considerations is that while a high level of translational accuracy and stability is clearly of positive benefit to the organism, there is likely to be a law of diminishing return, such that the costs of raising S and reducing R both increase sharply. In primitive organisms, when the fidelity of translation was presumably low, the benefit of increased efficiency and stability resulting from increased accuracy would initially have outweighed its cost. This principle might be expected to have resulted in a high degree of translational stability, especially in unicellular organisms where initiation of an error catastrophe would have disastrous consequences. In higher organisms, however, the selective balance would be different, because the great majority of cells are somatic cells, and in these it does not matter so critically if a proportion of them degenerate to error catastrophe.
C . EVOLUTIONOF AGING Aging may be defined most conveniently as a process which renders individuals more susceptible as they grow older to the various factors which may cause death (Medawar, 1946; Maynard Smith, 1962; Kirkwood, 1984a). Although there are limitations to this definition based on survivorship (see Kirkwood, 1984a, for review), it is commonly recognized that it adequately describes the generalized deterioration which occurs in higher animals after they have passed their reproductive prime. Although it is not uncommon to find the term aging applied widely to plants, invertebrates, and certain microorganisms, there are major difficulties in adapting a single term to describe meaningfully an extreme diversity of life-history patterns (Kirkwood and Cremer, 1982; Kirkwood, 1984a). In this section we consider only the aging of higher organisms which undergo repeated sexual reproduction during their lifetime, and we ex-
106
T. B . L. KIRKWOOD ET AL.
clude therefore all organisms with asexual life-histories, as well as organisms, such as Pacific salmon, which reproduce only once and then die in a determinate fashion. The aging of an organism which reproduces repeatedly has long been recognized as an evolutionary puzzle, since it would clearly be more advantageous to the individual to survive and reproduce indefinitely (Medawar, 1952; Williams, 1957; Hamilton, 1966; Kirkwood, 1977, 1981; Kirkwood and Holliday, 1979). The theory that aging is an evolved mechanism to promote species’ adaptability rests on the weak principle of group selection (Maynard Smith, 1976), and is not generally tenable. Therefore, it is recognized that evolution of aging is explained either as arising through the failure of natural selection to prevent it (see Medawar, 1952) or as a by-product of selection for some other trait (see Williams, 1957; Kirkwood and Cremer, 1982). The disposable soma theory of aging (Kirkwood, 1977, 1981; Kirkwood and Holliday, 1979) provides evolutionary support for the idea that aging is due to an energy-saving strategy of investing only the minimum necessary amount of resources in the maintenance of somatic cells. This in turn provides support for the hypothesis, suggested by Orgel (1963, 1973; see also Medvedev, 1962), that aging is due to instability in the cellular translation process. The starting point for the disposable soma theory is the observation, made orginally by Weismann (see Kirkwood and Cremer, 1982), that in multicellular organisms cells may be classed into two primary groups, reproductive cells and somatic cells. Reproductive cells constitute essential links in the chain of inheritance of genetic information and need a high degree of translational stability, since progressive deterioration in fidelity would be disastrous. By contrast, among somatic cells the future lineage of any single cell is limited since it is certain that, even in the absence of aging, the body of which it forms a part will eventually die. In somatic cells there is therefore not the same need for long-term stability, and this leaves the organism freer to vary its investment in translational accuracy and stability so as to maximize its lifetime reproductive success, or fitness. A trade-off which is of particular significance here is between the organism’s investment in reproductive output and its investment in translational stability. The larger the fraction of resources invested in translational stability, the smaller the investment in reproduction (and conversely), irrespective of whether resources are abundant or scarce. At either extreme, the organism pays a penalty. If investment in translational stability is too little, cells will become rapidly destabilized and death will result very soon. If investment in translational stability is too much, reproduction will be severely retarded. In consequence, the strategy of maximum fitness is one which optimally balances survival against reproduction, and it may be shown that the ideal investment of resources in mechanisms to promote survival is likely to be less than would be required for the organism to survive indefinitely (Kirkwood, 1981). To the extent that sur-
CELLULAR TRANSLATION PROCESS
107
viva1 is dependent on the preservation of translational stability, the disposable soma theory predicts therefore that aging may result from an increasing proportion of somatic cells becoming destabilized. The theory also predicts that longer lived species have, in general, a higher level of translational stability than shorter-lived species (Kirkwood and Holliday , 1979).
IV. Nature of Proof or Disproof Over the 20 years since the error catastrophe hypothesis was put forward, it has become increasingly clear that it is an extremely difficult concept to test. There are two main reasons for this. First, the error level in translation is so low that it is hard to measure it sufficiently accurately to be confident of detecting a change (see following section). Second, it is unclear how great an increase in error frequency would be necessary to produce an error catastrophe; it is possible that even quite a small increase could have a profound effect on cell viability. The term “error catastrophe” is probably quite misleading in that it suggests a state of presumably obvious disintegration. In fact, a relatively modest increase in error level from, say, to l o p 3 mistakes per amino acid residue would reduce the proportion of error-free protein from 99 to 90% for proteins of 100 residues, and from 90 to 37% for large proteins of 1000 residues. Such a change, together with a corresponding increase in the cell’s rate of protein turnover as scavenging mechanisms try to keep pace with new errors, would substantially reduce a cell’s metabolic efficiency and may be sufficient to prevent it from undergoing further cell division. At the same time, an increased level of faulty enzymes would render more likely one of the serious faults which kills a cell or blocks it from cycling. Despite these difficulties, numerous attempts have been made to measure the accuracy of the translation process and to determine the extent of error propagation in present-day cells (for earlier reviews, see Rothstein, 1977; Gershon, 1979; Medvedev, 1980; Gallant, 1981; Laughrea, 1982). These studies have followed two principal lines. In the first, which relates directly to the error theory of aging, cells such as mammalian fibroblasts which routinely undergo clonal extinction have been examined for evidence that the error frequency increases during clonal growth. In the second, cells such as bacteria, which would be regarded as normally stable, have been subjected to treatments known to increase the frequency of mistranslation in attempts to induce translational instability. That the results have been interpreted in conflicting and occasionally irreconcilable ways is due to the fact that the predictions which have been tested have tended to be indirect and they have not always been sufficiently clearly formulated. The most direct form of experimental test would be to measure the rates at
108
T. B. L. KIRKWOOD ET AL.
which various cell types make errors and to follow how these rates changed during clonal growth, or after sudden perturbations of fidelity. If it could be established unequivocally that the error rates do not progressively change, then the hypothesis of translational instability would be effectively disproven. On the other hand, if the error rate were found to increase, this would strongly support the error propagation hypothesis; final proof would require demonstration that active, erroneous information handling molecules were being made and that other changes which might affect translational accuracy could effectively be ruled out. Current methods for measuring error rates are not without limitations, however, and various indirect tests are therefore widely used. For example, if error propagation does take place an increasing proportion of molecules of each specific type of protein will contain one or more amino acid substitutions, and these may be detectable through a lowering of the average specific activity of the protein, or an increase in its thermal instability. Alternatively, viruses have been used as probes of translational instability on the grounds that if translational stability has broken down, then an increasing proportion of defective virus particles should be produced. The drawback to any indirect test is, however, that a positive or negative result may be an artifact of the system, rather than a consequence of the truth or falsehood of the primary hypothesis. Gallant (1981) has stressed the logical fallacy of basing any firm conclusion on an experiment of the indirect type, where more than one explanation can usually be offered for any given observation. Nevertheless, indirect tests do have a role to play in adding to or subtracting from the total weight of evidence which bears on the validity of error catastrophe hypothesis. A special kind of indirect test involves comparative study of cells from different species or of different types. For instance, the disposable soma theory suggests that if translational instability is the primary cause of aging, then cells from long-lived species should be more stable than cells from short-lived species (Kirkwood and Holliday , 1979). Similarly, somatic cells from multicellular organisms are predicted to be less stable than unicellular organisms and this too should be amenable to test. Finally, transformed cells, grown either from a malignant tumor or obtained by treatment of normal cells with carcinogens, grow indefinitely in culture in contrast to their normal counterparts; this last difference is explicable in terms of the error theory if either selection against destabilized cells is stronger in more rapidly growing transformed cell populations because translational efficiency becomes rate-limiting for cell division (Holliday, 1975; Kirkwood and Holliday, 1975b), or reactivation occurs in transformed cells of special error-correcting mechanisms which normally operate only in germ cells (Kirkwood, 1977). It should be noted that each of these comparative predictions relates primarily to the stability of translation, whereas comparisons are most easily made of the accuracy of translation. The HKH model illustrates how
109
CELLULAR TRANSLATION PROCESS 1.0 I
I
/
FIG.6. The adaptor model of error propagation (see Section 11) indicates that accuracy and stability of translation are to some extent independent. Curves A and B have the same level of stable accuracy, qstable, but curve A has a lower threshold than curve B and is therefore more stable. Such a situation could arise if curve A represents a translation apparatus with lower values of R and S than curve B. (Only the upper right regions of the curves are shown.)
stability and accuracy, although related, are also partially decoupled (Fig. 6), and it is thus possible that accuracy might be more or less the same, while stability is quite different. To conduct a proper comparative analysis of translational stability it is necessary, therefore, to measure protein turnover rates (see Section VI) as well as the error rates in protein synthesis. Finally, tests can be made of models of error propagation as well as of the general predictions of the error catastrophe hyopthesis itself (see, for example, Gallant and Prothero, 1980). These tests have the advantage that models generally make quantitative predictions, so that tests can be more exact, but it is mistaken to regard a test of a model as necessarily the same thing as a test of the theory which the model represents. In fact, for reasons discussed earlier, none of the error propagation models to date can be regarded as sufficiently realistic to warrant detailed quantitative test. This does not mean, however, that the models have no role to play in experimental test of the error propagation hypothesis. For example, the HKH adaptor model provides a useful background to a series of experiments in Escherichia coli in which aminoglycoside antibiotics, which increase the rate of ribosomal mistranslation. were added to the medium in an
110
T. B. L. KIRKWOOD ET AL 1.o
FIG.7. Effect of streptomycin on the accuracy of translation in a stable organism (e.g., E . coli), according to the adaptor model (see Section 11). Without streptomycin, the accuracy remains at the stable point A on the upper curve. Adding streptomycin decreases specificity, S, and displaces the curve DA down to the curve CB. Accuracy falls to the lower stable point, C. On removing streptomycin, curve CB is restored to DA and accuracy returns to the stable point A. In practice, there would be time lags while streptomycin is taken up by the cells and diluted out from them, so the changes in accuracy between A and C, and back again, would not follow exactly the indicated pathways but would be more gradual. If curve CB had no stable point (not shown) accuracy would fall progressively after addition of streptomycin and would only return to A on removal of streptomycin if it had not already fallen below qthCerhold for curve DA. Note that the figure shows only the upper right regions of the curves. (From Kirkwood, 1980.)
attempt to induce error catastrophe (see Section VII). The effect of adding antibiotic to the culture medium is roughly equivalent to decreasing the specificity parameter, S, causing a downward displacement of the curve. Depending on the dose and cell strain used, this may or may not eliminate the stability of translation, but even if it does not, qstableis reduced (Fig. 7). On removing the antibiotic the system should return to the original qstableprovided accuracy has not already fallen below the normal qthreshold. In this way judicious experimentation can test the general stability characteristics of the model, as defined by the critical points shown in Fig. 7. However, the precise pathways taken between the points will depend both on the kinetics of uptake and clearance of antibiotic from
CELLULAR TRANSLATION PROCESS
111
the cell and on more complex realities of the translation process than are currently represented in the model.
V. Measurement of Error Rates A. PROTEINERRORS A number of factors combine to make the determination of protein errors a difficult and challenging problem. Mistakes in transcription or translation will introduce errors at random sites in the polypeptide chain and any residue may be replaced by several alternatives. Purification and detection are therefore much more complex than with isozymes or mutant proteins. In the latter two cases, proteins differ by unique substitutions and can often be resolved into distinct bands or peaks by chromatography or electrophoresis. No such bands will appear when the substitutions are random, and the likelihood of losses during purification is very great. Despite these difficulties, several approaches have been described which, potentially at least, could detect error rates as low as one mistake in 103-105 codons read. None of these as yet provides a complete solution to the problem, but all are capable of further refinement and they have provided the most incisive data available at present. The principles on which they are based and their attendant advantages or disadvantages will be discussed in this section. 1, Radiolabeling of Proteins with Specific Amino Acids
When the structural gene of a protein does not specify the insertion of a particular amino acid, it will appear in the polypeptide only through a decoding mistake. This leads to a relatively straightforward method of measuring errors. The normally missing amino acid is supplied in radioactive form during protein synthesis, the protein purified to homogeneity, and its radioactivity determined. The assays can be made very sensitive by using amino acids of high specific activity. This approach can further be modified for use on proteins which do contain all 20 amino acids. Most proteins can be hydrolyzed enzymatically into specific polypeptide fragments. If any fragment does not contain all 20 monomers it could be used to determine error frequencies after labeling with the missing amino acid (Loftfield and Vanderjagt, 1972; Bouadloun et al., 1983). While it has great potential, there are several difficulties inherent in this whole approach. The most serious problem is the need to purify the protein so that traces of other polypeptides, containing the labeled amino acid, do not interfere. Errors which change the properties of molecules so that they no longer copurify with the correctly assembled polypeptide will not be detected. Further, isolation
112
T. B . L. KIRKWOOD ET AL
procedures are relatively laborious and this limits the number of samples which can be processed. This is particularly true when the protein has to be degraded into peptides which themselves have to be purified. Finally, the choice of proteins is somewhat limited, for most polypeptides do contain all 20 amino acids and the larger proteins give rise to a forbidding array of proteolytic fragments. Radiolabeling has been used to study accuracy in the synthesis of a mammalian histone lacking methionine (Buchanan and Stevens, 1978; Medvedev and Medvedeva, 1978), of flagellin from E. coli which has no cysteine (Edelman and Gallant, 1977a), of the E. coli ribosomal protein L7/L12 lacking cysteine (Bouadloun et al., 1983) and of chymotryptic and tryptic peptides from rabbit hemoglobin and chick ovalbumin which do not contain leucine, isoleucine, or valine (Loftfield, 1963; Loftfield and Vanderjagt, 1972). The results obtained from these proteins are summarized in Table I and discussed in a later part of this section. 2. Miscoding for Nonsense and Frameshgt M ufations Mutations can alter certain amino acids to read as end of translation signals (nonsense codons). They can also, by the addition or deletion of bases, shift the reading frame to produce garbled messages. One feature of both these mutational changes, not necessarily shared by missense mutations, is that when they occur in the structural gene of an enzyme they lead to a completely inactive protein being formed. Active enzyme will then only be made after a decoding mistake, and enzyme assays can be used to measure error rates (Gorini, 1970; Gallant and Foley, 1980). With some enzymes, particularly hydrolases like P-galactosidase and alkaline phosphatase, activities as low as l o p 5 of the normal can be detected (Gallant and Palmer, 1979; Rosenberger et al., 1980). This approach has the advantages of requiring only small amounts of material and of being far simpler and more convenient than any other so far described. Its main drawback is that it does not measure the substitution of one amino acid for another but, rather, errors in discrimination between release factors and noncognate amino acids (nonsense mutations) or the rate at which irregular ribosome movements and transcriptional errors occur (frameshift mutations). These aspects of fidelity are not necessarily the same. Further clarification of the suitability of multimeric enzymes, such as P-galactosidase and alkaline phosphatase, for this assay is also needed. Each decoding mistake produces a single subunit which will become active only when combined with other subunits, themselves produced at rare and random intervals. Ideally, an enzyme should be used that is active as a monomer. To date, readthrough assays with nonsense and frameshift mutations have been limited to prokaryotes and lower eukaryotes, since it is only in these organisms that such mutations have been detected with certainty. Results are given in Table 1.
TABLE I ERRORRATESDURING PROTEIN SYNTHESIS B Y VIGOROUSLY GROWING OR YOUNGCELLS ~~
Assay method
Organism
Incorporation of normally missing amino acids
Man Human fibroblasts Mouse E . coli E. coli
Misincorporation at a specific codon Readthrough of nonsense mutations
Protein
AA measured
Hemoglobin Histone HI
Ile Met Met CYS
Rabbit E. coli E . coli
Histone HI Flagellin Ribosomal protein L7lL12 Hemoglobin MS2 coat protein P-Galactosidase
E. coli
Alkaline phosphatase
“Misassignmentslcodons read
CYS
Val for Ile Lys for Asn
-
Error rate“ 3 x 10-5
<
10-5
- 10-5 1.5 x
1.3-4.0
X
IOW3
3 x 10-4 2 x 10-3 1.6-3 x 10-4
2x10-3-3
X
References Popp et a/. (1976) Buchanan et (11. (1980) Medvedev (1980) Edelman and Gallant (1977a) Bouadloun et al. (1983) Loftfield and Vanderjagt (1972) Parker e t a ! . (1980) Gallant and Foley (1980); Rosenberger et a/. (1 980) Gallant and Palmer (1979); Rosenberger et a / . (1980)
1I4
T. B. L. KIRKWOOD ET AL.
3. Measurement of the Fidelity of Specific Components of the Translation Apparatus Instead of measuring the natural frequency of altered molecules in vivo, a number of attempts have been made to measure the fidelity of components of the translation apparatus isolated and tested in vitro. Attempts were made with complete protein synthesizing systems supplied with artificial templates (Buchanan et al., 1980; Wojtyk and Goldstein, 19801, but until very recently, such experiments were difficult to interpret because the in v i m error levels were orders of magnitude greater than those in vivo. Now, however, considerable improvements in the performance of in vitro protein synthesizing systems from prokaryotes have been reported. Kurland and his associates (Jelenc and Kurland, 1979) have achieved error rates as low as 10-3-10-4 in the translation of synthetic messages, and it is likely that these modified protein-synthesizing systems will prove useful tools to study translational stability in the future.
4. Amplification of Errors by Amino Acid Starvation A quite different approach to measuring the accuracy of protein synthesis in mammalian cells has been attempted by Harley et al. (1980, 1981). They have starved cells of a particular amino acid and determined the rate at which another amino acid, specified by a related codon, is incorporated in its place. Using a theoretical model of the protein synthesis system to simulate the effects of starvation (Harley et al., 1981), they extrapolated from the elevated error levels measured under starved conditions to estimate the error level occurring under normal conditions. Several factors, however, make the interpretation of such results very difficult. First, starvation in prokaryotes activates systems which markedly alter the accuracy of translation (Gallant and Foley, 1980) and it is possible that analogous responses also occur in eukaryotes. Second, their model of protein synthesis does not allow for the possibility that ribosomes, blocked in their movement because of delays at “hungry” codons, might simply fall off the message. Premature dissociation of ribosomes from the message may have a marked effect on the rates of protein synthesis used in their calculations (see Menninger, 1983). Third, their model depends quite critically on rather simplistic assumptions about the interactions of cognate and noncognate aminoacyltRNAs with the ribosome, and it makes no allowance for possible effects of competition between cognate and noncognate aminoacyl-tRNAs under unstarved conditions. The idea of using amino acid starvation to estimate error rates under normal conditions is an interesting one, but at present the estimation procedure relies so heavily on unproven assumptions that the method cannot yet be regarded as reliable. Despite claims to the contrary (Harley et al., 1980), the method is particularly unsuited to testing whether error propagation takes place, since erroneous components of the translational system may alter the acceptance rate of cognate amino acids and this cannot be measured by starvation.
CELLULAR TRANSLATION PROCESS
I15
5 . Comparison of Methods The results listed in Table I show that the fidelity of protein synthesis measured by the different methods can vary within strikingly wide limits. Such marked variations are obviously of major significance in considerations of the error propagation hypothesis. It must be remembered, however, that the experiments cited in Table I used different organisms, proteins, and methods. This makes it difficult to express the results in comparable units. When misassignments at a specific codon are measured, as for example in the misreading of mutations or the substitution of lysine for asparagine, the error rate can be expressed as a fraction of the codons read. By their ingenious use of protein chemistry, Kurland’s group (Bouadloun et al., 1983) were able to do the same when measuring the insertion of a normally missing amino acid. But in other instances where the incorporation of a normally absent amino acid was followed (methionine in histone, isoleucine in human hemoglobin), this could not be done. With no information concerning the site where the noncognate amino acid was inserted, errors have to be expressed as a fraction of all the codons in the protein message. Since only some codons are likely to be misread as methionine or isoleucine this will overestimate accuracy; it is in these cases that the highest accuracies were found. No simple correction may be applied, for the various codons related to the missing amino acid may be mistranslated at different rates. Higher levels of accuracy also tended to be found when the proteins were purified to homogeneity for the assays and the dangers of losing erroneous molecules were greatest. In the case of histone H1, insertion of a methionine may result in exclusion of the erroneous molecules from chromatin. Finally, there is the very real possibility that erroneous polypeptides are scavenged at rates which vary with the organism and the protein species. All these considerations suggest that the data in Table I may exaggerate the variability of error rates during protein synthesis. The most that can be said at present is that misassignments in vigorously growing cells are unlikely to be more frequent than 1 in 500 codons read and could be up to two orders of magnitude less frequent. ERRORS B. TRANSCRIPTION
Measuring transcriptional accuracy presents even greater problems than those associated with translational fidelity and the data available are, to say the least, extremely limited. For a clearer view of the difficulties involved, it appears useful to consider first the kinds of transcriptional mistakes that are possible. Different classes of transcriptional error can, in fact, have different implications for decoding stability. One error class would be a base misinsertion in a messenger (m) RNA. This error can be amplified, as the message undergoes successive rounds of translation, to give a number of incorrect protein molecules. Such an event would be
I16
T. B. L. KIRKWOOD ET AL.
what Hasegawa et al. (1984) have called a type 2 error and which has been discussed in a previous section. In prokaryotes a single RNA polymerase transcribes all RNA species and the accuracy of messages is likely to be the same as that of stable RNA. In eukaryotes, however, different polymerases transcribe ribosomal (r), transfer (t), and mRNA (Roeder, 1976) and the frequency of this error class will depend on the fidelity of the polymerase involved. A second class would be a base misinsertion in a tRNA or in one of the RNA components of ribosomes. In this case there will be no amplification as these species are not translated. But a base substitution in the anticodon or in another sensitive site of a tRNA could turn it into a suppressor tRNA (Ozeki et al., 1980) and this would give rise to type 1 errors. Similarly, base substitutions in rRNA would give rise to type 1 and not type 2 errors. In eukaryotes, the accuracy of tRNA and rRNA will reflect the fidelity of their respective polymerases and may be different from each other or from that of mRNA. A third class of errors would involve incorrect initiation and termination of transcription. Both these processes require that the polymerase accurately recognizes specific DNA promotor and terminator base sequences. Errors could lead to RNA molecules with altered structure and function. Recognition errors could further play a special role in cells with developmental programs. There is evidence that at least some developmental stages depend on promotor selection by RNA polymerases, the promotor choice being influenced by DNA methylation patterns (Doerfler, 1983) or by the recognition of specific u factors (Haldenwang and Losick, 1979). Finally, it is now clear that many RNA molecules must undergo posttranscriptional modifications before they become functional. Processing is essential for eukaryotic messages whose introns have to be removed with precision before they are translated (Nevins, 1983) and also for tRNA and rRNA species (Young et al., 1980). The above considerations indicate that no single method is likely to detect all the errors associated with transcription. In fact, the only data available at present concern the frequency of base misinsertions in natural messages and in products copied from synthetic DNA templates.
1. In Vivo Rates of Transcriptional Errors Direct sequencing of natural messages is far too insensitive to detect random errors and investigators have had to search for quite different approaches to measure transcriptional accuracy in growing cells. So far, only two methods with the required sensitivity have been described. One of these depends on measuring the mutation rates of RNA viruses. Mutations in RNA genomes can be so readily demonstrated as to convince investigators that the mutation rates must be much higher than in their DNA counterparts
CELLULAR TRANSLATION PROCESS
117
(Kawai and Hanafusa, 1973; Domingo et al., 1976; Darlix and Spahr, 1983). However, accurate and quantitative determinations of the rates are beset by technical difficulties. Up to the present, rate measurements appear to have been attempted only with RNA phages in E. coli (Domingo et af., 1976). The major problem with this system is that mutant, revertant, and wild-type phages have different replication rates; to extract the essential figure, the mutation rate per phage generation, one has to correct for the differences. The results put the base misinsertion rate of the phage polymerase at approximately l o p 4 . A second method, again so far used only with E. coli, depends on the readthrough of nonsense codons artificially introduced into the structural genes of enzymes. Any active enzyme formed will then be the result of either a transcriptional or a translational error. To measure transcriptional accuracy one has first to eliminate the contribution from mistranslation. This can be done by making use of some of the well-studied properties of the E . coli lac operon (Rosenberger and Foskett, 1981; Rosenberger and Hilton, 1983). The results indicate that E. coli RNA polymerase misinserts bases at rates varying between 2 X lop4 and 1 x l o p 5 , accuracy depending on the site of the mutation and the nature of the nonsense codon. Additional and wider scope for this approach should result from the recent cloning of release factor genes on multicopy plasmids (Weiss et al., 1984). A high cellular concentration of release factors greatly reduces or eliminates translational errors at nonsense codons and the residual activity in enzyme nonsense mutations will be related to polymerase errors. Perhaps the main conclusion to be drawn from these very limited data is that in vivo base misinsertion rates, in E . coli at least, can be as high as 2 x 10V4.This brings them close to translational error rates, and since they will be amplified by translation they could have marked effects on individual cells.
2 . In Vitro Rates of Transcriptional Errors RNA polymerases can be quite readily purified from a variety of sources and many aspects of their enzymology have been studied very intensively. Unfortunately for the present review, the accuracy of transcription has not been one of these aspects. The very small number of fidelity studies reported have used the purified E . colt enzyme and synthetic polydeoxyribonucleotide templates to measure the frequency of base misinsertions. The templates contained 2 out of the 4 normal bases and the incorporation of a noncognate base served to measure errors. The misincorporation frequencies appeared to be a function not only of the template and the ribonucleotide, but also of the incubation conditions and the enzyme preparation (Krakow et a l . , 1976). For example, the error frequencies for the same template and the same noncognate base (GTP into a d[A - TI,, template) have varied from 5 X ]OW4 (Bick, 1975) to 2.5 X (Springate and Loeb,
118
T. B. L. KIRKWOOD ET AL.
1975). The misincorporation rates for different bases also appear to vary over this same range, mispairing of C for U being the most frequent (Springate and Loeb, 1975). Thus, for the E . coli enzyme at least, the error rates found in the test tube correspond reasonably well with estimates from growing cells. This is encouraging for future studies with purified polymerases from different sources; in vitro studies may be the simplest way of comparing various cell types. The error rates found so far are quite high, which is in keeping with the fact that RNA polymerases show no exonucleolytic proofreading capacity.
VI. Turnover of Proteins and Messages All living cells studied so far contain a formidable number of different proteases and our knowledge of these enzymes has been extensively reviewed (Pine, 1972; Goldberg and Dice, 1974; Goldberg and St. John, 1976; Holzer and Heinrich, 1980; Hershko and Ciechanover, 1982; Makrides, 1983; these reviews cite additional reviews and symposia). As discussed in the above references, proteases fulfill a wide spectrum of functions. These include the nonspecific turnover of cell components, the processing of precursor molecules, the transport of proteins, the breakdown of ingested proteins, the specific hydrolysis of at least some abnormal proteins, and morphogenetic changes in organisms as different as bacteriophages and mammalian sperm. It has been mentioned already that scavenging mechanisms for the preferential elimination of abnormal molecules would result in increased translational stability, and in this section we make a brief review of the data and we consider the further possibility that nonspecific turnover may have effects on stability too. There are basically two ways in which the average rates of nonspecific turnover of proteins and their messages (mRNAs) can be varied. In the first, the rates of turnover of both kinds of molecule are altered in concert. If this happens the essential kinetics of error propagation remain the same, but the generation time of the translation apparatus is changed in relation to chronological time. This means that if there is a given probability of destabilization occurring randomly in any one generation of the translation apparatus, the probability of destabilization per unit of chronological time will either increase or decrease in the same direction as the turnover rate. The second alternative is that the ratio of the turnover rates for proteins and messages changes. If, for example, the protein turnover rate is raised relative to the message turnover rate (and other factors remain the same), the number of protein molecules synthesized from each molecule of mRNA will increase. This would have the effect of making individual type 2 errors both less frequent and more serious, so the distribution of error levels within a population of cells will be more widespread. In general, a higher ratio of
CELLULAR TRANSLATION PROCESS
119
message/protein turnover rates will give rise to a population which is more homogeneous, while a lower ratio will give rise to a population in which a few individuals have markedly higher error levels than the average. In mammalian cells the ratio of message/protein turnover rates is substantially lower than in exponentially growing E . coli (for data, see Lewin, 1974; Makrides, 1983), and it is possible that this difference is related to differences in translational stability. In determining the relevance of protein turnover to error propagation the most important points to be clarified experimentally are whether cells can specifically degrade erroneous proteins and whether they can increase their basal proteolysis in response to an error stress. Both prokaryotes and eukaryotes produce a number of proteases which show marked specificity. They are presumed to be responsible, under normal conditions, for the processing or breakdown of specialized proteins with short half-lives (Holzer and Heinrich, 1980; Hershko and Ciechanover, 1982). They are also capable of hydrolyzing highly abnormal proteins such as those produced by the addition of amino acid analogs. It is still quite unclear how these enzyme systems distinguish their substrates from other proteins; however, it has been established that by no means all altered proteins are degraded at accelerated rates. Until it is known which alterations increase the probability of a protein to be degraded and by how much, the mere existence of specific proteases does not help very much in assessing the relation between protein turnover and translational stability. A different approach to clarifying the importance of protein degradation in error feedback would be to see if turnover increases under conditions where abnormal proteins might accumulate, such as in senescent cells. Experiments of this kind involve considerable technical difficulties and these have been reviewed by Makrides (1983). In addition, both prokaryotic and eukaryotic cells can increase their nonspecific turnover of proteins markedly when subjected to various, mostly nutritional stresses (Hershko and Ciechanover, 1982). Presumably because of these complications, senescence has been reported to lead to increased protein turnover, decreased protein turnover, and no change in protein turnover (Makrides, 1983). The ability of cells to influence error feedback by selective proteolysis thus remains a potentially important unknown.
VII. Evidence from Prokaryotes Some of the most interesting data on translational stability come from a series of experiments on E . coli. In these studies, low levels of aminoglycoside antibiotics, such as streptomycin, have been used to increase protein synthetic errors artifically in order to find out whether a normally stable cell can be pushed toward error catastrophe. Streptomycin at low concentration acts only on ribosomes and decreases their fidelity (Gorini, 1974).
120
T. B . L. KIRKWOOD ET AL.
The first experiment on these lines was by Branscomb and Galas (1975). They grew E . coli in medium containing low concentrations of streptomycin and reported that after addition of the drug they detected a progressive increase in the proportion of heat-labile P-galactosidase, together with a delayed progressive decrease in culture growth rate and increase in the number of visibly sick cells. The decrease in growth rate commenced only after about 15 cell population doublings and prior to this time the cells seemed outwardly normal, suggesting that streptomycin was not exerting an immediate toxic effect. In all cases, recovery of the cultures occurred eventually as cells took over the population which were resistant to low level streptomycin. Edelman and Gallant (1977b) similarly treated cultures of E . coli with low concentrations of streptomycin and measured the subsequent misincorporation of radiolabeled cysteine into the protein flagellin which normally contains no cysteine. They observed that following treatment with streptomycin, the misincorporation of cysteine increased over time by up to a 40-fold factor, but thereafter remained stable. They concluded from the failure of this large increase in errors to push the cells into translational instability that the translation process in E. coli is very stable. In subsequent studies, a similar conclusion was reached using readthrough of a specific nonsense codon as a measure of errors (Gallant and Palmer, 1979; Gallant and Foley, 1980). Rosenberger et al. (1980) have employed a similar readthrough assay, however, and obtained different error propagation kinetics, strongly suggestive of error catastrophe. Between sublethal and obviously lethal concentrations of streptomycin, there was a range where E. coli cells grew for 15 or more generations in the presence of the antibiotic and then lost viability (Rosenberger, 1982). During this period the error rate appeared to increase exponentially, with a maximum at the point where massive cell death occurred. The possibility of obtaining these different kinds of result is, in fact, a prediction of error propagation models (see Section 11), and these results suggest therefore that the kinetics of error accumulation after a perturbation are a powerful tool for determining the parameters of translational stability (Kirkwood, 1980; Gallant and Foley, 1980; Gallant and Prothero, 1980). Gallant and Prothero (1980) have, in fact, used data from studies with streptomycin in E . coli to conduct tests on the mathematical models of error propagation. They concluded that their data were consistent with the model of Eq. (2) (Section 11) only for parameter values far removed from any risk of instablity. They further observed that, under these circumstances, Eq. (2) is little different in its properties from the simpler Eq. (1). Given that the data they analyzed showed no evidence for translational instability, the latter conclusion is unsurprising, since the difference between the models is only apparent in the region where Eq. (2) yields downward curvature toward qthreshold. In the vicinity of qstableEqs. ( I ) and (2) are barely distinguishable (see, also, Kirkwood, 1980), but if translational instability can be induced, then Eq. ( 1 ) is not adequate. In
CELLULAR TRANSLATION PROCESS
121
fitting Eq. (1) to the data, Gallant and Prothero (1980) estimated a value of CY 5 0.5 and it is interesting that although this is well within the range producing stable translation, it is large enough to indicate that a significant amount of translational error feedback is taking place.
VIII. Evidence from Eukaryotes A. FUNGI
As we have seen, the theory of error propagation is able, in principle, to explain the fact that some populations of organisms can survive indefinitely, whereas others die out. This is well illustrated in the fungi, where most species can propagate themselves indefinitely by asexual means, but some have vegetative cells with limited growth potential. The most thoroughly studied species in this group is Podospora anserina (Marcou, 1961). Individual haploid ascospores give rise to populations of multinucleate cells, or hyphae, which grow at a constant rate for a given period of time, but these invariably become senescent and further growth ceases. Just as in higher organisms, survival depends on sexual reproduction, which by-passes the aging process. The studies with Podospora, as well as those with Aspergillus gluucus (Jinks, 1959), are of particular importance, since they demonstrate that senescence has a cytoplasmic, rather than a nuclear origin, and also that it is dominant or invasive in heterokaryons between normal and aged cells. It has also been shown in Podospora that longevity is progressively reduced as the incubation temperature is increased. In relation to the instability of the translation process, two important questions arise. First, is the finite growth of vegetative cells of fungi due to the accumulation of errors in macromolecules? Second, will genetic or environmental factors which increase error levels change a potentially immortal population of cells into one which has limited growth potential? The answer to the first question is as yet unknown, but recently it has been demonstrated that senescence is accompanied by gross changes in the mitochondria1 genome, particularly the amplification of defined sequences and the loss of others (Cummings et a/., 1979; Jarnet-Vierny et al., 1980; Belcour et al., 1981; Wright et al., 1982). This explains the cytoplasmic basis of the phenotype, the dominance of senescence in heterokaryons, and the cessation of growth in an obligate aerobic organism. It does not, however, necessarily account for the early commitment to the senescent condition, which is well documented (Marcou, 1961), or its inevitability in a population of cells where selection for respiratory competence should be possible. Abnormalities in mitochondria may be one important consequence of a general failure to maintain the integrity of macromolecules. It has been shown that low
122
T. B . L. KIRKWOOD ET AL
concentrations of amino acid analogs accelerate the senescence of Podospora, suggesting that defective proteins may also be involved in natural aging (Holliday, 1969), but no detailed biochemical studies of possible changes in proteins have so far been reported. Much more information has been obtained with the related fungus, Neurospora crassa, which grows indefinitely, although eventually the growth rate may become quite irregular (McDougall and Pittenger, 1966; Bertrand et al., 1968). Two mutants of Neurospora have been isolated which have characteristics of senescence similar to Podospora. The first is called nd (natural death) which was originally discovered by Sheng (1951). nd resembles wild-type strains of Podospora in that its longevity is greatly increased at low incubation temperatures. The hypothesis that nd cultures die through error propagation was tested by experiments on phenotypic suppression (Holliday , 1969). Certain auxotrophic mutants exist in yeast and Neurospora, which will grow on minimal medium in the presence of error-promoting agents such as 5-fluorouracil (FU), 8azaguanine, or paromomycin (Barnet and Brockman, 1962; Sundaram, 1967; Palmer et al., 1979; Singh et al., 1979; Masurekar et al., 1981). nd was combined with an adenine auxotroph (ade3B) which was known to grow slowly in the presence of FU, in the absence of adenine. If nd was indeed an errorpromoting strain, then it should allow the ade3B mutant to grow in the absence of both adenine and FU. This was shown to be the case. Moreover, during the growth of nd, the extent of suppression of ade- increased, at least during the first half of its lifespan, suggesting that the level of errors was also increasing. In another experiment, the growth rate of the ade- nd+ strain was accurately monitored in the presence of FU. Since growth depends on translational errors, there will initially be selection for the more error-prone cells. Furthermore, with error feedback growth should accelerate, but then slow down as errors become inhibitory. An accelerating growth rate followed by death was in fact observed (Holliday, 1969). It has also been shown that nd accumulates altered glutamic dehydrogenase (GDH), as shown by an increase in its heat-lability and the accumulation of inactive enzyme cross-reacting material (Lewis and Holliday , 1970). Whereas the biochemical basis of the defect in nd is as yet unknown, another mutant of Neurospora which shows clonal aging has been characterized. Printz and Gross (1967) found that leu-5 contained abnormal enzyme molecules, and presented evidence that its phenotype was due to the synthesis of an altered leucyl tRNA synthetase, presumably with reduced specificity. This phenotype is more extreme at 35 than 25"C, and Printz and Gross reported that cultures had limited growth at the higher temperature. The possibility that this limit to growth was due to an error catastrophe was tested in experiments in which an antiserum was used to measure the proportion of inactive GDH molecules (Lewis and Holliday, 1970). On shifting cultures from 25 to 35"C, a proportion of inactive
CELLULAR TRANSLATION PROCESS
123
molecules appeared within a few hours, presumably due to an initial increase in translational errors, but this stayed constant during a 48-hour period of growth. Subsequently, the proportion of inactive molecules increased rapidly and growth ceased after 78 hours. The kinetics of the appearance of an inactive GDH strongly suggested that an error catastrophe was responsible for the death of the culture. It is also important to note the leu-5 grown at 35°C was highly mutable, which suggested that the replication of DNA was much less accurate than at 25"C, or in wild-type strains. B. MAMMALIAN CELLS The limited replication of mammalian fibroblasts in tissue culture is widely accepted as a model of cellular aging (Hayflick and Moorhead, 1961; Hayflick, 1965, 1977; Cristofalo and Stanulis-Praeger, 1982). Cultures obtained from human donors of increasing age show altered colony size distributions (Smith et al., 1978) and exhibit a significant reduction in overall division potential (Hayflick, 1965; Schneider and Mitsui, 1976; Goldstein et al., 1978; Martin et al., 1970, 1981), while fibroblasts derived from different mammalian species show a positive correlation of in vitro lifespan with specific longevity (Rohme, 1981). Also, cells from patients with Werner's syndrome, a genetic disorder characterized by many features of accelerated aging, have greatly reduced lifespans (Martin et al., 1970; Salk et al., 1981; Thompson and Holliday, 1983). Each of these findings supports the association of the limited division potential of fibroblasts with the aging of the body as a whole, although it may still be argued that the cessation of growth is due to a process of terminal differentiation in which the cells merely enter a noncycling state (Martin et aE., 1974; Bell et al., 1978). In this review, we shall be concerned only with testing the hypothesis that fibroblasts cease to divide as a consequence of translational instability, although if this view is correct it will ultimately be necessary to establish more clearly the link between cell aging in vitro and in vivo (see, for example, Kirkwood, 1984b). A direct prediction of the error catastrophe hypothesis is that fibroblast cultures should be found to contain progressively more erroneous protein as they traverse their in vitro lifespan. One of the first tests was to screen for an increase in the fraction of heat-labile enzymes (Holliday and Tarrant, 1972), and significant reduction in the heat-stability of the enzymes, glucose-&phosphate dehydrogenase (G6PD) and 6-phosphogluconate dehydrogenase (6PGD), was detected during the last 15 passages of the lifespan of foetal lung fibroblast strain, MRC-5. Many naturally occurring variants of G6PD are known in man and about half of them are heat-labile (McKusick, 1978). Therefore random amino acid substitutions should also produce a proportion of heat-labile molecules. Studies of naturally occurring variants have also shown that they sometimes have altered substrate specificity, and it was confirmed that the altered fraction of enzyme in
124
T. B. L. KIRKWOOD ET AL.
senescent fibroblasts was preferentially able to use the analog substrate, deoxyglucose-6-phosphate. Evidence that the unstable fraction is due to errors in synthesis comes from studies with the RNA base analog, 5-fluorouracil (Holliday and Tarrant, 1972), and more recently, with the aminoglycoside antibiotic, paromomycin (Holliday and Rattan, 1984), which has been shown to reduce the fidelity of translation by eukaryotic ribosomes in vitro (Wilhelm et al., 1978a,b; Buchanan et al., 1980). Results with G6PD and their interpretation as evidence that errors in proteins are occurring are, however, controversial (Pendergrass et al., 1976; Duncan et a/. , 1977; Kahn et al., 1977; for a review, see Holliday and Thompson, 1983), and at best the evidence is only indirect. Another prediction of the protein error theory is that antibiotics such as paromomycin (Pm), if it increases translation errors in vivo, should accelerate many of the normal features of fibroblast aging. It was found that MRC-5 cells grown in the presence of the antibiotic are unaffected for many generations of growth, but then adopt the morphological characteristics of senescence much sooner than control cultures, including the accumulation of autofluorescent age pigments (Holliday and Rattan, 1984). Moreover, the long-term effects of Pm are not removed by returning the cells to normal medium. It was also shown that as the cells age they appear to become progressively more sensitive to the effects of Pm. These results, together with the earlier experiments with FU, and more recent ones with the gentamycin derivative, G418 (Holliday, unpublished observations), are certainly in accord with the error theory. Immunological methods have also been used to search for inactive enzyme molecules (CRM) in senescent cells. Lewis and Tarrant (1972) reported that the ratio of lactic dehydrogenase (LDH) specific activity to CRM declines steeply during the senescence of MRC-5, but the antiserum used in these studies was not obtained from purified enzyme, so nonspecific precipitation may have occurred. In another study, Shakespeare and Buchanan (1978) raised antiserum against purified phosphoglucose isomerase (PGI), and used a technique which would have detected about 15% inactive molecules. Neither inactive CRM nor heatlabile PGI was detected in cell-free extracts from senescent cultures. Although the error theory predicts that all proteins should be altered to some extent in senescent cells, it was pointed out by Lewis (1972) that one might well expect major differences in the extent of this effect. For example, some proteins containing amino acid substitutions may be rapidly degraded by proteases, whereas others may be much more resistant (see Section VI). More specific measurement of the accuracy of protein synthesis during the in vitro lifespan of fibroblast cultures has been attempted by Buchanan and Stevens (1978), Buchanan et al. (1980), Wojtyk and Goldstein (1980), and Harley et al. (1980). These studies have each failed to detect any significant increase in error frequency during aging of the cultures, but in all cases limitations in the technique for measuring errors have meant that the negative result does not constitute refutation of the error catastrophe hypothesis. Buchanan and Stevens (1978)
CELLULAR TRANSLATION PROCESS
125
attempted measurement of misincorporation of radiolabeled methionine into histone H1 but, since protein synthesis itself was inhibited when highly radioactive methionine was used, they were unable to detect misincorporation in either old or young cells. Buchanan et al. (1980) developed a cell-free protein synthesis system using translation systems obtained from fibroblasts of different ages to decode an artificial poly(U) template. Even under standard optimized conditions, however, the resulting in vitro error level of approximately lo-* was very much higher than the expected in vivo level. Thus, if the real error level in young cells is, say, and in old cells is 10W3, this difference might very well not be discernible. Wojtyk and Goldstein (1980) used a similar method, but without the standardization employed by Buchanan et al. (1980); indeed, in a later study by Wojtyk and Goldstein (1982) the protein synthesizing activity varied by a factor of 800-fold. Finally, the limitations of the method employed by Harley et al. have been described above (Section V). More incisive evidence against the error catastrophe hypothesis comes from studies using viruses as probes of errors in macromolecular synthesis. Holland et al. (1973) infected young and old populations of fibroblasts with three types of virus and found no specific differences in yield, heat stability, or in the case of polio virus, mutation frequency. This suggests that error propagation is not significantly affecting the pathways of intracellular information transfer. However, most viruses greatly overproduce the subunits they require for assembly of completed particles, and it is possible that strong selection against the inclusion of defective molecules in the assembly process will mask any moderate increase in error level. In this connection, it may be of significance that in studies with polio and vesicular stomatitis virus Pitha et al. (1974, 1975) were unable to detect alterations in heat stability or yield until the last passage before the cultures ceased growing, when virus yield was sharply reduced. Fulder (1977) examined the reversion frequency of three ts mutants of Herpes simplex virus, which were grown in young and old MRC-5 fibroblasts and found that in one case, reversion frequency was elevated in senescent cells; in another, it was reduced, and in a third, it was unchanged. Experiments with virus probes offer a very interesting approach to the study of translational stability, but clearly they need to be interpreted with caution. The best evidence for increasing protein errors in fibroblasts comes from studies on the fidelity of DNA polymerase OL (Linn et al., 1976; Murray and Holliday, 1981). If error propagation does occur, random errors in the synthesis of DNA polymerases should affect fidelity of DNA replication and this should result in an increasing rate of mutation, Fulder and Holliday (1975) showed previously that during the in vitro lifespan of fibroblasts an approximately exponential rise in the frequency of variant cells with an increased level of G6PD could be detected (see also Fulder, 1979). [It should be noted that later studies by Gupta (1980), using better defined mutational markers, demonstrated only a linear increase in mutation frequency, but Gupta’s studies extended only through
126
T. B . L. KIRKWOOD ET AL.
the first 70% of culture lifespan and therefore excluded the period during which the rise detected by Fulder and Holliday became clearly distinguishable from a straight line.] Linn et al. (1976) examined partially purified polymerase (Y from young and old MRC-5 cells using an artificial DNA template and found that the polymerase from old cells was several times more error-prone than the polymerase from young cells. Murray and Holliday (1981) extended these studies by testing enzyme from cell populations of various ages in a better defined assay system and observed an approximately exponential increase in error frequency with cell age. DNA polymerase y was also affected. Much of this study was concerned with various important controls, which effectively ruled out a variety of possible experimental artifacts. Although these studies establish that a key information-handling system does indeed become more error-prone during the lifespan of fibroblast populations, it remains to be proven that this is a consequence of translational instability. It is, for instance, possible that proteolytic cleavage of the enzyme yields active, but less accurate derivatives (Krauss and Linn, 1981; Hubscher et al., 1981). It is striking that the changes that have been documented during the long-term growth of cultured human fibroblasts have most often been detected only at the latter part of the lifespan (usually the last 5-15 population doublings in a total in vitro lifespan of 50-65 population doublings), which is much more compatible with a progressive or exponential change in phenotype, rather than a linear one. This is true for chromosome abnormalities (Saksela and Moorhead, 1963; Thompson and Holliday, 1975; Miller et al., 1977), the frequency of cell variants (Fulder and Holliday, 1975; Fulder, 1979), increased autofluorescence (Rattan et al., 1982), proportion of heat-labile G6PD and 6PGD (Holliday and Tarrant, 1972) or LDH cross-reacting material (Lewis and Tarrant, 1972), and the reduced fidelity of DNA polymerase a (Murray and Holliday, 1981). All these studies are compatible with the view that aging is associated with the progressive breakdown in the structural integrity of macromolecules (see Holliday, 1984). Numerous studies have been reported on tests of the error theory of aging in whole animals, but since these require additional, and sometimes complex consideration of the growth dynamics of cell populations in vivo, we shall not attempt to review them here. Recent reviews have been provided by Gallant (1981) and Laughrea (1982), and more extensive discussion is to be published elsewhere (Holliday and Kirkwood, 1984).
IX. Conclusions During the two decades since Orgel (1963) first pointed out the risk of instability in the cellular translation process, this concept has generated a great deal
CELLULAR TRANSLATION PROCESS
127
of interest. This is not at all surprising, since the idea relates both to the origin of life and to its finitude, namely, aging. Error propagation would have been the major hurdle to be crossed in the early evolution of a genetic decoding apparatus, and it is clear from the elaborate enzymic editing mechanisms in present-day cells that, on the one hand, high fidelity macromolecular synthesis is costly, and on the other, the price is worth paying. This makes it plausible, at least, that multicellular organisms have found it evolutionarily advantageous to trim their investment in somatic translational accuracy to the point where error catastrophe is a real risk (Kirkwood, 1977). The fact that the theory has proved as controversial as it has is due, perhaps, to the complexity of its predictions. Initial studies sought evidence of major changes in proteins, when in reality it now seems more likely that quite modest increases in error rates could seriously incapacitate and eventually kill a cell. For instance, a small fraction of erroneous membrane ATPases may be sufficient to collapse ion gradients, a single altered repressor molecule may block transcription of an essential gene, or a minor proportion of nucleases or topoisomerases with altered specificities could irreversibly damage the genome. If each ribosomal protein pool contained only 2% of malfunctioning molecules, the assembly of a ribosome with perfect copies of its 50 different proteins would occur only one time in three. For these reasons, results showing that erroneous proteins do not accumulate during clonal growth need to be interpreted with great caution when insensitive or unproven methods have been used. Models of the dynamics of error propagation have a useful role to play in elucidating the general stability characteristics of the translation process, and they may also aid in the design and interpretation of critical experiments (Kirkwood, 1980). Present models are of limited realism, however, and future models will be needed which can take account of error feedback at different levels of information transfer and which represent more flexibly the effect of protein synthetic errors on enzyme activity and specificity. A recent, interesting application of error feedback theory was made by Hasegawa et al. (1984) who argued that the more rapid evolution of tRNAs and ribosomal proteins in animal mitochondria than in their cytoplasmic counterparts may have arisen because animal mitochondria import information-handling enzymes from the cytoplasm and are, therefore, less at risk of error catastrophe. This places the mitochondria1 translation process under less strict evolutionary constraint. Weighing evidence for and against the natural occurrence of error catastrophe is likely, at this juncture, to prove premature. On the one hand, negative reports are common, and these have led some investigators to conclude that the error theory is no longer tenable. However, as we have remarked already, the limitations on techniques mean that the negative results which have so far been obtained fall a good deal short of serious falsification of any major prediction of the theory. On the other hand, the positive evidence, although in some instances
128
T. B. L. KIRKWOOD ET AL.
quite strongly suggestive, is derived mainly from indirect methods and therefore needs further examination. The present situation can best be summed up by observing that the theory is now seen to be more complex than was originally recognized. Although it is known from the quantum nature of chemical reactions and the behavior of mutant enzymes that error feedback must take place in the cellular translation process, it is still not known whether in present-day cells this feedback is so small as to be unimportant, whether it may actually be the primary mechanism of cellular aging, or whether the truth lies somewhere in between. Progress in theoretical understanding and in improving techniques is continually being made, however, and in E . coli it does appear that error feedback is sufficiently great to be measurable. Given the similarities between the decoding systems of all living organisms, it may well turn out that the risk of translational instability has considerable significance in most species.
REFERENCES Alberts, B., Bray, D., Lewis, J . , Raff, M., Roberts, K., and Watson, J . D. (1983). “Molecular Biology of the Cell.” Garland, New York. Barnet, W. E . , and Brockman, H . E. (1962). Biochem. Biophys. Res. Commun. 7, 199-203. Belcour, L . , Begel, O., Mosse, M.-O., and Vierny, C . (1981). Curr. Genet. 3, 13-22. Bell, E., Marek, L. F., Levinstone, D. S., Merrill, C., Sher, S . , Young, I. T., and Eden, M. (1978). Science 202, 1158-1 163. Bertrand, H . , McDougall, K. J . , and Pittenger, T. H . (1968). J . G m . Microbiol. 50, 337-350. Bick, M. D. (1975). Nucleic Acids Res. 2 , 1513-1523. Bouadloun, F., Donnen, D., and Kurland, C. G. (1983). E M B O J . 2, 1351-1356. Branscomb, E. W . , and Galas, D. J. (1975). Nature (London) 254, 161-163. Buchanan, J. H . , and Stevens, A . (1978). Mech. Ageing Dev. 7, 321-334. Buchanan, J . H., B u m , C. L., Lappin, R. I., and Stevens, A. (1980). Mech. AgeingDev. 12, 339353. Bukhari, A . I . , and Zipser, D. (1973). Nature (London) New Biol. 243, 238-241. Burnet, F. M. (1974). “Intrinsic Mutdgenesis: A Genetic Approach to Ageing.” Wiley, New York. Crick, F. H . C. (1968). J . Mol. Biol. 38, 367-379. Crick, F. H . C., Brenner, S . , Klug, A., and Pieczenik, G. (1976). Orig. Life 7, 389-397. Cristofalo, V. J . , and Stanulis-Praeger, B. M. (1982). Adv. Cell Cult. 2 , 1-68. Cummings, D. J., Belcour, L., and Grandchamp, C. (1979). Mol. Gen. Gene!. 171, 239-250. Curtis, H . 1 . (1966). “Biological Mechanisms of Aging.” Thomas, Springfield, Illinois. Darlix, J.-L., and Spahr, P.-F. (1983). Nucleic Acids Res. 11, 5953-5967. Doerfler, W. (1983). Annu. Rcv. Biochem. 52, 93-124. Domingo, E., Flavell, R. A., amd Weissman, C. (1976). Gene 1, 3-25. Duncan, M. R., Dell’Orco, R. T., and Guthrie, D. C. (1977). J . Cell. PhVsiol. 93, 49-56. Edelman, P., and Gallant, J . A. (1977a). Cell 10, 131-137. Edelman, P., and Gallant, J . A. (1977b). Proc. Nut/. Acad. Sci. U . S . A . 74, 3396-3398. Eigen, M., and Schuster, P. (1979). “The Hypercycle.” Springer-Verlag, Berlin and New York. Fersht, A. R. (1981). Proc. R. Soc. London Ser. B 212, 351-379. Fersht, A. R . , and Dingwall, C. (1979). Biochemisrry 18, 2627-2631.
CELLULAR TRANSLATION PROCESS
129
Fulder, S. J. (1977). Mech. Ageing Dev. 6, 271-282. Fulder, S. 3. (1979). Mech. Ageing Dev. 10, 101-1 15. Fulder, S . J., and Holliday, R. (1975). Cell 6, 67-73. Gallant, J. A. (1981). In “Biological Mechanisms of Aging” (R. T. Schimke, ed.), pp. 373-381. Publication No 81-2194. National Institute of Health, Bethesda, Maryland. Gallant, J. A., and Foley, D. (1980). In “Ribosomes: Structure, Function and Genetics” (G. Chambliss, G. R. Craven, J. Davies, L. Kahan, and M. Nomura, eds.), pp. 615-640. Univ. Park Press, Baltimore, Maryland. Gallant, J. A., and Palmer, L. (1979). Mech. Ageing Dev. 10, 27-38. Gallant, J. A,, and Prothero, J. (1980). J . Theor. Biol. 83, 561-578. Gershon, D. (1979). Mech. Ageing Dev. 9, 189-196. Goel, N. S., and Ycas, M. (1975). J. Theor. Biol. 54, 245-282. Goldberg, A. L. (1972). Proc. Natl. Acad. Sci. U . S . A . 69, 422-426. Goldberg, A. L., and Dice, J. F. (1974). Annu. Rev. Biochem. 43, 835-869. Goldberg, A. L., and St. John, A. C. (1976). Annu. Rev. Biochem. 45, 747-803. Goldstein, S., Moerman, E. J., Soeldner, J. S., Gleason, R. E., and Barnett, D. M. (1978). Science 199, 781-782. Gorini, L. (1970). Annu. Rev. Gener. 4, 107-134. Gorini, L. (1974). I n “Ribosomes” (M. Nomura, A. Tissieres, and P. Lingyel, eds.), pp. 791-803. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Gupta, R. S. (1980). J. Cell. Physiol. 103, 209-216. Haldenwang, W. C . , and Losick, R. (1979). Nature (London) 282, 256-260. Hamilton, W. D. (1966). J. Theor. Biol. 12, 12-45. Harley, C. B., Pollard, J. W., Chamberlain, J. W . , Stanners, C. P . , and Goldstein. S. (198O).Proc. Natl. Acad. Sci. U.S.A. 77, 1885-1889. Harley, C. B., Pollard, J. W., Stanners, C. P., and Goldstein, S. (1981). J . Biol. Chem. 256, 10786- 10794. Hasegawa, M., Yano, T., and Miyata, T. (1984). J . Mol. Evol., 20, 77-85. Hartman, H. (1975). Orig.Life 6, 423-427. Hayflick, L. (1965). Exp. Cell Res. 37, 614-636. Hayflick, L. (1977). Handb. Biol. Aging pp. 159-186. Hayflick, L., and Moorhead, P. S. (1961). Exp. Cell Res. 25, 285-621. Hershko, A., and Ciechanover, A. (1982). Annu. Rev. Biochem. 51, 335-364. Hoffmann, G. W. (1974). J . Mol. B i d . 86, 349-362. Holland, J. J., Kohne, D., and Doyle, M. V. (1973). Nature (London) 245, 316-319. Holliday, R. (1969). Nature (London) 221, 1224-1228. Holliday, R. (1975). Fed. Proc. Fed. Am. Soc. Exp. Biol. 34, 51-55. Holliday, R. (1984). Monogr. Dev. Biol. 17, 60-77. Holliday, R., and Kirkwood, T. B. L. (1981). J. Theor. Biol. 93, 627-642. Holliday, R., and Kirkwood, T. B. L. (1983). J. Theor. Biol. 103, 329-330. Holliday, R., and Kirkwood, T. B. L. (1984). “Error Theories of Ageing.” Oxford University Press, London and New York, in preparation. Holliday, R . , and Rattan, S. I. S . (1984). Monogr. Dev. Biol. 17, 221-233. Holliday, R., and Tarrant, G. M. (1972). Nature (London) 238, 26-30. Holliday, R., and Thompson, K. V . A. (1983). Gerontology 29, 89-96. Holzer, H., and Heinrich, P. C. (1980). Annu. Rev. Biochem. 49, 63-91 Hopfield, J. J. (1974). Proc. Nurl. Acad. Sci. U.S.A. 71, 4135-4139. Hiibscher, U., Spanos, A , , Albert, W . , Grummt, F., and Banks, G. R. (1981). Proc. Nurl. Acad. Sci. U.S.A. 78, 6771-6775. Jarnet-Vierny, C . , Begel, O., and Belcour, L. (1980). Cell 21, 189-194.
I30
T. B. L. KIRKWOOD ET AL.
Jelenc, P. C., and Kurland, C. G. (1979). Proc. Natl. Acad. Sci. U.S.A.. 76, 3174-3178. links, J. L. (1959). J . Gen. Microbiol. 21, 397-409. Kahn, A., Guillouzo, A., Liebovitch, M. P., Cottreau, D., Bourel, M., and Dreyfus, J.-C. (1977). Biochem. Biophys. Res. Commun. 77, 760-766. Kawai, S . , and Hanafusa, H. (1973). Proc. Natl. Acad. Sci. U.S.A. 70, 3493-3497. Kirkwood, T. B. L. (1977). Nature (London)270, 301-304. Kirkwood, T. B. L. (1980). J . Theor. B i d . 82, 363-382. Kirkwood, T. B. L. (1981). I n “Physiological Ecology: An Evolutionary Approach to Resource Use” (C. R. Townsend and P. Calow, eds.), pp. 165-189. Blackwell, Oxford. Kirkwood, T. 8. L. (1984a). Handb. B i d . Aging, in press. Kirkwood, T. B. L. (1984b). Monogr. Dev. B i d . 17, 9-20. Kirkwood, T. B . L., and Cremer, T. (1982). Hum. Genet. 60, 101-121. Kirkwood, T. B. L., and Holliday, R. (1975a). J . Mol. Biol. 97, 257-265. Kirkwood, T. B. L., and Holliday, R. (1975b). J . Theor. B i d . 53, 481-496. Kirkwood, T. B. L., and Holliday, R. (1979). Proc. R. Soc. London Ser. B. 205, 531-546. Krakow, J. S . , Rhodes, G., and Jovin, T. (1976). In “RNA Polymerase” (R. Losick and M. Chamberlin, eds.), pp. 127-158. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Krauss, S. W., and Linn, S. (1982). Biochemistry 21, 1002-1009. Laughrea, M. (1982). Exp. Gerontol. 17, 305-317. Lewin, B . (1974). “Gene Expression,” Vol. 2, pp. 266-268. Wiley, New York. Lewis, C . M. (1972). Mech. Ageing Dev. 1, 43-47. Lewis, C. M., and Holliday, R. (1970). Nature (London) 228, 877-880. Lewis, C. M., and Tarrant, G. M. (1972). Nature (London) 239, 316-318. Linn, S., Kairis, M., and Holliday, R. (1976). Proc. Natl. Acad. Sci. U.S.A. 73, 2818-2822. Loftfield, R. B . (1963). Biochem. J . 89, 82-92. Loftfield, R. B., and Vanderjagt, D. (1972). Biochem. J . 128, 1353-1356. Maaloe, 0. (1979). In “Biological Regulation and Development” (R. Goldberger, ed.), Vol I , pp. 487-537. Plenum, New York. Makrides, S. C. (1983). B i d . Rev. 58, 343-422. Marcou, D. (1961). Ann. Sci. Nut. Bot. 11, 653-764. Martin, G. M.. Sprague, C. A., and Epstein, C. J. (1970). Lab. Invesr. 23, 86-92. Martin, G. M., Sprague, C. A., Norwood, T. H., and Pendergrass, W. R. (1974). Am. J . Pathol. 74, 137-154. Martin, G. M., Ogbum, C. E., and Sprague, C . A. (1981). In “Aging: A Challenge to Science and Scoiety. Vol 1. Biology” (N. W. Shock and M. Marois, eds.), pp. 124-135. Oxford Univ. Press, London and New York. Masurekar, M., Palmer, E., Ono, B. I . , Wilhelm, J . M., and Sherman, F. (1981). J . Mol. B i d . 147, 381-390. Maynard Smith, J. (1962). Proc. R. Soc. London Ser. B 157, 115-127. Maynard Smith, J. (1976). Q. Rev. Biol. 51, 277-283. McDougall, K. M., and Pittenger, T. H. (1966). Genetics 54, 551-565. McKusick, V. A. (1978). “Mendelian Inheritance in Man,” 5th ed., pp. 732-745. Johns Hopkins Univ. Press, Baltimore, Maryland. Medawar, P. B. (1946). Mod. Quart. 2 (New Ser.), 30-49. Medawar, P. B. (1952). “An Unsolved Problem in Biology.” Lewis, London. (Reprinted in “The Uniqueness of the Individual.” Methuen, London, 1957). Medvedev, Zh.A. (1962). In “Biological Aspects of Aging”, ( N . W. Shock, ed.), pp. 255-266. Columbia Univ. Press, New York. Medvedev, 2h.A. (1980). Mech. Ageing Dev. 14, 1-14.
CELLULAR TRANSLATION PROCESS
131
Medvedev, Zh.A., and Medvedeva, M. N. (1978). Gerontologist 18, 100. Menninger, J. R. (1983). J . Mol. Biol.,l71, 383-399. Miller, R. C., Nichols, W. W., Pottash, J . , and Aronson, M. M. (1977). Exp. Cell. Res. 110, 6373. Miller, S . L., and Orgel, L. E. (1973). “The Origins of Life on Earth.” Prentice Hall, New York. Mizutani, H., and Ponnamperuma, C. (1977). Orig. Life 8, 183-219. Morley, A. A. (1982). J . Theor. Biol. 98, 469-474. Murray, V., and Holliday, R. (1981). J . Mol. B i d . 146, 55-76. Nevins, I. R. (1983). Annu. Rev. Biochem. 52, 441-466. Ninio, J. (1975). Biochimie 57, 587-595. Orgel, L. E. (1963). Proc. Natl. Acad. Sci. U.S.A. 49, 517-521. Orgel, L. E. (1970). Proc. Narl. Acad. Sci. U.S.A. 67, 1476. Orgel, L. E. (1973). Nature (London) 243, 44-445. Ozeki, H., Inokuchi, H., Yamao, F., Kodaira, M., Sakano, H., Ikemura, T., and Shimura, Y. (1980). In “Transfer RNA” (D. Soll, J . M. Abelson, and P. R. Schimmel, eds.), pp. 341-362. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Palmer, E., Wilhelm, J . M., and Sherman, F. (1979). Nature (London) 277, 148-149. Parker, J., Johnston, T. C., and Borgia, P. T. (1980). Mol. Gen. Gener. 180, 275-281. Pendergrass, W. R., Martin, G. M., and Bornstein, P. (1976). J . Cell. Physiol. 87, 3-14. Pine, M. J. (1972). Annu. Rev. Microbiol. 26, 103-126. Pitha, J . , Adams, R., and Pitha, P. M. (1974). J . Cell Physiol. 83, 21 1-218. Pitha, I., Stork, E., and Wimmer, W. (1975). Exp. Cell Res. 94, 310-314. Popp, R. A,, Bailiff, E. G., Hirsch, G. P., and Conrad, R. A. (1976). Inrerdiscip. Top. Gerontol. 9, 209-2 18. Printz, D. B., and Gross, S. R. (1967). Generics 55, 451-467. Rattan, S. I. S . , Keeler, K. D., Buchanan, J . H., and Holliday, R. (1982). Biosci. Rep. 2, 561-567. Roeder, R. G . (1976). In “RNA Polymerase” (R. Losick, and M. Chamberlin, eds.), pp. 285-330. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Rohme, D. (1981). Proc. Narl. Acad. Sci. U.S.A. 78, 5009-5013. Rosenberger, R. F. (1982). IRCS Med. Sci. 10, 874-875. Rosenberger, R. F., and Foskett, G. (1981). Mol. Gen. Genet. 183, 561-563. Rosenberger, R. F., and Hilton, J. (1983). Mol. Gen. Genet. 191, 207-212. Rosenberger, R. F., Foskett, G . , and Holliday, R. (1980). Mech. Ageing Dev. 13, 247-252. Rothstein, M. (1977). Mech. Ageing Dev. 6, 241-257. Saksela, E., and Moorhead, P. S. (1963). Proc. Narl. Acad. Sci. U.S.A. 50, 390-395. Salk, D., Au, K., Hoehn, H., Stenchever, M. R., and Martin, G . M. (1981). Cyrogenet. Cell Genet. 30, 108-117. Savageau, M. A., and Freter, R . R. (1979). Proc. Natl. Acad. Sci. U.S..A 76, 4507-4510. Schneider, E. L., and Mitsui, Y. (1976). Proc. Nail. Acad. Sci. U.S.A. 73, 3584-3588. Shakespeare, V., and Buchanan, J. H. (1978). J . Cell. Physiol. 94, 105-115. Sheng, T. C. (1951). Genetics 36, 199-212. Singh, A . , Ursic, D., and Davies, J . (1979). Nature (London) 277, 146-148. Smith, I. R., Pereira-Smith, 0. M., and Schneider, E. L. (1978). Proc. Narl. Acad. Sci. U.S.A. 75, 1353-1356. Springate, C. F., and Loeb, L. A. (1975). J . Mol. Biol. 97, 577-591. Sundaram, T. K. (1967). Biochim. Biophys. Acra 138, 611-613. Szilard, L. (1959). Proc. Nail. Acad. Sci. U.S.A. 45, 30-45. Thompson, K. V. A . , and Holliday, R . (1975). Exp. Cell Res. 96, 1-6. Thompson, K. V. A . , and Holliday, R. (1983). Gerontology 29, 73-82. Watson, J. D. (1976). “Molecular Biology of the Gene,” 3rd ed. Benjamin, Menlo Park, California.
132
T. B. L. KIRKWOOD ET AL.
Waxman, L., and Goldberg, A . L. (1982). Proc. Nutl. Acud. Sci. U.S.A. 79, 4883-4887. Weiss, R. B., Murphy, J. P., and Gallant, J. A . (1983). 1.Barteriol., 158, 362-364. Wilhelm, J. P., Pettitt, S . E., and Jessopp, J . J. (1978a). Biochemistry 17, 1143-1149. Wilhelm, J. P., Jessopp, J. J., and Pettitt, S. E. (1978b). Biochemistry 17, 1149-1153. Williams, G. C. (1957). EvoLution 11, 398-41 1 . Woese, C. R. (1967). “The Genetic Code.” Harper, New York. Wojtyk, R. I . , and Goldstein, S. (1980). J . Cell. Physiol. 103, 299-304. Wojtyk, R. I., and Goldstein, S. (1982). J. Cell Bid. 97, 704-710. Wright, R. M., Hormm, M. A . , and Cummings, D. J. (1982). Cell 29, 505-515. Young, R. A., Bram, R. J., and Steitz, J. A. (1980). In “Transfer RNA” (D. Soll, J. N. Abelson, and P. R. Schimmel, eds.), pp. 99-106. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York.
INTERNATIONAL REVIEW OF CYTOLOGY. VOL. 92
Chromosome and DNA-Mediated Gene Transfer in Cultured Mammalian Cells A. J. R.
DE
JONGE'AND D. BOOTSMA
Department of Cell Biology and Genetics, Erasmus Universiv, Rotterdam, The Netherlands Introduction ................................... A. DNA-M Gene Transfer (DMGT) . . . . . . . . . . . . . . . B. Chromosome-Mediated Gene Transfer (CMGT) . . . . . . . . . 11. Technical Aspects of DMGT and CMGT . . . . . . . . . . . . . . . . . . . . . . A. DNA-Mediated Gene Transfer ................... B. Chromosome-Mediated Gene T ................... C. Choice of Recipient Cells in DMGT and CMGT . . . . . . . . . . . . 111. Uptake and Expression of Donor Genetic Materials in Recipient Cells. . . . . . . . . . . . . . . . . .......... A. Transgenome Size and Gene C B. Transgenome Stability and Linkage to Host Cell Chromosomes C. Cotransfer of Nonselected Genes . . . . . . . . . . . . . . . . . . . . . . . . . 1V. Applications of Gene Transfer in the Genetic Analysis of Mammalian Cells. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Intrachromosomal Gene Mapping. . . . . . . . . . . . . . . . . . . . . . . . . B. The Isolation of Eukaryotic Genes.. . . . . . . . . . . . . . . . . . . . . . . C. Studies of the Regulation of Eukaryotic Gene Expression ..... D. Studies of DNA Recombination and Ligation in Mammalian Cells . . . . . . . . . . . .... V. Conclusions and Future Prospects. ............................ References . . . . . . . . . 1.
I33 134 135 135 136 137 138 139 139 143 146 147 147 149 150 152 154 156
I. Introduction In analogy with the genetic transformation of bacteria using isolated prokaryotic DNA (Avery et d., 1944) Szybalska and Szybalski (1962) provided the first evidence for genetic transformation of cultured human cells using isolated human DNA. It was not until some 15 years later that DNA-mediated transformation of mammalian cells became a reproducible technique. Meanwhile, transformation using isolated metaphase chromosomes, first described by McBride and Ozer (1973), was developed and today, isolated eukaryotic DNA as well as 'Present address: Department of Microbiology and Parasitology, Free University of Amsterdam, Medical Faculty, Amsterdam, The Netherlands. 133 Copyright 0 1984 by Academc Prcss. Inc AIt rights uf reproducuon in any form reserved. ISBN 0-12-364492-5
134
A. J. R. DE JONGE AND D. BOOTSMA
well as metaphase chromosomes are very useful vectors in the transfer of genes in mammalian cell systems. In discussing these methods, a distinction can conveniently be made between the donor cell providing the genetic material transferred, the recipient cell which receives the donor genetic material, and the resulting transformant cell containing the recipient cell genome plus a variable amount of donor genetic material (usually referred to as the transgenome). The transfer and continued expression of donor genes generally occur at such a low frequency that it can only be monitored using marker genes which confer viability on transformants in selective culture conditions.
(DMGT) A. DNA-MEDIATEDGENETRANSFER Szybalska and Szybalski ( 1962) found evidence for genetic transformation in an intraspecific cell system using DNA from a human cell line and recipient cells of human origin. Although reversion of the mutation in the recipient cells could not be completely excluded in this system, gene transfer was the most plausible explanation on the basis of a quantitative comparison with appropriate controls. DMGT in cultured mammalian cells became a reproducible technique (Wigler er al., 1978) when the importance of coprecipitating the DNA with calcium phosphate, as developed by Graham and Van der Eb (1973) for the assay of viral DNA infectivity, was realized. There has since been a rapid increase in both the number of genes transferred and the applications of the DMGT methodology to study several aspects of the genetic organization of mammalian cells. The increasing interest in DMGT may be attributed at least in part to the relative simplicity of the methodology, the fact that DNA can be isolated from virtually every population of cells (prokaryotic as well as eukaryotic), and the possibilities of manipulating the DNA (for instance with recombinant DNA procedures) before it is added to cultured recipient cells. Thus DMGT has become an important tool to study the regulation of eukaryotic gene expression, e.g., by testing for the expression of in vitro manipulated genes (Mellon et al., 1981; Dierks et al., 1981). Also the transfer of characterized DNA molecules can be used to study DNA amplification (Roberts er al., 1983), DNA recombination (Small and Scangos, 1983; Robert de Saint Vincent and Wahl, 1983; Goodenow et al., 1983), and ligation (Miller and Temin, 1983) in cultured eukaryotic cells. Another application of DMGT is in the identification, isolation (by molecular cloning), and characterization of eukaryotic genes (Perucho er al., 1980; Lowy et al., 1980). A number of cellular oncogenes have been identified by using these gene transfer techniques (Cooper, 1982). Examples of these applications will be discussed in Section IV.
GENE TRANSFER IN MAMMALIAN CELLS
135
GENETRANSFER (CMGT) B . CHROMOSOME-MEDIATED When isolated metaphase chromosomes are incubated with cultured recipient cells, they can be taken up by the cells and broken down into chromosomal fragments. Occasionally, a fragment is transported into the nucleus where expression of donor genes can take place to generate a transformant cell. Since the first convincing report of CMGT (McBride and Ozer, 1973), a number of investigators have used this technique to transfer various selectable marker genes into suitable recipient cells. Usually< submicroscopic donor chromosome fragments (microtransgenomes) are retained in CMGT-generated transformants; a minority contain one (occasionally more) donor chromosome fragment(s) large enough to be cytologically detectable (macrotransgenome). Cotransfer of genes neighboring the selectable marker gene has been reported and the study of CMGTgenerated transformants has allowed the regional mapping of such syntenic genes. In the current CMGT technique, the chromosomes are added to recipient cells as a coprecipitate with calcium phosphate so that DMGT and CMGT have evolved to very similar techniques, major differences being the size of the DNA molecules and the structural composition of the donor genetic material added to the recipient cells. In previous reviews of CMGT (Willecke, 1978; McBride and Peterson, 1980; Klobutcher and Ruddle, 1981) and DMGT (Pellicer et al., 1980; Scangos and Ruddle, 1981; Gwss and Khoury, 1982), the data on these techniques have been summarized gparately. Here we review similarities and differences between them and discuss current and potential applications in studies addressing eukaryotic gene structure, functioning, and organization.
11. Technical Aspects of DMGT and CMGT For DMGT, high-molecular-weight DNA can be obtained from any population of cells, euSTyotic as well as prokaryotic, using standard isolation procedures consisting of cell lysis, protease, and RNase treatments as well as extraction and precipitation of the DNA. The average size of DNA molecules isolated from eukaryotic cells is usually 50-80 X lo3 base pairs. For CMGT, complete eukaryotic chromosomes can be obtained from populations of suitable donor cells arrested in the metaphase of the cell cycle. Rupture of the cellular membrane liberates the chromosomes which are separated from other cellular material using centrifugation techniques. Only a relatively small number of established or transformed cell lines has been used as CMGT donor cells because they meet the requirements of a high proliferative activity and the desired response to mitotic inhibitors (arrest in the metaphase).
136
A. J. R. DE JONGE AND D. BOOTSMA
A. DNA-MEDIATED GENETRANSFER The calcium phosphate coprecipitation technique as introduced by Wigler et al. (1978) is the most widely used method of eukaryotic DMGT. Mass populations (>lo6) of cells are treated with the coprecipitate and reported transfer frequencies for one gene using total donor cellular DNA range from lo-’ to less than 10 - ’. Comparisons of transfer frequencies obtained in different laboratories are difficult because of variations in technique and cell lines used and because the independent origin and true transformant nature of every clone obtained has not always been established. Using cloned selectable marker genes inserted in viral or plasmid DNA molecules, transfer frequencies up to 10W3can be obtained. Three dominant vector systems should be mentioned in this respect: vectors containing the prokaryotic dihydrofolate reductase gene (DHFR) introduced by O’Hare e/ al. (1981), the xanthine-guanine phosphoribosyl transferase gene (XGPRT; Mulligan and Berg, 1981), and the phosphotransferase gene (PTR; Colbirre-Garapin e/ al., 1981). These genes confer resistance to recipient cells for methotrexate, mycophenolic acid and the antibiotic G-418, respectively. The viral or plasmid DNA is usually coprecipitated with calcium phosphate after mixing with carrier DNA. Several adjuvants have been used to enhance the transformation frequency. Treatment with dimethyl sulfoxide (DMSO) resulted in an increase in the number of transformants using the transfer of the hamster thymidine kinase (TK) gene into mouse LTK- cells (Lewis et al., 1980). It did not significantly increase the transfection frequency in a gene transfer system using Chinese hamster (CHO) cells (Abraham et al., 1982). Recently a high efficiency of polyoma DNA transfection was observed after treatment of the recipient cells with the lysosomal enzyme inhibitor chloroquine (Luthman and Magnusson, 1983). Its applicability for the transfer of nonviral DNA sequences to mammalian cells has to be demonstrated. Using micromanipulation techniques, a solution of DNA can be directly introduced into the nucleus of individual cultured recipient cells by injection via a glass microcapillary needle (Graessmann e/ al., 1980). By i b c t i n g solutions of viral (Graessmann et al., 1977) or plasmid (Capecchi, 1980) DNA, where a relatively large fraction of the DNA molecules consists of the sequences of interest, many copies of a gene can be introduced into each nucleus. Expression of the donor DNA can then be observed in virtually every injected cell and up to 20% may grow out to transformant clones (Capecchi, 1980; Anderson et al., 1980). However, expression of genes after microinjection of total genomic DNA isolated from eukaryotic cells has not been reported. So far, microinjection has mostly been applied in studies of the organization and expression of viral genomes. With the increasing number of plasmids and viral vectors carrying selectable markers and eukaryotic genes, there are several potentially useful applications in studies of the eukaryotic cellular genome as
GENE TRANSFER IN MAMMALIAN CELLS
137
well. Microinjection may extend the applicability of DMGT to cells not amenable to the calcium phosphate technique and to genes for which no selective system is available. Since only small amounts of DNA are needed for microinjection (per cell a few femtoliters of a 1 mg/ml DNA solution are injected) the technique may be applied to assay the biological activity of DNA when only small amounts can be obtained, for example in DNA fractionation studies. In this context it should be mentioned that microinjection is also used to introduce DNA sequences into fertilized eggs of frog or mouse origin, in order to generate strains of laboratory animals carrying specifically constructed foreign DNA sequences in all their cells (Rusconi and Schaffner, 1981; Costantini and Lacy, 1981; Wagner et al., 1981). Such techniques constitute a promising new approach to the study of developmental biology. Various other methods have been developed for the introduction into mammalian cells of eukaryotic genes carried on viral or plasmid vectors. Gene transfer was accomplished after the addition, as a calcium phosphate coprecipitate, of intact bacteria carrying plasmids (Schaffner, 1980) or of intact recombinant bacteriophages (Lowy et al., 1980; Ishiura et al., 1982) to recipient cells. Efficient gene transfer was claimed after fusion of recipient cells with protoplasts of bacteria containing recombinant plasmids (Schaffner, 1980; Robert de Saint Vincent et al., 1981; Robert de Saint Vincent and Wahl, 1983). Another method of DMGT was based on infection of recipient rat cells with in vitro assembled polyoma-like particles (Slilaty and Aposhian, 1983). With the methods mentioned so far, transfer frequencies at least as high as those obtained with the standard calcium phosphate technique have been observed and a stable mode of gene expression can be achieved. Gene transfer using a DEAE dextran-DNA solution, a technique originally used in DNA-mediated viral infection of cells, has been reported to result in only a transient expression of donor genes in 0.11% of the cells during the first days after transfer (Milman and Herzberg, 1981). Stable transformants which could be isolated under selective conditions were not obtained. So far limited use has been made of the transfer techniques mentioned in this paragraph.
B . CHROMOSOME-MEDIATED GENETRANSFER In most early CMGT experiments, transfer frequencies of about l o p 7 were obtained by incubating isolated chromosomes with a suspension of recipient cells in the presence of poly-L-ornithine to enhance the association of chromosomes and cells. Comparisons with controls omitting the polycation were not made. Wullems et al. (1975) found a 10-fold increase in the number of clones arising in selective medium when the fusogen Sendai virus was added to the mixture of chromosomes and recipient interphase cells. With recipient cells in mitosis, a 30fold increase was observed. In later experiments using the same transfer system
138
A . J . R. DE JONGE A N D D. BOOTSMA
we have not been able to confirm these results so that the efficiency of these modifications is questionable. Most efficacious (CMGT frequencies of about 10- in several laboratories) has been the administration of chromosomes to recipient cell monolayers as a chromosome-calcium phosphate coprecipitate with a subsequent treatment of the recipients with DMSO, as introduced in CMGT by Miller and Ruddle (1978) and further analyzed by Lewis et al. (1980). This procedure has since become general practice so that isolated metaphase chromosomes and purified genomic DNA are now introduced into recipient cells with the same coprecipitation technique. For a number of genes, CMGT occurs at a strikingly higher frequency than DMGT in the same donor and recipient cell system. We have repeatedly transferred the gene coding for hypoxanthine phosphoribosyl transferase (HPRT) into mouse A9(HPRT - ) cells using chromosomes isolated from human HeLa cells at frequencies of > lop6.In later experiments using HeLa-DNA, transfer occurred only in a few experiments at a frequency of about 10F8. Similar results were reported by Lewis et al. (1980) in a comparative study of CMGT and DMGT of Chinese hamster genes encoding thymidine kinase or methotrexate resistance into mouse LTK- cells. It is possible that non-DNA chromosomal constituents (RNA or protein), or the manner in which the DNA is packaged in metaphase chromosomes, offer some protection against degradation in the recipient cell. Alternatively, the higher transfer frequency may be due to a larger size of the DNA molecules in isolated chromosomes. It is also possible that a fraction of purified DNA may not be active in transfer of genetic information due to damage introduced by the DNA isolation procedure. C. CHOICEOF RECIPIENTCELLSIN DMGT
AND
CMGT
High frequencies of gene transfer have been obtained only with a very limited number of recipient cell lines. This has been a serious limitation in the applicability of the technique. Most studies using total genomic DNA transfers into cells derived from the murine L929 cell line (for example LTK- , A9, B82), have been performed at reproducible transfer frequencies of l o p s to lop6. The identification and isolation of viral and cellular oncogenes, believed to be involved in cellular malignant transformation, have depended almost exclusively on the use of murine NIH3T3 cells as DMGT recipients. The proficiency for genetic transformation combined with the cell growth characteristics necessary to monitor neoplastic transformation has allowed the transfer and expression of oncogenes into NIH3T3 cells at frequencies >10W5. DMGT and CMGT into recipients of other origin have generally occurred at substantially lower frequencies. An important case in point is the gene transfer into cells of Chinese hamster ovary (CHO) origin. CHO cells are one of the most widely used experimental
GENE TRANSFER IN MAMMALIAN CELLS
139
tools in somatic cell genetic analysis and a variety of characterized mutant cell lines is available. However, CHO cells have been poor recipients in CMGT and in DMGT using total genomic DNA, usually transformed at 50- to 100-fold lower frequencies than LTK - cells. Elaborate studies attempting to improve this (Lewis et al., 1980; Linsley and Siminovitch, 1982; Abraham et al., 1982; Nairn et al., 1982) have been partially successful (Nairn et al., 1982). Important for the enhancement of transfer frequencies has been the discovery of clonal variation in the transformability of murine (Corsaro and Pearson, 1981) as well as CHO cells (Nairn et al. , 1982). Even sister subclones of LTK- can show a substantial (10to 20-fold) difference in DMGT of the same gene (Corsaro and Pearson, 1981). In view of this clonal variation, it may be worthwhile to attempt the selection of high transfer subclones of more cells lines valuable in gene transfer. Such selection could be effected using transfers of a dominant acting selectable gene such as ouabain or methotrexate resistance. Alternatively, the isolation of suitable mutant sublines from a high transfer cell line such as LTK-- may suffice. We have isolated several LTK-HPRT- double mutant cell lines and have used one of these, LTHI, in DMGT of the human HPRT gene (de Jonge et al., 1982). Another complication in gene transfer is the varied response recipient cells of different origin show toward a DMSO treatment after the addition of DNA or chromosomes. In mouse L-derived cells, the treatment enhances transformation efficiency, in human cells it is only useful at shorter treatment times with a higher concentration (Gross et al., 1979; Gross Lug0 and Baker, 1983), while in CHO cells no effect is apparent (Srinivasan and Lewis, 1980; Abraham et al., 1982). Furthermore, we have repeatedly observed 8- to 10-fold reductions in CMGT of the human gene for thymidine kinase (TK1) into Swiss mouse 3T3TK- fibroblasts when a DMSO posttreatment is given (unpublished results). It appears necessary to determine the optimal transfer conditions, including the effects of adjuvants such as DMSO, for each recipient cell line used.
111. Uptake and Expression of Donor Genetic Materials in Recipient Cells
A. TRANSGENOME SIZEAND GENECOPYNUMBER Studies addressing the state of donor genetic material in transformant cells (discussed at length in previous reviews by Willecke, 1978; McBride and Peterson, 1980; Pellicer et al., 1980; Klobutcher and Ruddle, 1981; and Scangos and Ruddle, 1981) have revealed an extensive heterogeneity with respect to size as well as organization and mode of propagation of transgenomes. The largest reported transgenome is probably an apparently complete human X-chromosome, transferred in our laboratory by CMGT into Chinese hamster-human cell hybrids as recipient cells (Wullems et al., 1976) with selection for the
140
A. J. R. DE JONGE AND D. BOOTSMA
FIG. I . Visualization of DNA-mediated gene transfer (DMGT) by in situ hybridization and Southern blotting. (A) I n sifu hybridization of 'H-labeled human repetitive DNA (Cot-I) to chromosomes of a CHO cell in a clone selected for methotrexate resistance after transfection with the bacterial dihydrofolate reductase gene (DHFR, in plasmid pHG) in the presence of human genomic
141
GENE TRANSFER IN MAMMALIAN CELLS
expression of HPRT. In all 16 transformants analyzed, the human X-linked genes for HPRT, glucose-6-phosphate dehydrogenase (G6PD), a-galactosidaseA (a-gal-A), and phosphoglycerate kinase (PGK) were expressed and karyotypic analysis revealed the presence of the human X-chromosome. In subsequent elaborate attempts using the same cell system, we have not been able to generate clones and other studies of CMGT into rodent-human cell hybrids have not been reported. We assume that the clones isolated by Wullems et af. (1976) represent an unexplained exceptional case of transfer. A minority of CMGT transformants carry a macrotransgenome detectable using alkaline Giemsa staining of metaphase spreads (Miller and Ruddle, 1978; Klobutcher and Ruddle, 1979, Klobutcher et al., 1980; Olsen et al., 1981). The transgenome carried by most CMGT transformants and all DMGT transformants studied so far is apparently too small to be detected with the alkaline Giemsa staining method. In a few transformants, obtained after DMGT of cloned selectable donor genes, microtransgenomes have been located with the more sensitive in situ hybridization technique using radiolabeled plasmid DNA as a probe (Huttner et al., 1981; Robins et al., 1981; Robert de Saint Vincent and Wahl, 1983). An example of visualization and characterization of microtransgenomes after DMGT, performed in our institute by Westerveld and Hoeijmakers, is presented in Fig. 1. Reiterated human DNA sequences (Cot- 1 DNA) were used to detect a human transgenome in CHO cells, which is not visible with the alkaline Giemsa technique. This method may have general utility, especially when specific probes are not available. A transgenome present in mouse Swiss 3T3TK- cells after CMGT using HeLa chromosomes is shown in Fig. 2 . In this case (transformant 60(5))the transgenome could not be distinguished by chromosome staining techniques. However, in situ hybridization with human Cot- 1 DNA demonstrated the presence of a large piece of human DNA independent of a mouse chromosome. By using electrophoretic and immunological procedures it was shown that the transformant contained in addition to the human TK gene ( X I ) also the human genes coding for galactokinase (GALK) and acid a glucosidase (GAA), all three genes being located on chromosome 17 in human cells. Further refinements of the in situ hybridization methodology which allow ~
~~
~
~~
~~~
DNA (cotransfer). The sequence divergence of repetitive DNA between human and Chinese hamster is sufficient to demonstrate the sites where the human sequences have been integrated. The marked spots of autoradiographic grains at the distal ends of two Chinese hamster chromosomes indicate the presence of a large amount of human DNA (arrows). (B) Southern hybridization demonstrating the presence of transfected DNA in the clone of A. DNA from this clone (lanes 1) and from the nontransfected Chinese hamster cell line (lanes 2) were digested with the restriction endonuclease EcoRI, size-fractionated on a 5% agarose gel, blotted onto nitrocellulose, and hybridized with 32Plabeled pHG (a) and human repetitive DNA (Cot-], b) followed by autoradiography. The multiple copies of the vector and the large amount of human DNA are probably the result of amplification of the transfected integrated DNA.
142
A. J . R . DE JONGE AND D. BOOTSMA
FIG.2. Visualization of a human transgenome in the genome of transformant 60(5) by in situ hybridization. Thymidine kinase (TK) positive transformants of TK- Swiss mouse 3T3 cells were selected in HAT medium after chromosome mediated gene transfer (CMGT) using isolated human HeLa chromosomes. 3H-labeled human repetitive DNA (Cot-1) was hybridized to chromosomes of the transformants. One chromosome contains sequences that hybridize with the Cot- I DNA (arrow). This chromosome is absent after selection for loss of the human TKI gene in medium with bromodeoxyuridine (not shown). This transformant has also harbored the human genes for galactokinase (CALK) and acid a-glucosidase (CAA)by cotransfer with the TKI gene (see Table I); these three genes are located on human chromosome 17. (Experiments performed in collaboration with Dr. C. Bartram in our laboratory.)
detection of single-copy DNA sequences (Malcolm et al., 1982) should eventually make detection of even the smallest transgenomes possible. In addition to these methods of visualization, transgenome sizes have also been estimated from frequencies of cotransfer of closely linked syntenic donor genes (see Section III,C) and from analytical DNA-DNA hybridization of radiolabeled purified unique sequence donor cell DNA with transformant cell DNA (Olsen et al., 1981). With appropriate probes, the latter method allows accurate estimates of the amount of donor genetic material retained in transformant cells as well as providing data on the copy number of the donor cell sequences. Such studies should provide further insight in the structure of transgenomes generated by DMGT and CMGT. Many transformants generated by DMGT of plasmid-
GENE TRANSFER IN MAMMALIAN CELLS
143
borne selectable genes have been shown to carry concatanated multiple copies of plasmid sequences. Concatanates appear to be the most frequently generated type of DMGT-derived transgenomes. In some CMGT-generated transformants, the presence of multiple copies of donor genes, but not their mutual linkage, has been demonstrated (Scangos et al., 1979). Furthermore, the accumulation of multiple copies has been proposed to account for the overexpression of donor genes and as a mode of propagation of the selectable marker gene in unstable transformants. However, various observations indicate that in many CMGT transformants, donor genes are present (or expressed) at the single-copy level (Miller and Ruddle, 1978; Klobutcher et al., 1980; Olsen et al., 1981). In conclusion, a variety of transgenomes with regard to the amount as well as the organization of the donor genetic material has been observed. Also, no fundamental differences between DMGT- or CMGT-derived transgenomes are evident but the larger size of some CMGT-derived transgenomes and the apparent preponderance of single-copy transgenomes in CMGT may indicate that donor chromosomal material has a larger average size and/or a reduced susceptibility to degradation and reconstruction in comparison to isolated donor DNA. B. TRANSGENOME STABILITY AND LINKAGE TO HOSTCELLCHROMOSOMES Transient expression in recipient cell populations has been observed to take place during the first days after DMGT of various mammalian genes. The phenomenon is reminiscent of the abortive transformation observed in transfection of viral DNA. It is presumed to reflect the presence of the genes on donor DNA sequences which lack elements necessary for their continued propagation in a relatively large fraction of the recipient cell population. Such sequences would therefore be progressively lost during continued culture, with a concomitant loss of gene expression. Ultimately, only the few cells carrying the donor genes in a propagatable fashion would continue expression and be selectable as transformant cells. Both stable and unstable modes of propagation have been observed. When cultured in nonselective medium, stable transformants retain the expression of donor genes indefinitely in all progeny cells. In unstable transformants, expression is progressively lost at a rate of about lo-* to 10- per cell generation. In some cases, loss of the donor phenotype was shown to be accompanied by the physical loss of the donor gene(s), but other modes cannot be excluded. In unstable transformants, the transgenome is believed to be propagated as an extrachromosomal genetic unit reminiscent of the bacterial episome. Direct evidence for this mode of propagation has come from the observation of free macrotransgenomes in unstable CMGT-generated transformants. Such macrotransgenomes are usually observed at 1 copy per cell and many of them have centromere-like constrictions. Klobutcher el al. ( 1980) have suggested that these free macrotransgenomes possess donor centromeres which provide for a normal
144
A. J. R . DE JONGE AND D. BOOTSMA
distribution to daughter cells at mitosis. The xenogenic nature of the donor centromere in the recipient cell would then be responsible for the instability of the transgenome, exactly analogous to the unstable retention of chromosomes of one of the parental cells in interspecific cell hybrids. Transformants carrying macrotransgenomes with centromere-like constrictions often do not express all syntenic donor genes known to be located proximal to the selected gene. An obvious explanation for this is the fragmentation of donor chromosomes and a rejoining of fragments. Linkage of a selected gene to the centromeric region of another donor chromosome could occur. Another indication of rearranged donor sequences in macrotransgenomes is the cotransfer of an asyntenic unselected donor gene, which segregates with the selectable gene and the macrotransgenome (Klobutcher and Ruddle, 1979). Although other explanations, for example the transfer of a rearranged donor chromosome cannot be excluded, it is evident that the conservation of donor chromosomal sequence organization in macrotransgenomes cannot simply be assumed. Molecular analysis using nucleic acid probes specific for certain chromosomal regions should be useful in establishing the frequency with which such rearrangements occur. The manner in which other free macrotransgenomes and microtransgenomes, which are believed to be acentric, are propagated in unstable transformants is much less clear. For such genetic elements there is no known mechanism of equal distribution over descendent cells. The favored hypothesis proposes that unequal segregation during successive rounds of replication and cell division early in the history of these transformant cell lines leads to the accumulation of multiple copies of the transgenome in some cells and no transgenomes in others. This would give the population as a whole the characteristic of instability. Evidence for the accumulation of donor genes in unstable transformants has been found in several laboratories (see for review of this work Klobutcher and Ruddle, 1981). Two lines of evidence support the appearance of cells that do not express the selectable donor gene in clones generated from unstable transformants. In selective medium the negative cells die and clones of highly unstable transformants would grow slower than clones of more stable transformants. Lewis et al. (1980) have proposed this explanation for the significant larger clone size of CMGT transformants versus DMGT transformants obtained with the same cell system. In nonselective medium, mozaic clones composed of positive and negative cells would be generated. Bacchetti and Graham (1977) have observed a number of such clones after subcloning unstable DMGT transformants in nonselective medium. Both direct and indirect evidence has been obtained for the presence of multiple copies of the donor selectable gene in some unstable transformants grown under selective conditions. However, apart from a few transformants observed to carry duplicate macrotransgenomes, it is not clear whether the multiple copies
GENE TRANSFER IN MAMMALIAN CELLS
145
are linked into one or a few concatanate transgenomes, or whether many separate copies are propagated although the observed instability is more easily understood by assuming concatanate transgenomes. Furthermore, concatanate transgenomes have been observed in many DMGT transformants while for most CMGT transformants studied, the data obtained are consistent with the retention of only one copy of the donor sequence. Hence, the mechanisms for the generation and propagation of putative acentric transgenomes are still uncertain. Also, different mechanisms may be involved for the different types of transgenomes generated in DMGT and CMGT. A better understanding of these mechanisms is necessary to increase the experimental control over the gene transfer process and further extend the utility of transfer technology. Unstable transformant cell populations have often been observed to convert to a stable mode of expression upon prolonged culture. The reverse process, conversion of stable into unstable expression, has not been described. The generally accepted interpretation of the stabilization process is that at a low frequency an unstable free transgenome becomes associated with a host cell chromosome, probably by covalent integration. Although such linkage is a rare event, the stable transformant generated would distribute the transgenome efficiently to all progeny cells and would, with time, become the predominant cell type of the population. Integration seems to occur at a random moment in the history of a transformant cell line as is indicated by the isolation of stable cell lines, in which linkage has apparently occurred at an early stage, and of unstable populations, in which it has been possible to study the stabilization process. These studies and the abundant evidence for the integration of the transgenome into host cell chromosomes in stable transformants have been reviewed before (McBride and Peterson, 1980; Klobutcher and Ruddle, 1981; Scangos and Ruddle, 1981). In summary, it has been shown that integration can take place in many different sites of the host genome. Rearrangement and loss of transgenome sequences can occur during the integration process. Rearrangement of host cell chromosomes carrying transgenomes has also been reported. Although denoted as stable, integrated transgenomes do not seem to be carried as stably as endogenous host cell genes. As suggested by Fournier et al. (1979), host cell chromosomal lability may be a prerequisite for, or a consequence of, the association with transgenomes . Mechanistic details of the integration process are unknown but recent evidence indicates that mammalian cells are capable of precise homologous recombination between donor and host cell chromosomal sequences (Goodenow et al., 1983). Further investigation will have to elucidate the general mechanisms operating in transgenome integration. Characterized nucleic acid sequences constitute a powerful tool in such studies. Fundamental differences with regard to integration of DMGT- and CMGT-derived transgenomes are not apparent at present. The preponderance of stable transformants isolated after DMGT of plasmid-borne
146
A. J . R . DE JONGE AND D. BOOTSMA
selectable genes without genomic carrier DNA (Wigler et al., 1979; Huttner et al., 1981) may be explained by the necessity of transgenome integration into host cell chromosomes in order to acquire linkage to eukaryotic origins of DNA replication in such an experimental design. In most protocols of DMGT using plasmid molecules, cellular genomic DNA is added as carrier and the selectable gene can then become linked to a carrier-DNA fragment containing an origin of replication and thus be propagated in an unstable fashion, at least for some time. Since integration is often accompanied by the loss of transgenome sequences, the stabilization event can be used to construct deletion maps of large macrotransgenomes expressing syntenic donor genes which allow a regional mapping of those genes. C. COTRANSFER OF NONSELECTEDGENES In DMGT as well as CMGT, there is ample evidence for the retention of more donor genetic material than only those sequences necessary for the propagation and expression of the selectable gene. DMGT using mixtures of plasmids results in concatanate transgenomes in which nonselective plasmid sequences have become linked to selected donor genes. This discovery has advantageously been used to permanently introduce nonselectable plasmid sequences into recipient cells by simply coprecipitating an excess of these plasmids with plasmids containing a selectable gene (Wigler et al., 1979; Wold et al., 1979). Even in DMGT using genomic donor cell DNA, cotransfer and expression of nonselected donor genes have been observed at a low frequency. Peterson and McBride (1980) reported one out of 87 LTK- DMGT transformants selected for the expression of Chinese hamster TK, which also expressed the Chinese hamster donor gene for galactokinase known to be linked to the TK locus. Similarly, in a study of DMGT into LTK- cells using human Hela-DNA, we have found expression of human GALK in 2 independent transformants out of 17 in which human TKI is expressed. As discussed by Peterson and McBride (1980), the average size of donor DNA molecules in DMGT ( lo5 nucleotide base pairs) is at least an order of magnitude smaller than the estimated intergenic TKI -CALK distance in the donor genome and at least two explanations for the cotransfer are possible. First, the intergenic distance may be less than estimated and second, the cotransfer observed may well be the result of a fortuitous linkage of donor DNA molecules in the recipient cell. In a study of 15 T K + transformants of LTK- cells, Warrick et al. (1980) did not find cotransfer of GALK or 23 other genes assayed, but one transformant expressed the asyntenic gene for esterase-D. In CMGT, cotransfer of nonselected asyntenic genes has also been observed at similar low frequencies, but cotransfer of closely linked syntenic genes occurs at relatively high frequencies. Various laboratories have reported a cotransfer frequency of about 25% for the human TK and GALK loci and the average CMGT-
GENE TRANSFER IN MAMMALIAN CELLS
147
derived transgenome was estimated to be 25-33% longer than the distance between these two loci (in different studies determined at 0.04 or 0.2% of the haploid genome). Thus the average size of CMGT-derived transgenomes may be estimated at 0.05 to 0.25% of the haploid human genome, or 2 X lo6 to 7 X lo6 nucleotide base pairs (see McBride and Peterson, 1980). Evidently further studies on the chromosome- and DNA-mediated cotransfer of closely linked genes should establish the generality of these findings, but the data obtained so far indicate a considerable difference in the average sizes of the donor genetic material in CMGT and DMGT. In CMGT transformants carrying macrotransgenomes, cotransfer of syntenic genes is often observed. Upon back-selection for loss of expression of the selectable gene, expression of the cotransferred gene(s) and the visible transgenome are usually also lost. This concomitant loss has often been interpreted as an indication of transgenome integrity. However, in addition to the evidence of rearrangements in macrotransgenomes mentioned before, a few CMGT macrotransgenomes expressing normally asyntenic donor genes have also been observed, again indicating the recombination of donor chromosomal fragments. When complications such as these are taken into account, CMGT transformants expressing genes known to be syntenic can be used to regionally map those genes.
IV. Applications of Gene Transfer in the Genetic Analysis of Mammalian Cells A. INTRACHROMOSOMALGENEMAPPING CMGT provides methods for fine structure mapping of chromosomes, particularly for the regional mapping at a level of resolution intermediate between those obtained with restriction endonuclease analysis and somatic cell hybridization (see Ruddle, 1981). The loss of donor genetic material from a macrotransgenome expressing nonselected donor genes which can occur upon integration into a host cell chromosome (stabilization) is used to construct a deletion map of that macrotransgenome (Klobutcher and Ruddle, 1979). In constructing such a deletion map, it is necessary to know on which side of the macrotransgenome the deletions have occurred. In addition to the chromosome banding technique used by Klobutcher and Ruddle (1979) to determine this, other methods such as in situ hybridization to metaphase chromosomes of nucleic acid probes specific for certain chromosomal regions, should extend the applicability of this method to chromosomal regions which lack a conspicuous banding pattern. A serious complication of the method is the occurrence of donor sequence rearrangements during gene transfer, which could lead to false localizations.
148
A. J . R . DE JONGE AND D. BOOTSMA
Intrachromosomal gene mapping by CMGT can also be achieved by determining the cotransfer frequencies of nonselected genes syntenic with the selected marker gene in primary transformants. These frequencies reflect the relative positions and distances of the genes in that linkage group (Klobutcher and Ruddle, 1981). Assuming that donor chromosome breakage and degradation occur at random during CMGT, a gene located far from the marker gene should be cotransferred at a lower frequency than a gene close to the marker gene. In view of the evidence that previously separate donor sequences can become linked in the recipient cell, this method will probably be restricted to genes for which relatively high cotransfer frequencies are observed. Furthermore, the cotransfer frequency of a gene should be calculated on the basis of multiple cotransformants expressing that gene, which could necessitate the analysis of large numbers of transformants. We have analyzed a number of CMGT transformants for the cotransfer of human acid a-glucosidase (GAA) with human thymidine kinase (TKI) and galactokinase (CALK). All 3 loci were mapped before to region 17q21-q25 (Human Gene Mapping 6, 1982). After transfer of isolated human HeLa-S3 chromosomes to Swiss mouse 3T3TK - cells, 40 independent transformants expressing human TKI were isolated, 10 of which coexpressed human GALK, as determined by electrophoresis. The 10 TK+GALK+ cotransformant cell lines and 10 TK+GALK- transformant cell lines were analyzed for the expression of GAA with an immunological procedure using a mouse anti-human GAA serum. The results are presented in Table I. These results indicate a close linkage of the GAA, the TKI, and the GALK loci on human chromosome 17. Of the 10 randomly picked TK +GALK- transformants tested, only one was GAA+ . Therefore, a reliable cotransfer frequency of GAA with TK cannot be calculated from the data we obtained. The one TK+GALK-GAA+ transformant obtained indicates a gene order placing the TK locus in between the GAA and CALK loci. However, in view of the high TK+GALK+GAA+ cotransfer frequency, a higher number of TK+GAA+ transformants would be expected for such an TABLE I COTRANSFER OF THREEGENESLOCATED ON HUMAN CHROMOSOME 17 (TK, CALK A N D GAA), AFTER CMGT USING HeLa CHROMOSOMES IN Swiss MOUSE TK- 3T3 CELLS Transformant phenotype TK GALK - GAA TK CALK GAA TK + GALK + GAA TK+ GALK- GAA + +
+
~
+
~
+
Number of clones
GENE TRANSFER IN MAMMALIAN CELLS
149
ordering. Clearly, the determination of the precise ordering of these three genes will require the analysis of a larger number of transformants. GENES B. THEISOLATIONOF EUKARYOTIC DMGT has been an essential tool in the molecular cloning of a number of selectable eukyarotic genes. A nice example is reported by Perucho et al. (1980) who used a “plasmid rescue” procedure to isolate the chicken TK gene. They ligated the bacterial plasmid pBR322 to restriction fragments of chicken DNA and isolated primary TK+ transformants of LTK-. After a second round of DMGT using DNA from primary transformants, some secondary TK+ transformants were found which carried an apparently intact pBR322 sequence linked to the selected chicken TK gene. DNA isolated from such a transformant was used to transfer this concatanate into E . coli, selecting for an antibiotic resistance marker of pBR322. Finally, the E. coli transformants obtained were tested for the ability to transfer TK back into LTK- . As stated by the authors, this method is in principle suited to the isolation of any selectable eukaryotic gene but the isolation of chicken TK has been greatly facilitated by the small size of the gene, the availability of a very strong selective system and an excellent DMGT recipient cell line. In a similar approach, Lowy et al. (1980) used the ligation of pBR322 to restriction fragments of Chinese hamster DNA to isolate the Chinese hamster APRT gene. In the secondary APRT+ transformant of LTK- APRT- cells, the antibiotic resistance marker and the prokaryotic origin of replication in pBR322 had been deleted so that the “plasmid rescue” procedure could not be followed. Instead, a recombinant phage library of chromosomal DNA from this secondary transformant in A phage Charon 4A was constructed from which one clone containing pBR322 sequences was identified by plaque hybridization. This clone was subsequently shown to contain the functional Chinese hamster APRT gene. A number of cellular oncogenes, probably involved in malignant transformation, has been identified and cloned using DMGT into mouse NIH3T3 cells. A recent example is the isolation of the oncogene of a human neuroblastoma cell line by Shimizu et al. (1983). A “suppressor-rescue’’ procedure was followed in which restriction fragments of a primary DMGT transformant were ligated to a restriction fragment of pBR322 containing a bacterial tRNA amber suppressor gene (sup F). Two successive rounds of DMGT yielded secondary transformants containing a single copy of the sup F gene flanking the oncogene. From secondary transformant DNA a A Charon 4A phage library was constructed. Since A Charon 4A carries 2 amber mutations, propagation of the library in suppressorfree bacterial hosts could be used to select the phages that expressed the sup F gene. DNA from 4 such phages was assayed for oncogenic activity on NIH 3T3 cells and DNA from 2 of these was found to be active, suggesting that these
150
A. J . R . DE JONGE AND D. BOOTSMA
phages contain a biologically active clone of the oncogenic sequences of the neuroblastoma cell line. These studies elegantly show the combination of bacterial genetics and recombinant DNA technologies with mammalian cell genetics to be a powerful tool in the genetic analysis of higher eukaryotes.
C. STUDIESOF THE REGULATION OF EUKARYOTIC GENEEXPRESSION 1. Analysis of DNA Sequences Required for the Transcription of Eukaryotic Genes In vitro and in vivo transcription studies of cloned eukaryotic (viral and mammalian) genes have shown sequences located in front of the actual protein coding region to play an important role in the efficient and precise initiation of mRNA synthesis. Various workers have performed such studies in vivo by DMGT of cloned genes with characterized changes in the 5’ extragenic region to investigate the effect of these changes on the transcription of the cloned gene. Both transient expression, where the gene is carried extrachromosomally at a high copy number (Mellon et al., 1981; Banerji et al., 1981) and stable expression, where the gene is integrated at a low copy number (Dierks et al., 198I), have been investigated. Expression has usually been investigated by S1 mapping analysis of the RNA products, which allows the determination of the 5‘ end of the transcripts with an accuracy of 1 to 2 nucleotides (Weaver and Weissmann, 1979). Grosveld et al. ( 1982a,b) have investigated the transient expression of various characterized mutant rabbit (3-globin templates linked to an SV40-pBR322 recombinant vector after DMGT into human HeLa cells. The rabbit P-globin RNA molecules produced during the first 3 days after transfer were analyzed by S1 mapping. A comparison of the level of transcription obtained from the various deletion mutants used showed the requirement for efficient transcription in vivo of both a region between 100 and 58 nucleotide base pair (bp) in front of the mRNA start site and the ATA box, an AT-rich sequence located at 25-30 bp in front of the start site. In addition, it was shown that deletion of the ATA box induces the production of (3-globin mRNA molecules with heterogeneous 5’ ends. Also, deletion of sequences downstream of the ATA box shifts the site of transcription initiation in vivo to a downstream position, the size of the shift being equal to the size of the deletion. The use of stable expression after DMGT is illustrated by a similar study by McKnight et al. (1981) of the sequence requirements for the in vivo expression of the herpes simplex virus (HSV) thymidine kinase gene by DMGT of characterized mutant templates into mouse LTK- cells. 2. Studies of the Mechanisms of X-Chromosome Inactivation
DMGT has been applied in the study of the molecular basis of X-chromosome inactivation (Lyon, 1961; Gartler and Andina, 1976) in eutherian mammals. In a
GENE TRANSFER IN MAMMALIAN CELLS
151
recent study, we have compared the transformation efficiency of the gene for HPRT carried on the active and inactive human X-chromosome (de Jonge et al., 1982). The HPRT locus is susceptible to inactivation and humans heterozygous for HPRT deficiency (the Lesch-Nyhan syndrome, L-N) show a mozaic pattern for HPRT expression in their cultured fibroblasts. After SV40 transformation of fibroblasts from an L-N heterozygote, we obtained clonally derived cell lines carrying the HPRT+ allele either on the active or on the inactive chromosome. DNA isolated from HPRT+ and HPRT- cells was used in DMGT of the active and inactivated HPRT gene, respectively, into HPRT-deficient mouse LTH 1 cells isolated in our laboratory. Transformants expressing human HPRT were obtained with both DNAs. The transformation frequency observed with the active and the inactivated HPRT gene was and 5 x 1 O p 7 , respectively. These experiments demonstrate that the HPRT gene on the inactivated human Xchromosome can be expressed efficiently after DMGT. In similar experiments, Venolia and Gartler (1983) have found that the inactivated HPRT was transformed at least 25 times less efficient than the active HPRT gene (2 X lo-* and 5X respectively). Although there are differences in procedures and cell lines used, it is not obvious how these could account for the marked difference in transformation efficiency of the inactivated HPRT gene observed in these two studies. In other experiments using DNA isolated from mouse-human hybrid cells with an active or an inactivated human X-chromosome, the inactivated X-DNA was ineffective while transformation with the active HPRT gene was obtained at frequencies ranging from 2.5 X l o p 7 to 25 X lo-’ (Liskay and Evans, 1980; Lester et al., 1982; Venolia et al., 1982). In one experiment using inactivated XDNA, Lester et al. (1982) did obtain 10 clones (frequency which aborted before they could be analyzed for expression of HPRT. In these colonies a transient expression of reactivated HPRT could have occurred. We have also obtained similarly abortive colonies using active as well as inactivated X-DNA (unpublished results). The true nature of such abortive colonies has not been established. However, they have been observed before in experiments aimed at transfer of HPRT (McBride and Ozer, 1973; Graf et al., 1979; Lester et al., 1980). Although a transient expression of the HPRT gene could account for the generation of abortive colonies, other explanations unrelated to HPRT expression must be considered equally likely. €hapman et al. (1982) reported that 1 out of 59 transformants isolated after DMGT of HPRT using DNA isolated from various tissues of adult mice expressed the inactivated allelic HPRT variant these mice carry. The lower efficiency in DMGT of the inactivated HPRT has been interpreted to indicate a modification of the DNA in inactive X-chromosomes, such as methylation of cytosine residues. Evidence that methylation may be involved in maintenance of the inactivated state has come from studies using 5-azacytidine, a cytosine ana-
152
A. J . R . DE JONGE AND D. BOOTSMA
log which cannot be methylated at the 5 position and therefore leads to demethylation of the DNA in which it is incorporated. From cultures of mousehuman cell hybrids with an inactive human X-chromosome only, clones expressing human HPRT can be isolated after treatment with 5-azacytidine at frequencies up to lo3 times the spontaneous frequency (Mohandas e t a / ., 1981; Lester et al., 1982; Marshall-Graves, 1982; Jones et a / . , 1982). In some clones, expression of another X-linked gene (G6PD or PGK) was also observed. DNA isolated from hybrid cells with 5-azacytidine reactivated HPRT functions in DMGT of that gene (Venolia et al., 1982; Lester et al., 1982). This was interpreted to indicate that the 5-azacytidine treatment has caused a structural change, possibly hypomethylation, in the DNA at or near the formerly inactivated HPRT gene which results in expression of the gene. A comprehensive discussion of X-chromosome inactivation and reactivation is beyond the scope of this review of gene transfer. Differences in the cell systems used in the studies mentioned may have contributed, at least in part, to the observed differences in the capacity of active, inactivated, and reactivated HPRT to function in DMGT. Studies of X-chromosome methylation including unsuccessful attempts at 5-azacytidine reactivation in primary human fibroblasts have indicated that if there is a relation between methylation and X-chromosome inactivation, this relation could be complex (Wolf and Migeon, 1982).
D. STUDIESOF DNA RECOMBINATION AND LIGATION I N MAMMALIAN CELLS The introduction of specifically tailored DNA molecules of defined structure into eukaryotic cells and restriction enzyme analysis of those sequences in transformants provides a powerful tool to investigate the enzymatic host cell processes involved in the joining of DNA sequences. These studies not only yield information on the capabilities and specificities of such processes in mammalian cells, but they also contribute to our understanding of the molecular events in which donor DNA molecules can become involved during gene transfer. Simultaneous introduction of DNA molecules carrying overlapping fragments of viral genomes have shown that recombination can occur in homologous regions (Upcroft e t a / . , 1980; Wake and Wilson, 1979; Young and Fisher, 1980) as well as in regions of little (Anderson et al., 1982; Stringer, 1982) or no homology (Folger et al., 1982). Further evidence for homologous recombination of donor DNA sequences in recipient cells has recently been obtained by DMGT of mixtures of two characterized plasmids with nonoverlapping deletions of the cloned HSV-TK gene into LTK- cells (Small and Scangos, 1983). Similar experiments were performed using the cloned Syrian hamster CAD gene which confers uridine prototrophy into CHO cells (Robert de Saint Vincent and Wahl, 1983). By selecting for the
GENE TRANSFER IN MAMMALIAN CELLS
153
expression of a functional resistance gene, transformants were isolated in both studies and restriction enzyme analysis of the plasmid sequences in those transformants showed that homologous, in a number of instances reciprocal, exchanges between the two donor DNA plasmid molecules had occurred to generate the functional gene. Goodenow et al. (1983), have used a cotransformation system with HSV-TK to introduce several truncated forms of the BALB/c H-2Ld gene of the murine major histocompatibility complex into mouse LTK- cells (H-2k haplotype). They have obtained TK cotransformants which contain complete hybrid transplantation antigen molecules of the donor haplotype. These cotransformants are generated at substantially lower frequencies than observed in cotransformation of a complete copy of the H-2Ld gene. The authors propose as the most plausible explanation a homologous recombination between host cell chromosomal and donor DNA sequences to generate hybrid templates from which the hybrid molecules observed are generated. Although integration of donor DNA sequences into host cell chromosomes is a frequently observed phenomenon in DMGT (and probably occurs often in CMGT as well), it has not before been possible to demonstrate site-specificity or a homology requirement for this process. With more cloned eukaryotic genes becoming available, it should be possible to determine if the observations of Goodenow et al. (1983) indicate a generally operative mechanism, or whether such events occur only in polymorphic multigene families such as those of the histocompatibility complex or the immunoglobulins for which special genetic mechanisms facilitating recombination exist. Very recently, Miller and Temin (1983) have used DMGT cotransformation of pairs of plasmids containing nonoverlapping complementary fragments of the spleen necrosis virus (SNV), an avian retrovirus, to study the ligation of physically unlinked DNA molecules in cultured chicken embryo fibroblasts and canine thymus cells. Transformation of a single linear plasmid or mixtures of circular plasmids was ineffective, but infectious virus was produced after introduction of various mixture of different linearized plasmids, with or without S1nuclease treatment. The plasmid mixtures were composed in such a way that ligation in both coding and noncoding sequences could be monitored and there was no essential difference in the efficiency of virus production. The results indicate that eukaryotic cells can perform both blunt-end and cohesive-end ligation at high efficiency, often without loss of DNA bases. The authors went on to show that the limiting step of the cotransformation process is apparently not the intracellular linkage, but rather the ability of a single cell to take up at least one copy of each of the two physically unlinked DNA molecules in a biologically active state. The recent experiments discussed in this section clearly indicate that host cell processes with a variety of specificities can act on donor DNA sequences. Stud+
154
A. J. R . DE JONGE AND D. BOOTSMA
ies such as these should provide further insight in the molecular mechanisms of gene transfer. Also, the role of these host cell processes in the metabolism of endogenous genetic material will be a very interesting field of future research.
V. Conclusions and Future Prospects Incubation of cultured mammalian cells with isolated metaphase chromosomes or eukaryotic DNA results under appropriate conditions in the transfer of functional eukaryotic genes. Transient expression of donor genes, measurable during the first days after transfer, can occur in a relatively large fraction of the recipient cell population. Cells with a more permanent mode of expression arise at such a low frequency that sensitive systems for their selection have to be employed. Thus, transfer has in general been restricted to selectable marker genes. Most details of the events in gene transfer are still unclear, but it is evident that recipient cells can process donor genetic material in various ways. Besides degradation, ligation as well as homologous and nonhomologous recombination of donor DNA sequence mutually and with host cell chromosomal sequences have been observed. In CMGT, these processes can often involve long donor sequences since cotransfer studies using genes of known proximity indicate that, on average, a sequence of about 5 X lo6 nucleotide base pairs remains intact during transfer. However, intact sequences at least an order of magnitude longer (macrotransgenomes) have also been observed in some transformants and shorter sequences may be retained intact at a high frequency. The maximal size of donor sequences retained intact after DMGT is in the order of lo5 nucleotide base pairs, a high estimate of the average size of isolated donor DNA molecules. The initial size and the structural composition of donor genetic material may influence the ultimate size of intact donor sequences as it is conceivable that nonDNA chromosomal constituents or helix folding offer some protection against breakage and degradation of donor chromosomal DNA. In both CMGT and DMGT, donor sequences can be ligated and recombined into compound transgenomes. Linkage appears to occur at random, creating novel gene sequences. These can be very useful (for example in generating stable cotransformants of a nonselectable gene) but they can also complicate linkage studies of syntenic genes by creating spurious synteny relationships. A transformant may contain multiple identical or different transgenomes. A transgenome can be propagated independently (unstable transformant) or covalently linked to host cell chromosomes (stable transformant). In most unstable transformants, the mechanisms of propagation of the transgenome are obscure and a transgenome loss-rate of lo-' to lo-* per cell generation is observed. Stabilization appears to occur at a random moment in the history of the transformant by integration into a host cell chromosome. The integration process
GENE TRANSFER IN MAMMALIAN CELLS
155
did not seem to have any site-specificity. However, recent evidence indicates that very precise homologous recombination between exogenous and host cell chromosomal DNA sequences is possible but the generality of this phenomenon has yet to be established. It is plausible that a site specificity of transgenome integration does exist but is usually masked by the complexity of the eukaryotic genome, where multiple sequences of sufficient homology with transgenome sequences may exist. As has been proposed before (Klobutcher and Ruddle, 1981), the multiplicity of transgenome integration sites could be advantageously used in the development of panels of stable transformant cell lines with a selectable gene integrated into previously unselectable linkage groups. Such panels would be useful for further regional mapping of chromosomes, using CMGT and would also facilitate assignment of genes to chromosomes in cell hybridization studies by providing a selectable marker on chromosomes which previously did not have such a marker. Isolation of a number of mammalian genes has been accomplished via the cloning of a complementary DNA (cDNA) after immunoprecipitation of polyribosomes, or other methods, and subsequent use of this cDNA in screenings of genomic clone banks. The alternative approach using DMGT can be used for genes which are not represented in relative abundance in cellular mRNA and should allow the isolation of selectable or otherwise identifiable genes which can be stably transferred. In these experiments cotransfer with vectors containing dominant selectable marker genes may facilitate the selection of high transfer clones. These transfer techniques not only provide a tool to isolate eukaryotic genes, but they also allow these genes-manipulated at will by the investigator-to be introduced into cells of various genetic make-up in order to investigate the regulation of gene expression. A wide range of applications of the gene transfer methodology is evident but the generality of many phenomena observed will have to be established and the rather narrow base of selectable genes and suitable recipient cell lines will have to be extended if this methodology is to be used to its full potential. Substantial variation has been observed in both transformability, even for closely related cell lines, as well as in the effect of adjuvants such as DMSO. Therefore, the conditions for optimal transfer will have to be determined separately for every new cell system introduced. The use of cloned dominant-acting resistance markers should be helpful in this respect. Although many details of the gene transfer process remain to be elucidated, this technique has, in combination with recombinant DNA and other molecular biological techniques, evolved in less than a decade from the first basic methodological studies to exciting applications in the analysis of the genetic organization of mammalian cells. Further understanding of the cellular processes involved and refinement of the gene transfer techniques will extend the utility and experimental control of the methodology.
156
A. J. R. DE JONGE AND D. BOOTSMA RWERENCES
Abraham, I., Tyagi, J. S., and Gottesman, M. M. (1982). Somar. Cell Genet. 8, 23-39. Anderson, R. A,, Krakauer, T., and Camerini-Otero, R. D. (1982). Proc. Narl. Acud. Sci. U.S.A. 79, 2748-2752. Anderson, W. F., Killos, L., Sanders-Haigh, L., Kretschmer, P. J . , and Diacuniakos, E. G. (1980). Proc. Narl. Acud. Sci. U.S.A. 77, 5399-5403. Avery, 0. T., MacLeod, C. M., and McCarty, M. (1944). J . Exp. Med. 79, 137-158. Bacchetti, S., and Graham, F. L. (1977). Proc. Nurl. Acad. Sci. U.S.A. 74, 1590-1594. Banerji, J., Rusconi, S., and Schaffner, W. (1981). Cell 27, 299-308. Capecchi, M. R . (1980). Cell 22, 479-488. Chapman, V. M., Kratzer, P. G., Siracusa, L. D., Quarantillo, B. A., Evans, R., and Liskay, R. M. (1982). Proc. Nut/. Acad. Sci. U.S.A. 79, 5357-5361. Colbkre-Garapin, F., Horodniceanu, F., Kourilsky, P., and Garapin, A-C. (1981 ). J . Mol. Eiol. 150, 1-14. Cooper, G . M. (1982). Science 217, 801-806. Corsaro, C. M., and Pearson, M. L. (1981). Somat. Cell. Genet. 7, 617-630. Costantini, F., and Lacy, E. (1981). Nature (London) 294, 92-94. Dierks, P., van Ooyen, A , , Mantei, N., and Weissmann. C. (1981). Proc. Natl. Acad. Sci. U.S.A. 78, 1411-1415. Folger, K. R., Wong, E. A,, Wahl, G., and Capecchi, M. R. (1982). Mol. Cell. Eiol. 2, 13721387. Fournier, R. E. K., Juricek, D. K., and Ruddle, F. H. (1979). Somat. Cell Genet. 5, 1061-1077. Gartler, S. M., and Andina, R. J. (1976). Adv. Hum. Genet. 7, 99-140. Goodenow, R. S., Stroynowski, I . , McMillan, M., Nicolson, M., Eakle, K., Sher, B. T . , Davidson, N., and Hood, L. (1983). Nature (London) 301, 388-394. Graessmann, A., Graessmann, M., and Mueller, C. (1977). Proc. Natl. Acud. Sci. U . S . A . 74, 483 1-4834. Graessmann, A,, Graessmann, M., and Mueller, C. (1980). In “Methods in Enzymology” (L. Grossman and K. Moldave, eds.), Vol. 65, pp. 816-825. Academic Press, New York. Graf, L. H., Urlaub, G., and Chasin, L. A. (1979). Somar. Cell Genet. 5, 1031-1044. Graham, F. L., and van der Eb, A. J. (1973). Virology 52, 456-467. Gross, T. A., Squires, S., Martin, P., and Baker, R. M. (1979). J . Cell Eiol. 83, 453a. Gross Lugo, T . , and Baker, R. M. (1983). Somar. Cell Genet. 9, 175-188. Grosveld, G . C., de Boer, E., Schewmaker, C. K., and Flavell, R . A. (1982a). Nature (London) 295, 120-126. Grosveld, G. C., Rosenthal, A., and Flavell, R. A. (1982b). Nucleic Acids Res. 10, 4951-4971. Gruss, P., and Khoury, G. (1982). Curr. Top. Microsc. 96, 159-170. Human Gene Mapping VI (1982). Cytogenet. Cet’t. Genet. 32 ( I -4). Huttner, K. M., Barbosa, J. A., Scangos, G. A., Pratcheva, D. D., and Ruddle, F. H. (1981). 1. Cell Eiol. 91, 153-156. Ishiura, M., Hirose, S . , Uchida, T., Hamada, Y . , Suzuki, Y., and Okada, Y . (1982). Mol. Cell Biol. 2, 607-616. Jones, P. A , , Taylor, S . M., Mohandas, T., and Shapiro, L. J . (1982). Proc. Natl. Acad. Sci. U.S.A. 79, 1215-1219. de Jonge, A. J. R., Abrahams, P. J., Westerveld, A , , and Bootsma, D. (1982). Nature (London) 295, 624-626. Klobutcher, L. A,, and Ruddle, F. H. (1979). Nature (London) 280, 657-660. Klobutcher, L. A , , and Ruddle, F. H . (1981). Annu. Rev. Eiochem. 50, 533-554.
GENE TRANSFER IN MAMMALIAN CELLS
157
Klobutcher, L. A , , Miller, C. L., and Ruddle, F. H. (1980). Proc. Null. Acud. Sci. U.S.A. 77, 3610-3614. Lester, S . C., LeVan, S. K . , Steglich, C., and DeMars, R. (1980). Somat. Cell Genet. 6, 241-259. Lester, S. C., Kom, N. J., and DeMars, R . (1982). Somat. Cell Genet. 8, 265-284. Lewis, W. H., Srinivasan, P. R., Stokoe, N., and Siminovitch, L. (1980). Somat. Cell Genet. 6, 333-347. Linsley, P. S . , and Siminovitch, L. (1982). Mol. Cell. Biol. 2, 593-597. Liskay, R. M., and Evans, R. J. (1980). Proc. Natl. Acad. Sci. U.S.A. 77, 4895-4898. Lowy, I . , Pellicer, A , , Jackson, J. F., Sim, G. K., Silverstein, S., and Axel, R. (1980). Celt 22, 817-823. Luthman, H., and Magnusson, G . (1983). Nucleic Acids Res. 11, 1295-1308. Lyon, M. F. (1961). Nature (London) 190, 372-373. Malcolm, S., Barton, P., Murphy, C., Ferguson-Smith, M. A,, Bentley, D. L., and Rabbitts, T. H. (1982). Proc. Nutl. Acud. Sci. U.S.A. 79, 4957-4961. Marshall-Graves, J. A. (1982). Exp. Cell Res. 141, 99-105. McBride, 0. W., and Ozer, H. L. (1973). Proc. Nutl. Acad. Sci. U.S.A. 70, 1258-1262. McBride, 0. W., and Peterson, J. L. (1980). Annu. Rev. Genet. 14, 321-345. McKnight, S . L., Gavis, E. R., and Kingsbury, R. (1981). Cell 25, 385-398. Mellon, P., Parker, V., Gluzman, Y., and Maniatis, T. (1981). Cell 27, 279-288. Miller, C. K., and Temin, H. M. (1983). Science 220, 606-609. Miller, C. L., and Ruddle, F. H. (1978). Proc. Nail. Acad. Sci. U.S.A. 75, 3346-3350. Milman, G., and Herzberg, M. (1981). Somat. Cell Genet. 7, 161-170. Mobandas, T., Sparkes, R. S . , and Shapiro, L. J. (1981). Science 211, 393-396. Mulligan. R. C., and Berg, P. (1981). Proc. Nutl. Acud. Sci. U.S.A. 78, 2072-2076. Nairn, R. S., Adair, G. M., and Humphrey, R. M. (1982). Mol. Gen. Genet. 187, 384-390. O’Hare, K., Benoist, C . , andBreathnach, R. (1981). Proc. Nurl. Acud. Sci. U.S.A. 78, 1527-1531. Olsen, A. S., McBride, 0. W., and Moore, D. E. (1981). Mol. Cell. Eiol. 1, 439-448. Pellicer, A,, Robins, D., Wold, B., Sweet, R., Jackson, J . , Lowy, I., Roberts, J. M., Sim, G. K., Silverstein, S . , and Axel, R. (1980). Science 209, 1414-1422. Perucho, M., Hanahan, O., Lipsich, L . , and Wigler, M. (1980). Nature (London) 285, 207-210. Peterson, J . L., and McBride, 0. W. (1980). Proc. Nutl. Acud. Sci. U.S.A. 77, 1583-1587. Robert de Saint Vincent, B., and Wahl, G. M. (1983). Proc. Natl. Acad. Sci. U.S.A. 80, 20022006. Robert de Saint Vincent, B., Delbriick, S . , Eckhart, W . , Meinkoth, J . , Vitto, L., and Wahl, G . (1981). Cell 27, 267-277. Roberts, J. M., Buck, L. B., and Axel, R. (1983). Cell 33, 53-63. Robins, D. M., Ripley, S., Henderson, A. S . , and Axel, R. (1981). Cell 23, 29-39. Ruddle, F. H. (1981). Nature (London) 294, 115-120. Rusconi, S., and Schaffner, W. (1981). Proc. Nutl. Acud. Sci. U.S.A. 78, 5051-5055. Scangos, G. A . , and Ruddle, F. H. (1981). Gene 14, 1-10. Scangos, G. A . , Huttner, K . M., Silverstein, S . , and Ruddle, F. H. (1979). Proc. Natl. Acud. Sci. U.S.A. 76, 3987-3990. Schaffner, W. (1980). Proc. Nad. Acad. Sci. U.S.A. 77, 2163-2167. Shimizu, K . , Goldfarb, M . , Perucho, M., and Wigler, M. (1983). Proc. Nutl. Acad. Sci. U.S.A. 80, 383-387. Slilaty, S. N., and Aposhian, H. V. (1983). Science 220, 725-727. Small, J . , and Scangos, G . (1983). Science219, 174-176. Srinivasan, P. R., and Lewis, W. H. (1980). In “introduction of Macromolecules into Viable Mammalian Cells” (R. Baserga, C. Croce, and G. Rovera, eds.), pp. 27-45. Liss, New York.
158
A. J. R. DE JONGE AND D. BOOTSMA
Stringer, J. R. (1982). Nature (London) 296, 363-366. Szybalska, E. H . , and Szybalski, W. (1962). Proc. Narl. Acad. Sci. U.S.A. 48, 2026-2034. Upcroft, P., Carter, B., and Kidson, C. (1980). Nucleic Acids Res. 8, 2725-2736. Venolia, L . , and Gartler, S. M. (1983). Nature (London) 302, 82-83. Venolia, L., Gartler, S. M . , Wassman, E. R., Yen, P., Mohandas, T., and Shapiro, L. J. (1982). Proc. Natl. Acad. Sci. U.S.A. 79, 2352-2354. Wagner, T. E., Hoppe, P. C., Jollick, J. D., Scholl, D. R., Hodinka, R . L., and Gault, J. B. (1981). Proc. Narl. Acad. Sci. U.S.A. 78, 6376-6380. Wake, C. T., and Wilson, J. H. (1979). Proc. Natl. Acad. Sci. U.S.A. 76, 2876-2880. Warrick, H., Hsiung, N., Shows, T. B., and Kucherlapati, R. (1980). 1. Cell Biol. 86, 341-346. Weaver, R. F., and Weissmann, C. (1979). Nucleic Acids Res. 6, 1175-1 193. Wigler, M . , Pellicer, A , , Silverstein, S., and Axel, R. (1978). Cell 14, 725-731. Wigler, M . , Sweet, R., Sim, G . K., Wold, B., Pellicer, A., Lacy, E . , Maniatis, T., Silverstein, S . . and Axel, R. (1979). Cell 16, 777-785. Willecke, K. (1978). Theor. Appl. Genet. 52, 97-104. Wold, B . , Wigler, M., Lacy, E., Maniatis, T., Silverstein, S . , and Axel, R . (1979). Proc. Natl. Acad. Sci. U.S.A. 16, 5684-5688. Wolf, S. F., and Migeon, B. R. (1982). Nature (London) 295, 667-671 Wullems, G. J . , van der Horst, J . , and Bootsma, D. (1975). Somat. Cell Genet. 1, 137-152. Wullems, G. J., van der Horst, J., and Bootsma, D. (1976). Somat. Cell Genet. 2, 359-371. Young, C. S. H., and Fisher, P. 8. (1980). Virology 100, 179-184.
INTERNATIONAL REVIEW OF CYTOLOGY, VOL. 92
DNA Methylation in Eukaryotic Cells AHARONRAZINAND HOWARDCEDAR Department of Cellular Biochemistry, The Hebrew University-Hadassah Medical School, Jerusalem, Israel I. Introduction . . . . . . . . . . .
..............................
11. The Methylation Pattern o aryotic DNA.. . . . . . . . . . . . . . . . . . . A. Clonal Inheritance of the Methylation Pattern . . . . . . . . . . .
B. Methylases.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Methylation at the Replication Fork ....................... 111. DNA Methylation and Gene Expression ....... ................ A. Correlation between Methylation and Gene Activity. . . . . . . . . . B. Cause and Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Mechanisms of Blocking Gene Activity. . . . . . . . . . . . . . . . . . . . D. Differentiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Demethylation Models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Main Concepts.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
159 160 163 163 165 166 I66 170 172 176 178 179 180 181
I. Introduction For the last two decades the phenomenon of DNA modification has been extensively explored. For many years the lack of appropriate techniques allowed only the detection and quantitation of methylated bases in DNA. Although a substantial body of information accumulated with regard to the overall content of methyl groups and their distribution along the chromosomes, its contribution to the understanding of the DNA methylation phenomenon has been very limited. The use of restriction enzymes and recombinant DNA technology has led to significant progress in the understanding of both the metabolism and function of DNA methylation. During the last couple of years several hundred papers have been published in the field of DNA methylation and many reviews summarize this vast literature (Taylor, 1979; Razin and Riggs, 1980; Drahovsky and Boehm, 1980; Burdon and Adams, 1980; Razin and Friedman, 1981; Ehrlich and Wang, 1981; Doerfler, 1981, 1983; Felsenfeld and McGhee, 1982; Riggs and Jones, 1983). This current review will focus primarily on DNA methylation in the eukaryotic cell. The establishment of a methylation pattern, its clonal inheritance, and possible biological functions will be dealt with in detail. 159 Copyright 0 1984 by Academic Press, Inc. All rights of rcproductton in any form reserved ISBN 0-12-364492-5
160
AHARON RAZIN AND HOWARD CEDAR
11. The Methylation Pattern of Eukaryotic DNA
The DNA of most eukaryotes is methylated at the 5 position of cytosine residues (5-methylcytosine, 5mC) in 5 ’CpG3’ containing sequences. The fact that the CpG dinucleotide has an axis of symmetry makes these eukaryotic methylated sites similar to those in prokaryotes, where the cytosine or adenine residues are always methylated on both DNA strands (Smith, 1979; Roberts, 1976; Razin et al., 1980). In a series of clever experiments using methylsensitive restriction enzymes, Bird (1978) showed that CpG sites in eukaryotic DNA are also symmetrically methylated and other studies have confirmed these results (Cedar et al., 1979). It should be noted that some workers have claimed TABLE I CpG METHYLATION TISSUEAND SPECIESS P E c i i x r r Y a Species Human Calf Rat Adult Fetal Mouse Adult somatic tissues
Germ line and embryonic cells
Cell lines
Sea urchin Drosophila
Tissue
CpG methylated (%)
T cells Placenta Thymus
82 47 85
Liver Spleen Liver
82 85 76
Liver Kidney Brain Lung Spleen Testes Sperm Placenta Yolk sac Embryo (9- 10 days) L cells 3T3 cells Teratocarcinoma cells F9 T cell line Ascites Sperm All {issues All developmental stages
78 82 75 76 82 69 67
49 53 77 65 70 65
70 67 39 <0.01
“DNA preparation and analysis of methylation at CpC sequences were according to Gruenbaum cr al. (1981b). All values in the table are average estimates obtained with at least 3 independent DNA preparations. The variability was S2%.
161
DNA METHYLATION IN EUKARYOTIC CELLS TABLE 11 METHYLATION PATTERNSOF “HOUSEKEEPINC” GENES Species
Gene
Hamster
aprt
Mouse
dhfr
rRNA genes Rat
rRNA genes
Amplified rRNA genes
Xenopus
rRNA genes
Somatic 5 S RNA genes
Pattern of methylation
5’ region hypomethylated in all tissues 5’ region hypomethylated in all tissues Hypomethylated in all tissues Hypomethylated in all tissues Hypermethylated in rat hepatoma cell line 60 bp repeated spacer hypomethylated in somatic tissues methylated in sperm Hypomethylated in liver
References Stein et al. (1983) Stein e t a / . (1983) Bird er al. (1981b) Reilly et al. (1982); Kunnath and Locker ( 1982) Tantravahi et al. (1981) Bird et a/. (1981a)
Sims el a/. (1983)
that 5mC can be found in other dinucleotide sequences. Several investigators have shown that certain CCGG sites in eukaryotic genes are not cleavable by the restriction enzyme MspI, suggesting the presence of CpC methylation, but it has recently been demonstrated that this is an artifact and at least one such site is indeed not methylated at this external cytosine residue (Keshet and Cedar, 1983; Busslinger et al., 1983). 5mC is indeed located in all cytosine-containing dinucleotides in higher plant DNA, but these methyl moieties are all found in the symmetrical prototype sequence CXG, where X can be any of the four major bases (Gruenbaum et al., 1981a). In this case also the cytosine residues on both DNA strands are methylated, although some hemimethylated sites have also been observed in plant satellite DNA (Deumling, 1981). The extent of methylation of CpG sequences is species specific (Table I). In addition, the representation of CpG sequences in the various species is inversely correlated with the extent of CpG methylation. It was, therefore, suggested that the under representation of CpG results from high mutability of methylated CpG (Bird and Taggart, 1980). However, these mutations must be rare, as otherwise the methylated CpGs would have been quickly eliminated. Instead, changes in the CpG content among the species is observed only on an evolutionary scale. The fact that CpG sequences are only partially methylated suggests that the pattern of methylation in any eukaryotic cell is not solely determined by the distribution of CpG sequences. In fact the analysis of methylation patterns in specific genes reveals a nonrandom distribution of methylated CpG sequences: In constitutive genes
162
AHARON RAZIN AND HOWARD CEDAR TABLE 111 METHYLATION PATTERNS OF Tissuti-S~~ciric GENES
Species Human
Rabbit
Rat
Gene P-Globin gene family Growth hormone and chorionic somatomammotmpin al and a2 type 1 collagen genes Cytochrome P-450 P-Globin genes
Liver protein 22 gene Insulin I and I1
Albumin gene
a-Fetoprotein gene
Cytochrome P-450 Mouse
Metallothionein I gene
J chain gene
Chicken
Cytochrome P-450 P-Globin genes a-Globin genes
Ovalbumin, ovotransferin, and ovomucoid
Pattern of methylation
References
Hypomethylated in erythroid cells Hypomethylated in pituitary and placenta, respectively
van der Ploeg and Flavell (1980) Hjelle et a/. (1982)
Transformed nonexpressing lung fibroblasts are hypermethylated
Parker er al. (1982)
Chen er a / . (1982) Hypornethylated in erythroid cells Hypermethylated in nonexpressing hepatoma cells Hypomethylated in expressing cells (insulinomas 1 and 4, line 38) Hypomethylated in expressing hepatoma cells Hypomethylated in adult liver: methylated in fetal liver Hypornethylated in expressing hepatoma cells Hypornethylated in adult liver; methylated in fetal liver Hypornethylated in adult “Ah responsive’’ C57BL/6N mouse liver Hypomethylated in fetal liver, yolk sac, endoderm, and afetoprotein-producing hepatoma Hypomethylated in expressing B cells
Waalwijk and Flavell (1978b); Shen and Maniatis (1980) Nahkasi et al. (1981) Cate ef al. (1983)
Ott et a/. (1982) Vedel er a/. (1983); Kunnath and Locker (1983) Vedel er a/. (1983) Vedel et a/. (1983); Kunnath and Locker (1983) Chen er a/. (1982)
Andrews et a/. (1982)
Yagi and Koshland (1981) Chen er a/. (1982)
Hypomethylated in erythrocytes Hypomethylated in erythroid cells Hypomethylated in oviduct
McGhee and Cinder (1979) Weintraub et al, (198 I); Sanders-Haigh er a/. (1982) Mandel and Chambon ( 1979)
163
DNA METHYLATION IN EUKARYOTIC CELLS TABLE 111 (Continued Species
Gene &Crystalline
Vitellogenin
Xenopus
a2 type I collagen PI-Globin Albumin Vitellogenin A1 and A2
Pattern of methylation Hypomethylated in embryonic lens; hypomethylated in somatic tissues 5' end hypomethylated in response to estrogen 5 ' region hypomethylated in all cells Hypomethylated in erythocytes Hypomethylated in hepatocytes Heavily methylated in all adult tissues
References Jones et al. ( I 98 1); Bower et al. (1983) Wilks et al. (1982)
McKeon et al. (1982) Gerber-Haber et al. (1983)
(genes that are expressed in all cell types) the same pattern of methylation appears to characterize all tissues (Table II), whereas tissue-specific patterns of methylation have been observed in gene sequences that encode tissue-specific functions (Table 111). These data suggest that at the time of cell differentiation a methylation pattern is established and clonally inherited in somatic tissues (Razin and Riggs, 1980).
A. CLONALINHERITANCE OF THE METHYLATION PATTERN Many observations suggest that each cell type has a fixed pattern of methylation which is clonally inherited in each cell generation. Direct evidence for clonal inheritance of a methylation pattern has been provided by DNA-mediated gene transfer experiments (Pollack et al., 1980; Wigler et al., 1981; Stein ef al., 1982a). In these experiments foreign DNA was inserted into mouse L cells and other cell types by DNA-mediated gene transfer. Unmethylated DNA remained unmethylated while methylated DNA remained methylated at the same sites for hundreds of cell generations (Stein et al., 1982a). It has been suggested that methylation inheritance may be achieved by a simple mechanism in which hemimethylated CpG sites in newly replicated DNA are preferentially methylated by a maintenance methylase (Fig. 1 ) . This mechanism is based on two central assumptions which have attained considerable experimental support. (1) The sites are methylated symmetrically on both DNA strands. (2) The preferred substrate for the maintenance methylase is hemimethylated DNA, whereas unmethylated sites are methylated very inefficiently. B. METHYLASES It is feasible that the major methylation activity of any cell occurs at hemimethylated sites created during the process of DNA synthesis. During replica-
164
AHARON RAZIN AND HOWARD CEDAR
FIG. 1 . The mechanism for the inheritance of the methylation pattern
tion, and after repair, the parental strand is methylated while the newly synthesized strand is still unmethylated. It stands to reason, therefore, that the native substrate for the cell methylase is hemimethylated DNA. In bacterial systems if it has been shown that the modification enzymes operating in E . coli B and E. coli K12; M. Eco B (Vovis et al., 1974) and M. Eco K (Burckhardt et al., 1981) methylate unmodified DNA very slowly after an initial lag of 10 minutes, while hemimethylated DNA is methylated by these enzymes rapidly with no apparent lag period. This feature probably plays an important role in restriction modification since, in order to protect the cellular DNA, newly replicating DNA must be rapidly methylated. Foreign DNA, which is generally unmethylated, could undergo methylation inefficiently allowing enough time for DNA restriction. Bacterial methylases that are not involved in restriction modification, the rnec and dam gene products of E . coli, have been studied extensively with respect to substrate specificity (Urieli-Shoval et ul., 1983). In contrast to Type I modification enzymes, both enzymes methylate unmethylated and hemimethylated double-stranded DNA with the same efficiency. The rnec methylase methylates CC,AGG sites at the internal cytosine residues (Schlagman et al., 1976; Razin et al., 1980) and the durn methylase modifies adenine residues at GATC sites (Geirer and Modrich, 1979; Herman and Modrich, 1982). The fact that these methylases methylate with the same efficiency hemimethylated and unmethylated sites suggests that the signal for methylation is the recognition sequence itself, while modification of the sequence has no effect on its specificity. This feature of the E . coli methylases enables their action in trans on extrachromosomal DNA entities, including recombinant DNA plasmids. In contrast to the prokaryotic methylatable sites which were mentioned above, not all eukaryotic CpG sequences are methylated. The methyl groups are, in fact, nonrandomly distributed in a cell-specific pattern which is clonally inherited. Thus the eukaryotic maintenance methylase must be capable of distinguishing between methylated CpG and nonmethylated CpG-containing sequences. This could easily be accomplished if the methylase recognizes and methylates almost exclusively hemimethylated sites which are produced during replication. This site specificity of the eukaryotic DNA methylase was indeed demonstrated in
DNA METHYLATION IN EUKARYOTIC CELLS
165
vitro. The initial methylation rate obtained with a mouse ascites tumor cell methylase preparation was found to be 100-fold higher with hemimethylated +X DNA as compared to single-stranded or double-stranded unmethylated +X DNA. Furthermore, the enzyme methylated the CpG sequences of the hernimethylated DNA specifically and completely (Gruenbaum et al., 1982). Similar results were obtained with methylases from other cells (Taylor and Jones, 1982). These properties of the enzyme are in accordance with the observed clonal inheritance of the methylation pattern in vivo (Pollack e f al., 1980; Wigler et al., 1981; Stein et al., 1982a). The specificity of the methylase for hemimethylated sites guarantees that the modification will be inherited at previously methylated sites, while unmethylated sites will remain unmethylated. In addition, the in vitro and in vivo experiments mentioned above emphasize the high specificity of the enzyme for CpG sequences. This seemingly absolute specificity of the methylase to CpG sequences is consistent with the requirement for symmetrical palindrome sequence to satisfy the model of clonal inheritance of a methylation pattern. Thus none of the other dinucleotide sequences CpA, CpT, or CpC will have a cytosine residue on the complementary strand (Stein et al., 1982a).
C. METHYLATION AT
THE
REPLICATION FORK
The interrelationship between DNA methylation and replication is of special interest since DNA methylation is a postreplicative event. The precise lag time between methylation and replication of DNA has been undecided for a long time. In E . coli it has been recognized for many years that methylation occurs at or near the replication fork (Lark, 1968; Billen, 1968) but no accurate estimation of the lag time was attempted. The rigorous determination of this lag became especially important when DNA methylation was suggested as a strand discrirnination mechanism in mismatch repair (Radman et al., 1978). This proposed model was based on the assumption that a substantial lag exists between methylation and replication that leaves long enough stretches of hemimethylated DNA to be recognized by the mismatch repair apparatus. We now realize that the lag in E . coli (Szyf et al., 1982) as well as in mouse L cells (Gruenbaum et al., 1983) is extremely short. This does not necessarily rule out the strand discrimination theory. Transiently hemimethylated sites might serve the same purpose. It has recently been reported that the theory holds true at least for durn methylation in E . coli (Pukkila et al., 1983). Several studies in the past aimed at the determination of the lag between replication and methylation in mammalian cells yielded conflicting results (Kappler, 1971; Adams, 1971, 1974; Evans et al., 1973; Drahovsky and Wachner, 1975). In one of the early studies, DNA methylation was found to be separated from DNA synthesis by as much as an hour (Burdon and Adams, 1969). This conclusion was drawn from the results of experiments showing that DNA meth-
166
AHARON RAZIN AND HOWARD CEDAR
ylation in mouse L cells continues at a substantial rate for several hours after the inhibition of DNA synthesis by hydroxyurea and that in mouse L cell cultures partially synchronized by aminopterin, the maximum rate of DNA methylation occurred about 1 hour after the maximum rate of DNA synthesis. Similar observations were made by Kappler with a mouse adrenal cell line. When a more direct method was used, however, a lag of approximately 1 minute between the incorporation of deoxycytidine into the DNA and its methylation was observed (Kappler, 1970). This last observation was recently substantiated by experiments following the methylation of incorporated dGTP which was introduced into L cells by cell permeabilization (Gruenbaum et a/., 1983). Since these studies were carried out using unsynchronous cultures, the conclusions apply only to methylation which takes place during cell S phase. It should be noted, however, that methylatiOKof DNA has been observed in G, (Adams, 1971) as well as during mitosis (Bugler et a/., 1980), and methylation of certain sequences has been recently reported to be delayed for up to several hours after strand synthesis but completed before the DNA is replicated in the next cell cycle (Woodcock et al., 1982). In a more careful study of the time course of this delayed DNA methylation, it was shown that the maximum rate of delayed methylation occurred in late S and early G,. The delayed methylation is completed before cells enter G , (Woodcock et al., 1983). The most important conclusion of these observations is that methylation is not necessarily completed during replication. Some sequences may be methylated at later times provided methylation is completed before the next round of replication. This late replication may be a quite efficient process. In an experiment using permeabilized cells when methylation was inhibited during replication by S-adenosyl-L-homocysteine, the resulting hemimethyiated sequences were methylated immediately following the release of inhibition (Gruenbaum et a/., 1983). This kind of late methylation may certainly be needed as part of repair mechanisms. Indeed, a recent study on methylation of repaired patches in human fibroblasts DNA reveals that methylation of repaired sequences is completed in less than 2 hours (Kastan et al., 1982). It should be emphasized that DNA repair without compatible methylation may lead to heritable loss of methylation at some sites which will result in a change of the methylation pattern.
111. DNA Methylation and Gene Expression
A. CORRELATION BETWEEN METHYLATION AND GENEACTIVITY The mode of specific gene expression in eukaryotic organisms has long been the subject of research and speculation. While almost all of the genetic information is present in every differentiated cell, each cell type expresses only a small
DNA METHYLATION IN EUKARYOTIC CELLS
167
subset of the genome pool. Although much of this information is expressed ubiquitously in many cell types, certain sequences are highly specific appearing as transcripts only in one tissue or one cell type. Thus the gene for hemoglobin is abundantly expressed in red blood cell precursors, while it appears to be completely silent in other tissues. The same is true for the chicken ovalbumin gene, which is transcribed exclusively in oviduct cells. In most cases of differential gene expression it has been found that the relevant gene is present in equal copy number and in the same arrangement in all tissues (For a review see Brown, 1981). These observations suggested that gene expression is not controlled by signals written into the sequence of nucleotides, but rather by external factors which exert their effect on the chromosomal conformation of these genes. These predictions have indeed been borne out by experimental findings which show that active genes have a different chromosomal disposition than other parts of the genome which are not being transcribed (for a review see Igo-Kemenes et at., 1982). This phenomenon is usually measured using the nuclear DNA probe DNase I. Weintraub and Groudine (1976) originally showed that whereas the hemoglobin gene is preferentially digested by DNase I in reticulocyte nuclei this same gene is not sensitive to nuclease digestion in nuclei from cells which do not actively express this gene. Identical results have been obtained for other specific genes and for the entire population of genes expressed in a particular cell type (Garel and Axel, 1976; Garel et al., 1977). Sensitivity of genes to DNase I and to other probes appears, in fact, to be closely correlated with the potential of that gene to be expressed. Certain experiments suggest that the high mobility proteins HMG 14 and 17 may play a role in maintaining a gene in its active conformation (Groudine et al., 1981). Other results show that the actual arrangement of the nucleosomes may be modified when a gene is activated. Despite all of these associations it is still not at all clear how changes in chromosome structure might be related to alterations in gene expression. More importantly, it is not at all obvious that chromosomal modifications affect changes in gene expression and are not just the result of these changes. Despite the generally accepted belief that the DNA sequence is preserved during differentiation, recent results strongly indicate that DNA may undergo modification during development. In specific cases, such as that of the immunoglobin gene complex, the differentiation of specific antibody-producing cells is determined by distinct rearrangements of the gene (Seidman and Leder, 1978). In many other cases, the DNA methylation patterns of specific genes have been found to undergo modulation in a tissue-specific manner (for a review, see Razin and Riggs, 1980). A compilation of all of these experiments covering various genes in different organisms indicates that active gene sequences are generally undermethylated with reference to the same gene sequences in tissues where it is not expressed or with reference to germ line DNA.
168
AHARON RAZIN AND HOWARD CEDAR
The assay for site-specific methylation in particular genes is based on the use of methyl-sensitive restriction enzymes (Bird and Southern, 1978). The isoschizomers MspI and HpaII are the most frequently used of these enzymes since both recognize and cut the CpG-containing tetranucleotide CCGG, but HpaII will not cleave this site if the internal cytosine is methylated (Mann and Smith, 1977). Since MspI is insensitive to this methyl moiety, these enzymes can be used in parallel to identify the number and position of these specific sites and their state of methylation (Waalwijk and Flavell, 1978a; Singer er al., 1979; Cedar et al., 1979). The use of the enzyme MspI provides a nice control for validating that changes in a particular restriction pattern are indeed due to DNA methylation. Other enzymes which are sensitive to “CpG in their restriction site may also be used if the digestion pattern of the gene in the genomic DNA is compared with that of the cloned gene. Since these cloned gene sequences are propagated in bacteria they lack the original CpG methylations but may contain bacterial-specific methylations which are irrelevant to animal cell DNA. In a typical experiment cellular DNA from various tissues and from sperm are digested with the appropriate restriction enzymes, size fractionated by agarose gel electrophoresis, blotted onto nitrocellulose filters, and hybridized to a specific radioactively labeled gene probe. The gene-specific restriction pattern is then observed by autoradiography of the hybridized filter. The 30-kb region of human DNA containing the p like globin genes (yA, yG, 6 , p) has been analyzed for methylation by van der Ploeg and Flavell (1980) using this technology. This region contains 17 CCGG sites which have been assayed for methylation in various human tissues. The results show a fairly consistent correlation between gene activity and undermethylation. Sites located in the vicinity of the adult gene, for example, are heavily methylated in various tissues and in fetal liver, but are relatively undermethylated in bone marrow precursor cells which are actively engaged in the synthesis of globin RNA. In contrast to this the fetal genes are undermethylated in the fetal liver and methylated in all adult tissues. Similar results have been obtained by other laboratories studying the globin gene family. In the rabbit globin gene complement Shen and Maniatis (1980) found three specific HpaII sites which are modulated during differentiation. The most clear-cut picture has been observed in the chicken globin system, where the globin DNA of active cells is completely unmethylated at HpaII sites, while other tissues contain methylations at these same sites (Weintraub el al., 1981). This clear differentiation between modified and unmodified sites in the chicken globin gene is probably attributable to the fact that it is easier to obtain pure populations of cells in this system. The implication, if this is correct, is that differentiation is accompanied by “all or none” site-specific changes in the methylation pattern. The tendency toward undermethylation in genes which are actively being expressed is also apparent for other specific gene sequences. The chicken
DNA METHYLATION IN EUKARYOTIC CELLS
169
ovalbumin gene, for example, is undermethylated in the oviduct in comparison to other chicken tissues (Mandel and Chambon, 1979). These authors point out, quite correctly, that there are three types of methylatable HpaII sites in all animal cell DNA. While a few sites are modulatable and differ in their degree of methylation in various tissues other sites remain either always methylated or always unmethylated. Thus the total number of modulatable locations in the animal cell genome may be relatively small. Another gene which shows that changes in methylation are correlated with differentiation, is the J protein gene in the immunoglobin system (Yagi and Koshland, 1981). This gene, whose protein product is involved in antibody processing, is expressed in the same plasma cells which specifically produce antibody molecules. Recent evidence clearly shows that this gene is undermethylated in plasma cells as compared to other tissues in which this gene is inactive. This correlation extends to other gene functions which have been found to be active in certain cells growing in culture. Hamster cells, for example, have been transformed by adenovirus and the viral sequences analyzed for methylation using the enzymes HpaII and MspI (Sutter and Doerfler, 1980). The viral functions which are actively expressed in these cells were found to be undermethylated, while other nonexpressing genes were heavily methylated at many sites in the DNA. A similar phenomenon was found for cells transformed with Herpes sairniri (Desrosiers et al., 1979). Furthermore, in cells infected with MuLV and actively expressing viral functions, these viral sequences were unmethylated while the endogenous mouse viral sequences are silent and thoroughly methylated (Harbers et al., 1981). In almost all studies dealing with the methylation pattern of specific genes the presence of methylation was determined by the use of the restriction enzyme HpaII and its isoschizomer Mspl. Despite its convenience it should be recalled that methyl moieties at these specific sites (CCGG) represent only a small fraction of the total methylations of any particular gene. If 5mC is generally found at CpG sequences, methylation at the CCGG site will represent only about 6% of the total methylated sequences. Thus the methylation pattern obtained with these enzymes provides only partial information and not necessarily a representative one. The use of additional methyl-sensitive enzymes such as AvaI or Hhal may provide some additional mapping information, but these sites are also widely spaced. Thus, in these studies, the methylation map is basically very sketchy and methylation in key regions of the genomes may go undetected if they lack convenient restriction sites. An additional problem with these results is that they relate to special genes which are generally expressed as abundant RNA species. These studies actually reveal nothing about the rare class of gene sequences which make up the majority of the expressed gene complement of any cell. Recent experiments indicate that the correlation between undermethylation and gene expression may indeed be a general phenomenon true of all CpG
170
AHARON RAZIN AND HOWARD CEDAR
residues in all active genes (Naveh-Many and Cedar, 1981). In these studies two separate approaches were used to measure the methylation state of active genes. The degree of methylation of total DNA was determined by a method which measures the percent modification of all CpG residues (Gruenbaum et al., 1981b). DNA is nicked and labeled in the presence of only one nucleotide I3*P1dGTP and the resulting DNA is subjected to nearest neighbor analysis. The ratio of labeled Cyt to 5mC on thin layer chromatograms conveniently provides direct information on the extent of CpG methylation. Since methyl moieties are found almost exclusively at this dinucleotide, this method essentially examines all potentially methylatable sites. Mouse liver DNA, for example, is 80% methylated by this criterion (Razin et al., 1984). Taking advantage of the well known observation that active genes are preferentially digested by DNase I in nuclei (Weintraub and Groudine, 1976), it was possible to selectively label the active regions of the DNA exclusively simply by using the nearest neighbor method on isolated nuclei. When this labeled DNA was assayed for methylation it was found that active regions were about 40% methylated, indicating that active genes are considerably undermethylated compared to total DNA (Naveh-Many and Cedar, 1981). An alternate technique for assaying methylation in active regions was to label total DNA with [32P]dGTP and isolate those sequences which are actively expressed in the cell by hybridization using excess total RNA from that particular cell (Naveh-Many and Cedar, 1981). This DNA corresponding to the transcribed genes was also undermethylated when compared to total DNA. While these results indicated that active genes are undermethylated they also showed that the demethylation is not complete. This either means that many sites in active regions remain methylated or that the criteria used for preferentially studying active genes were not selective enough. While all of the above results clearly show that there is a close correlation between gene expression and undermethylation, these experiments provide no information as to the role of methylation in the regulation of genes. More direct experiments are required in order to determine whether DNA modification is determining potential gene expression or simply a result of other factors which determine gene activity. According to some models, decisive alterations in the methylation pattern which occur during differentiation determine the tissue specificity of the expression of mammalian genes (Holliday and Pugh, 1975; Riggs, 1975). B . CAUSEAND EFFECT Another possible approach for studying the effect of methylation on gene expression involves the use of drugs which specifically cause DNA demethyla-
DNA METHYLATION IN EUKARYOTIC CELLS
17 1
tion. In an effort to study the effect of methylation on gene expression, for example, Christman et al. (1977) treated Friend erythroleukemia cells with agents that prevent methylation such as ethionine, butyrate, and dimethyl sulfoxide. After several generations in the presence of these hypomethylating agents, globin synthesis was detected. It should be noted, however, that the globin gene is expressed at low levels even in the absence of inducer. The most powerful and selective of these agents is thought to be 5-azacytidine which causes demethylation probably as a result of its incorporation into DNA (Jones and Taylor, 1980). In this model such 5-azacytidine-substituted DNA would not be methylated following replication, and this alteration would then be inherited in 50% of the daughter cells (Taylor and Jones, 1982). This drug has indeed been found to cause induction of the globin gene in Friend erythroleukemia cells (Creusot et al., 1982), to change the differentiation pattern of mouse embryo cells (Taylor and Jones, 1979), activate human X chromosome genes (Mohandes et al., 1981; Jones et al., 1982), and activate EBV virus in latently infected human lymphoid cell lines (Ben-Sasson and Klein, 1981). Two recent studies strongly support the hypothesis that DNA methylation may repress gene expression. In chicken fibroblast cells growing in culture the endogenous viral genes AEV are found to be inactive in the sense that viral RNA and proteins are not produced from these gene copies (Groudine et al., 1981). These genes are also heavily methylated at numerous HpaII sites. Treatment of such cells with 5-azacytidine leads to the induction of these genes and the synthesis of viral products and these changes are accompanied by a reproducible reduction in the methylation state of the genes. Similar results have been reported for the mouse metallothionein I gene (Compere and Palmiter, 1981). While cadmium or glucocorticoids generally cause the induction of this gene, one particular mouse tymoma cell line was found to be unresponsive to this treatment. Blot hybridization analysis revealed that these genes were indeed present in this cell line but were highly methylated at every HpaII site. Following treatment with 5-azacytidine and selection with cadmium, resistant inducible clones were indeed isolated and the metallothionein gene in these clones was completely unmethylated suggesting that the drug induced demethylation led to the inducibility of this gene. In spite of the apparent straightforward nature of experiments involving the use of 5-azacytidinq some words of caution are in order. While this drug does lead to DNA demethylation, the exact mechanism of its action has not been determined. One particular troubling observation is that the extent of DNA demethylation produced by this drug is far in excess of the amount of DNA substitution by 5-azacytidine, suggesting that replacement of 5-methylcytosine is not the major element in the demethylation. Experiments in vitro have demonstrated that this compound does not inhibit the DNA methylase reaction, but it may indeed inhibit some other cell functions (Adams et al., 1982). On the other
172
AHARON RAZIN AND HOWARD CEDAR
hand, DNA substituted with azaCtd irreversibly inhibits in vitro methylases that methylate cytosine residues like E. coli mec methylase, HpaII methylase and EcoRII methylase (Friedman, 198 1). Another serious criticism of experiments employing 5-azacytidine is that its action is apparently specific to certain areas of the genome. This was clearly seen in the studies of Groudine et al. (1981) in which Sazacytidine was shown to induce the chicken endogenous viral genes (AEV). While these genes underwent considerable demethylation, the globin genes in these same cells remained normally methylated. Since the proposed mechanism of action of 5-azacytidine would predict random demethylation of all areas of the genome, these results suggest that this drug may be causing induction by some other unknown mechanism.
C. MECHANISMS OF BLOCKING GENEACTIVITY 1. EfSect of Nucleosomal Structure
The eukaryotic genome is a complex structure which must be defined in terms of its chromosomal conformation as well as its DNA sequence. While all of the cellular DNA is probably packaged into a nucleosomal structure, discrete variations in this structure are found in the regions surrounding specific gene sequences. In particular actively expressed genes are probably in a more exposed chromosomal conformation than other areas of the genome as determined by the access of these genes to a variety of DNA probes including RNA polymerase and various nucleases (Weisbrod, 1982). The most useful of these probes is DNase I which preferentially digests active genes in whole nuclei preparations (Weintraub and Groudine, 1976). This one to one correlation between DNase I sensitivity and gene expression suggests that changes in chromosome conformation may be an integral part of the mechanism of gene regulation. Although it is known that HMG proteins 14 and 17 are involved in maintaining the DNase Isensitive structure (Groudine et d.,198l), the exact conformational alterations associated with active genes has not been elucidated. If as suggested by the evidence presented in this review methylation is involved in the process of gene regulation, it is intriguing to speculate that this DNA modification may operate by altering DNA protein interactions in the chromosome. While there is no direct evidence to support this theory, many studies suggest such a role for DNA methylation. The most direct evidence for a link between protein structure and DNA methylation comes from studies using various nucleases to probe the placement of methyl moieties in chromatin. Micrococcal nuclease, for example, which can maximally digest approximately 50% of the chromosomal DNA, solubilized only 25% of the DNA methyl groups, suggesting that methylated cytosines are preferentially protected by
DNA METHYLATION IN EUKARYOTIC CELLS
173
chromosomal protein and are therefore less susceptible to nuclease attack (Razin and Cedar, 1977). A study of the kinetics of this nuclease digestion confirmed this result and show that an inordinate percentage of the total DNA 5mC population is concentrated within nucleosomal core particles (Solage and Cedar, 1978). These experiments clearly show that DNA methylation is not randomly distributed with respect to nucleosome structure. Experiments with the enzyme DNase I lead to a similar conclusion. A clear correlation exists between DNA undermethylation and sensitivity to nuclear DNase I digestion (Naveh-Many and Cedar, 1981). This result is not surprising in light of the fact that both of these parameters are associated with active genes. No careful analyses of the effect of CpG methylation on DNA-protein interaction have yet been reported, but experiments in other systems suggest that methyl groups may be a factor in controlling protein-DNA interactions. The methyl group of one thymine residue in the lac operator sequence changes the affinity of the repressor for this DNA by a factor of 20 (Fisher and Caruthers, 1979). In this regard it should be pointed out that almost all restriction enzymes recognize the presence of methyl moieties in their recognition site and are specifically inhibited by this modification (Roberts, 1976). Recent exciting experiments have shown that the methylated polymer (dG-SmdC),, can be converted to Z-form DNA using salt concentrations three orders of magnitude less than that required to convert the unmethylated polymer (Behe and Felsenfeld, 1981). Furthermore, this ZDNA does not get readily packaged into a nucleosome structure (Nickol et al., 1982). While these simple polymers probably do not exist in vivo, it is not hard to imagine that a few CpG methylations may lead to localized alterations in the DNA and chromatin structure in normal cells. It has recently been shown that there are S1 sensitive sites present in specific genes cloned into supercoiled plasmids (Weintraub, 1983). These sites probably arise from the formation of palindromic structures, stretches of homopolymers or other potential deviations from pure B-form DNA. Furthermore, in some cases these sites have also been identified in isolated nuclei (Larsen and Weintraub, 1982). One prevailing theory suggests that this tertiary structure may dictate the chromosomal conformation which would then determine gene activity. If DNA methylation were to have an affect on local DNA tertiary structure, this might provide a possible explanation for the effect of this modification on gene regulation. It is clear that our knowledge of chromatin structure is not sufficient to enable one to understand all of the protein-DNA interactions. When HMG proteins are removed from nuclei by treatment with 0.4 M NaCl, the active genes undergo a conformational change and lose their sensitivity to DNase I digestion (Groudine et al., 1981). It is remarkable that reconstruction of these nuclei with added HMG proteins leads to the restoration of this sensitivity at the same identical sites. This suggests that other factors on the chromosome in the region of the
174
AHARON RAZIN AND HOWARD CEDAR
active genes direct the correct placement of the HMG proteins. Although we do not know the nature of these factors, it is tempting to speculate that DNA methylation may be involved. 2. Effect of Methylation on the Expression of a Gene
In order to conclusively show that the presence of methyl moieties on DNA directly effect expression one must study the expression of genes after artificially intervening by either adding or removing methyl groups from the specific DNA. Both of these approaches have been used and the results strongly suggest that methylation directly affects gene expression. The method of DNA-mediated gene transfer provides an ideal system for asking these questions. In this system a cell line carrying an auxotrophic mutation can be transformed with a naked DNA molecule carrying the missing gene activity (Pellicer et a f . , 1978). Using selective medium, cells receiving the added gene function grow into colonies which can be counted and subsequently isolated by growth into mass culture. Mouse tk- L cells, for example, have been efficiently transformed with the Herpes simplex virus (HSV) thymidine kinase (tk) gene using this technique of DNA mediated gene transfer. When the tk gene was methylated in vitro and used in a transforming efficiency assay, it was found that both the HSV and chicken methylated tk genes were 2-10 times less effective than their unmethylated counterparts (Pollack et al., 1980; Wigler er al., 1981). While apparently demonstrating the effect of methylation on gene expression, these studies suffer from two basic problems. Since transformation efficiency was used as an assay it is not clear whether the methylation in these experiments affected the expression of the tk gene or simply the process of gene transfer. Furthermore clones transformed with the methylated tk genes were found to have undergone massive demethylation in the region of the tk gene, suggesting that the cells may have some mechanism for demethylating genes which are subjected to strong selective pressure. When cells are transformed with a vector using selective pressure it was found that other DNA species present during the transfer procedure are incorporated into the resulting clones together with the vector DNA (Wigler et u l . , 1979). This procedure of cotransformation can be used to introduce any gene sequence into mammalian cells regardless of whether a selection procedure is available for this sequence. When methylated DNA is inserted into cells by this procedure, the cotransforming sequences retain 70-80% of their original methyl moieties (Pollack et al., 1980; Wigler et al., 1981; Stein et al., 1982a). Taking advantage of this fact, it was possible to introduce both unmethylated and in vitro methylated hamster aprt genes into tk- , aprt- mouse L cells, using tk selection to obtain transformed clones (Stein e f al., 1982b). Resulting colonies were grown to mass culture and then exposed to azaserine to select for the aprf+ phenotype. While almost 100% of those clones containing unmethylated uprt genes were found to
DNA METHYLATION IN EUKARYOTIC CELLS
175
be viable in azaserine selection, over 95% of the clones containing methylated uprr genes had the uprt- phenotype by this criterion. Furthermore these clones did not produce any aprt specific RNA. Blot hybridization analysis confirmed the presence of the uprt gene in all clones and the stable methylation pattern in those cells which received the in vitro methylated aprt genes. Although cells harboring methylated uprt genes were unable to grow in azaserine-containing medium, prolonged exposure to this drug revealed the presence of a small number of uprt+ revertants in the population. As confirmation of the correlation between gene expression and undermethylation it was observed that the uprt genes within these revertants underwent demethylation. It is noteworthy that this demethylation was always accompanied by gene amplification or some other reorganizational change (Stein et ul., 1982b). These studies indicate that methylation may repress the expression of genes in animal cells, but do not provide information on the mode of this inhibition. In order to test the possibility that methylation exerts its affect at the level of transcription, Vardimon et al. (1982) tested the effect of methylation on the expression of genes injected into Xenopus oocytes. In this system, injected genes are transcribed efficiently and in a manner similar to that observed in vivo. The E2B gene, involved in the replication of DNA in adenovirus-transformed cells, for example, serves as a template for a defined mRNA in these cells. This DNA is transcribed in Xenopus oocytes in exactly the same manner yielding a mRNA molecule of identical size and initiation point. This gene, following methylation in vitro using the HpuII methylase, was found to be inactive in the oocyte system. Controls using DNA methylated by other methylases which do not recognize CpG residues confirmed that this effect was due to the presence of methyl moieties at the proper sites. Similar findings have been observed by studying the transcription of SV40 genes (Fradin et ul., 1982). These experiments are particularly noteworthy since SV40 contains only one CCGG site which is methylatable by the HpuII methylase. Methylation at this unique site represses the SV40 late transcripts while there was no effect on the transcription of early region RNA. Since this site is located in the 5' leader sequence, these results suggest that methylation in this region is sufficient to inhibit gene expression. Since DNA injected into oocytes probably assumes a nucleosomal structure soon after insertion, it is not clear from these experiments whether methylation directly influences transcription or whether the primary effect is on chromosome conformation, which in turn influences the transcription process. This might be answered by studying the effect of methylation in cell free in vitro transcription systems. DNA methylation has also been shown to inhibit the infectivity of retroviruses. When a proviral genomic clone of the MuLV virus was methylated by a eukaryotic DNA methylase, its biological activity was reduced by over 3 orders of magnitude (Simon et ul., 1983). This result confirmed early reports that
176
AHARON RAZIN AND HOWARD CEDAR
methylated integrated copies of this virus were also noninfective (Stuhlmann et a l . , 1981; Harbers et a l . , 1981; Chumakov et a l . , 1982).
3 . X Chromosome Inactivation One of the most intriguing questions of molecular biology is the phenomenon of X chromosome inactivation found in many mammals. The mechanism of inactivation of one of the X chromosomes in female cells which takes place at early stages of embryogenesis has long been thought to involve alterations in the chromosome structure, since this process is accompanied by a striking condensation of the inactive material. One proposed model for X-inactivation was based on DNA methylation (Riggs, 1975). Recent evidence indeed supports that Xinactivation may involve modification of the DNA of this chromosome. Since until recently no genetic hybridization probes were available for analyzing the X chromosome, these conclusions are based on relatively indirect experiments which take advantage of the selectable hgprt gene located on this chromosome. Mohandes et al. (198 l), Jones et al. (1982), and Lester et al. (1982) constructed a cell line which contained one mutant hgprt and one normal hgprt gene present on the inactive X chromosome. Thus these cells were phenotypically hgprt- as determined by their inability to grow in selective medium (HAT). When such cells were treated with 5-azacytidine and exposed to HAT medium, hgprt+ cells could be selected suggesting that reactivation of the inactive X chromosome may be produced by demethylation of the DNA. Moreover, after treatment with 5azacytidine, the hgprt gene on the human X chromosome is expressed in gene transfer experiments (Venolia et a l . , 1982). In all these experiments there was no evidence, however, that this treatment caused demethylation of the relevant sequences. Further indirect evidence of the role of DNA modification in this process was provided by Liskay and Evans (1980) who showed that the hgprt gene present on the inactive X chromosome was unable to transfer hgprt- cells using DNA mediated gene transfer. This approach supports the hypothesis that X inactivation may be accompanied by changes in the DNA as well as the chromosome structure but no evidence was presented to suggest that DNA methylation is involved in this process. Decisive experiments on the nature of this inactivation process will be obtained only when molecular probes for genes on the X chromosome will be used. D. DIFFERENTIATION
The general picture of methylation inheritance in normal cells is that the pattern of methylation is generally stable and reproducibly transferred from generation to generation at each cell division. On the other hand certain specific methylatable sites have been found to be modulated during differentiation of the organism in a programmed manner such that specific active genes are relatively
DNA METHYLATION IN EUKARYOTIC CELLS
177
undermethylated in tissues in which they are expressed. These observations are consistent with a model of differentiation in which discrete changes in the methylation pattern dictate the characteristics of differentiating cells (Razin and Riggs , 1980). Although only a few genes have been analyzed in detail it is noteworthy that in many cases cell differentiation is accompanied by a reduction in the methyl complement of specific genes. This can be seen by comparing the methylation pattern of gene sequences in sperm DNA with the pattern observed in specific tissues (Ehrlich et al., 1982; Razin et al., 1984). The exact mechanism of this specific demethylation is not known but several types of models may be put forward to explain this phenomenon. The most attractive of these models is that following DNA replication certain sites fail to be methylated by the maintenance methylase of the cell. While resulting cells would contain hemimethylated sites, these sites would segregate during the next cell division to yield one cell methylated at these sites and one cell with unmethylated sites. This process will generate many different cell types each having a different methylation pattern as well as different biological characteristics. While there is no evidence for this theory it should be noted that the drug Sazacytidine, which depresses methylation following replication, indeed causes biological changes mimicking those of the differentiation process (Jones and Taylor, 1980). This model would predict that germ line DNA would be more heavily methylated than DNA from other animal tissues. This simple view is certainly not correct since a careful study of the total CpG methylation content of various mouse tissues showed that all adult tissues and 9-10 day embryos (75-80% CpG methylation) were considerably more methylated than sperm DNA (67%) (Table I). Surprisingly, the extraembryonic tissues were found to be very undermethylated (40%) in comparison to sperm (Razin et al., 1984). These results suggest that while somatic cells have a stable methylation pattern which is faithfully inherited from generation to generation, cells undergoing differentiation during early development are characterized by the ability to alter this pattern either by demethylation or by de novo methylation. In addition to the overall tissue specific demethylation events described above for certain genes, large demethylations have been observed in several systems. Mouse teratocarcinoma cells which are induced to differentiate by treatment with retinoic acid undergo a concomitant 30% reduction in their methylation content. This induction mimics the in vivo formation of primitive endoderm in early embryos and the observed demethylation evidently corresponds to the changes seen during the formation of yolk sac and placenta in vivo. This demethylation is spread out over a large number of sites in the mouse genome but does not seem to correlate with any changes in gene expression, and the mechanism of this process is not known. A large number of tumor cell lines are also undermethylated (Ehrlich et al., 1982; Diala and Hoffman, 1982; Mermod et al., 1983; Wilson
178
AHARON RAZIN AND HOWARD CEDAR
and Jones, 1983) and analysis of several human tumors shows that this lack of methyl moieties is observed in a large number of various genes (Feinberg and Vogelstein, 1983; Diala et al., 1983). Recent studies suggest that carcinogenic agents may lead to DNA demethylation by a variety of mechanisms including direct inhibition of the cell’s maintenance methylase (Wilson and Jones, 1983). The significance of these massive drifts in the DNA methylation pattern both in extraembryonic tissues and in tumorogenic cells is as yet unresolved, but these changes may reflect changes in gene expression. Although the analysis of the overall content of CpG methylation suggests that de now methylation may occur during normal animal development (Razin et al., 1984; Manes and Menzel, 1981), specific examples of such modifications are sparce. Certain satellite sequences were indeed found to be more methylated in specific tissues as compared to sperm DNA (Sturm and Taylor, 1981; Pages and Roizes, 1982; Gama-Sosa et al., 1983). Several lines of evidence suggest that cells during early development as opposed to somatic cells may have the potential for de novo methylation. When early mouse embryos before the stage of implantation are exposed to MuLV, copies of this virus are integrated into the cellular genome, but this does not result in a productive infection (Jahner er al., 1982). Analysis of DNA from such infected embryos revealed that these proviral sequences underwent de now methylation. This same process was also observed following the introduction of MuLV into teratocarcinoma cells (Stewart et al., 1982). In this case it was shown that the de novo methylation occurred slowly over a period of 8- 16 days and was probably not the direct factor which inhibited the expression of these virus sequences (Niwa et al., 1983; Gautsch and Wilson, 1983). Other foreign sequences introduced into teratocarcinoma cells by DNAmediated gene transfer also undergo de novo methylation (Cedar et al., 1983). Finally, it has been shown that DNA injected into mouse zygotes becomes methylated in many tissues in which the DNA is integrated (Palmiter et al., 1982). In all of these cases the modification was described for foreign DNA sequences and it cannot be concluded that this is a general process which occurs during normal development. While the above suggested differentiation models based on DNA methylation seem to be supported by various experimental approaches, it should be kept in mind that DNA methylation may not be the only mechanism involved in differentiation. Many insects, including Drosophila, lack any methyl moieties either as 5mC or 6mAde (Urielli-Shoval et al., 1982).
IV. Conclusions From the data discussed in this review it is clear that the DNA methylation pattern undergoes discrete changes, especially during the process of differentia-
DNA METHYLATION IN EUKARYOTIC CELLS
179
tion, and that these changes may be intimately related to gene expression. The mechanisms involved in the processes of de n o w methylation and demethylation are not well understood. Demethylation may occur by several possible mechanisms. The idea of a demethylase which could remove methyl groups from DNA has been suggested (Gjerset and Martin, 1982), but there is no further experimental support for such a mechanism either in bacterial or in eukaryotic cells. The most reasonable model to explain demethylation is that in which replication occurs in the absence of maintenance methylation. If the CpG site remains hemimethylated through the cell cycle, when the DNA segregates during replication and mitosis one daughter cell will inherit a fully methylated site while this site will be unmethylated in the sister cell. Although the factors which participate in generating specific demethylations are not known, it is tempting to speculate that these factors are associated with an interplay between DNA replication and methylation. A. DEMETHYLATION MODELS Since the mechanism of methylation is intimately connected to that of replication, one would predict that the degree of demethylation could depend on the rate of DNA replication on the one hand, and the intracellular level of DNA methylase on the other. Furthermore, other processes which occur simultaneously with replication, such as transcription, could also modulate the level of methylation at specific sites. Assuming that the DNA maintenance methylase is at a limiting level in the cell nucleus, an increase in the rate of DNA replication may lead to undermethylation of the DNA. This is probably the mechanism for demethylation of sequences which undergo amplification. Although most DNA sequences undergo replication once during the cell cycle certain regions may undergo multiple rounds of replication leading to an increase in their copy number in the genome. The ribosomal genes of Xenopus, for example, are amplified 100,000-fold at early stages of the development of this organism in preparation for the massive level of protein synthesis which takes place in early embryo. While the amplified genes are completely unmethylated, the original genomic unamplified gene is heavily methylated (Bird and Southern, 1978; Bird et al., 1981a). It is possible that the sudden increase in DNA replication during the amplification process caused an effective demethylation. A similar demethylation phenomenon was observed for an introduced methylated a p t gene which underwent amplification during selection for a p t + clones (Stein et al., 1982b). Furthermore, viral sequences in many instances were found to be methylated when integrated while unmethylated in their extrachromosomal state (Gunthert et al., 1976; Desrosiers et al., 1979; Cohen, 1980). While this demethylation might have been due to rapid replication, another possibility is that there is no maintenance methylation of extra-
180
AHARON RAZIN AND HOWARD CEDAR
chromosomal DNA. In all of these cases one could speculate that rapid replication in a situation where the DNA methylase is limiting leads to a specific demethylation event. It has recently been shown that the DNA methylases in E . coli are indeed at a limiting level. Extrachromosomal DNA such as A phage DNA and pBR322 in fact undergo specific demethylation during the phase of rapid replication of these DNAs (Szyf et al., 1984). Demethylation in general has been associated with specific active genes and a large body of evidence shows a clear correlation between these two processes. Although methylation has been shown to inhibit gene activity, there is no clear cut evidence showing that in vivo demethylation precedes changes in gene activity and a considerable amount of data actually suggest that the undermethylation of certain active genes actually occurs following the activation of these sequences. In light of the close relationship between replication and methylation, it is tempting to speculate that methylation may be inhibited in regions of DNA which are undergoing massive transcription. Over several generations this would result in a gene-specific pattern of undermethylation, and would explain the low methylation level observed over the transcribed regions of genes such as globin (McGhee and Ginder, 1979; van der Ploeg and Flavell, 1980; Sanders-Haigh et al., 1982; Shen and Manniatis, 1980; Weintraub et al., 1981), ovalbumin, (Mandel and Chambon, 1979), &crystalin (Jones et al., 1981), and others in the tissues in which these genes are expressed. Experimental support for this theory comes from studies of the methylation pattern of liver-specific genes. The chicken villogenin gene (Wilks et al., 1982) and the rat albumin gene (Vedel et al., 1983; Andrews et al., 1982) were both found to be active in late embryo livers, yet these genes were fully methylated at this stage (as determined by analysis of certain specific restriction sites). The demethylation of these genes is only observed in adult livers of these organisms. The rat gene for PEPCK, although not active until birth, can be induced to almost full activity in late embryo livers and is also heavily methylated at this stage (Benvenisty et al., 1984). Furthermore, the fully methylated Xenopus vitellogenin genes can be stimulated to full expression by estrogen treatment (Gerber-Haber et al., 1983). These studies clearly suggest that changes in DNA methylation can follow alterations in gene expression. In this regard, it is interesting that unlike other tissue-specific genes the gene for muscle ci-actin is heavily methylated in all tissues of the rat including skeletal muscle (Shani et al., 1984). Although this gene is heavily transcribed in muscle cells we suggest that demethylation is impossible in this system since this gene is activated only after the muscle cells have ceased to divide. In the absence of replication, demethylation presumably cannot occur. B. MAINCONCEPTS The main concepts emerging from the data reviewed in this article are the following: (1) DNA methylation is eukaryotic cells serves as a gene silencing
DNA METHYLATION IN EUKARYOTIC CELLS
181
mechanism. (2) This function is accomplished by a tissue-specific DNA methylation pattern which is established in the embryo and subsequently inherited in the somatic tissue by a maintenance methylase. (3) The formation of the pattern involves changes in the methylation of specific sites which could be achieved by an interplay between replication and transcription rates, on one hand, and the intracellular level of the DNA methylase on the other hand.
ACKNOWLEDGMENTS We are grateful to J. Yisraeli and M. Szyf for helpful suggestions and to Caroline Gopin for preparing the manuscript. This work was supported by N.I.H. Grant GM 20483 and U.S.-Israel Binational Science Foundation Grant 2333180.
REFERENCES Adams, R. L. P. (1971). Biochim. Biophys. Actu 254, 205-212. Adams, R. L. P. (1974). Biochim. Biophys. Actu 335, 365-373. Adams, R. L. P., Fulton, J., and Kirk, D. (1982). Biochim. Biophys. A m 697, 286-294. Andrews, G . K., Dziadek, M . , and Tamaoki, T. (1982). J . B i d . Chem. 257, 5148-5153. Behe, M., and Felsenfeld, G. (1981). Proc. Natl. Acud. Sci. (I.S.A. 78, 1619-1623. Ben-Sasson, S . A., and Klein, G. (1981). Int. J . Cancer 28, 131-135. Benvenisty, N., Mencher, D., Meyuhas, O., Razin, A., and Reshef, L. (1984). Proc. Narl. Acud. Sci. (I.S.A., in press. Billen, D. (1968). J . Mol. Biol. 31, 477-486. Bird, A. P. (1978). J . Mol. Biol. 118, 49-60. Bird, A. P., and Southern, E. M. (1978). J . Mol. Biol. 118, 27-47. Bird, A. P . , and Taggart, M. H. (1980). Nucleic AcidsRes. 8, 1485-1497. Bird, A. P . , Taggart, M. H., and Gehring, C. A. (1981a). J . Mol. Biol. 152, 1-17. Bird, A., Taggart, M., and Macleod, D. (1981b). Cell 26, 381-390. Bower, D. J., Errington, L. H., Cooper, D. N . , Morris, S., and Clayton, R. M. (1983). Nucleic Acids Res. 11, 2513-2527. Brown, D. D. (1981). Science 211, 667-674. Bugler, B., Bertaux, O., and Valencia, R. (1980). J . Cell. Physiol. 103, 149-157. Burckhardt, J . , Weisemann, J . , and Yuan, R. (1981). J . B i d . Chem. 256, 4024-4032. Burdon, R. H . , and Adams, R. L. P. (1969). Biochim. Biophys. Acru 174, 322-329. Burdon, R. H . , and Adams, R. L. P. (1980). Trends Biochem. Sci. 5, 294-297. Busslinger, M., de Boer, E., Wright, S . , Grosveld, F. G . , and Flavell, R. A. (1983). Nucleic Acids Res. 11, 3559-3569. Cate, R. L . , Chick, W . , and Gilbert, W. (1983). J . Biol. Chem. 258, 6645-6652. Cedar, H., Solage, A . , Glaser, G., and Razin, A. (1979). Nucleic Acids Res. 6, 2125-2132. Cedar, H., Stein, R., Gruenbaum, Y., Naveh-Many, T., Sciaky-Gallili, N., and Razin, A. (1983). Cold Spring Harbor Symp. Quant. Biol. 47, 605-609. Chen, Y.-T., Negishi, M., and Nebert, D. W. (1982). DNA 1, 231-238. Christman, J . K., Price, P., Pedriman, L., and Acs, G. (1977). Eur. J . Biochem. 81, 53-61. Chumakov, I . , Stuhlmann, N., Harbers, K., and Jaenisch, R. (1982). J . Virol. 42, 1088-1098. Cohen, J . C. (1980). Cell 19, 653-662. Compere, S. J., and Palmiter, R. D. (1981). Cell 25, 233-240.
182
AHARON RAZIN AND HOWARD CEDAR
Creusot, F., Acs, G., and Christman, J. K. (1982). J . Biol. Chem. 257, 2041-2048. Dackowski, W., and Morrison, S. L. (1981). Proc. Natl. Acad. Sci. U.S.A. 78, 7091-7095. Desrosiers, R. C., Mulder, C., and Fleckenstein, B. (1979). Proc. Natl. Acad. Sci. U.S.A. 76, 3839-3843. Deumling, B. (1981). Proc. Natl. Acad. Sci. U.S.A. 78, 338-342. Diala, E. S . , and Hoffman, R. M. (1982). Biochem. Biophys. Res. Commun. 107, 19-26. Diala, E. S., Cheah, M. S. C., Rowitch, D., and Hoffman, R. M. (1983).J . Natl. Cancer Inst., 71, 755-764. Doerfler, W. (1981). J. Gen. Virol. 57, 1-20. Doerfler, W. (1983). Annu. Rev. Biochem. 52, 93-124. Drahovsky, D., and Boehm, T. L. J . (1980). Int. J . Biochem. 12, 523-528. Drahovsky, D., and Wachner, A. (1975). Naturwissenschaften 62, 189-190. Ehrlich, M., and Wang, R. Y. H. (1981). Science 212, 1350-1357. Ehrlich, M., Gamn-Sosa, M. A., Huang, L. H., Midgett, R. M., Kuo, K. C., McCune, R. A., and Gehrke, C. (1982). Nucleic Acids Res. 10, 2709-2721. Evans, H. H., Evans, T. E., and Littman, S. (1973). J . Mol. Biol. 74, 563-572. Feinberg, A. P., and Vogelstein, B. (1983). Nature (London) 301, 89-91. Felsenfeld, G., and McGhee, J . (1982). Nature (London) 296, 602-603. Fisher, E. F., and Caruthers, M. H. (1979). Nucleic Acids Res. 7, 401-416. Fradin, A,, Manley, J. L., and Prives, C. L. (1982). Proc. Natl. Acad. Sci. U.S.A. 79,5142-5146. Friedman, S. (1981). Mol. Pharmacol. 19, 314-320. Gama-Sosa, M. A,, Wang, R. Y. H., Kuo, K. C., Gehrke, C. W., and Ehrlich, M. (1983). Nucleic Acids Res. 11, 3087-3095. Garel, A , , and Axel, R. (1976). Proc. Natl. Acad. Sci. U.S.A. 73, 3966-3970. Garel, A., Zolan, M., and Axel, R. (1977). Proc. Natl. Acad. Sci. U.S.A. 74, 4867-4871. Gautsch, J. W., and Wilson, M. (1983). Nature (London) 301, 32-37. Geirer, G. E., and Modrich, P. (1979). J. Biol. Chem. 254, 1408-1413. Gerber-Haber, S . , May, F. E. B . , Westlay, B, R., Felber, B . K . , Hosbach, H. A., Andres, A.-C., and Ryffel, G . V. (1983). Cell 33, 43-5 1. Gjerset, R. A , , and Martin, D. W. (1982). J . Biol. Chem. 257, 8581-8583. Groudine, M., Eisenman, R., and Weintraub, H. (1981). Nature (London) 292, 311-317. Gruenbaum, Y., Naveh-Many, T., Cedar, H., and Razin, A. (1981a). Nature (London) 292, 860862. Gruenbaum, Y., Stein, R., Cedar, H., and Razin, A . (1981b). FEBS Lett. 124, 67-71. Gruenbaum, Y., Cedar, H., and Razin, A. (1982). Nature (London) 295, 620-622. Gruenbaum, Y., Szyf, M., Cedar, H., and Razin, A. (1983). Proc. Nut/. Acad. Sci. U.S.A. 80, 49 19-492 1. Gunthert, U., Schweiger, M., Stupp, M., and Doerfler, W. (1976). Proc. Natl. Acad. Sci. U.S.A. 73, 3923-3927. Harbers, K., Schnieke, A , , Stuhlmann, H., Jahner, D., and Jaenisch, R. (1981). Proc. Null. Acud. Sci. U.S.A. 78, 7609-7613. Herman, G. E., and Modrich, P. (1982). J . Biol. Chem. 257, 2605-2612. Hjelle, B. L., Phillips, J . A,, 111, and Seeburg, P. H. (1982). Nucleic Acids Res. 10, 3459-3474. Holliday, R., and Pugh, J. E. (1975). Science 187, 226-232. Igo-Kemenes, T., Horz, W., and Zachau, H. (1982). Annu. Rev. Biochem. 51, 89-121. Jahner, D., Stuhlmann, H., Stewart, C. L., Harbers, K., Loehler, J . , Simon, I., and Jaenisch. R. (1982). Nature (London) 298, 623-628. Jones, P. A,, and Taylor, S. M. (1980). Cell 20, 85-93. Jones, P. A,, Taylor, S. M., Mohandas, T., and Shapiro, L. J. (1982). Proc. Natl. Acad. Sci. U.S.A. 79, 1215-1219.
DNA METHYLATION IN EUKARYOTIC CELLS
183
Jones, R. E., DeFeo, D., and Piatigorsky, J. (1981). J . Biol. Chem. 256, 8172-8176. Kappler, J. W. (1970). J . Cell. Physiol. 75, 21-32. Kappler, J. W. (1971). J . Cell. Physiol. 78, 33-40. Kastan, M. B., Gowans, B. J., and Lieberman, M. W. (1982). Cell 30, 509-516. Keshet, E., and Cedar, H. (1983). Nucleic Acids Res. 11, 3571-3580. Kunnath, L., and Locker, J. (1982). Nucleic Acids Res. 10, 3877-3892. Kunnath, L., and Locker, J. (1983). EMBO J . 2, 317-324. Lark, C. (1968). J . Mol. Biol. 31, 389-400. Larsen, A,, and Weintraub, H. (1982). Cell 29, 609-622. Lester, S . C., Kom, N. J., and Demars, R. (1982). Somut. Cell Genet. 8, 265-284. Liskay, M. R., and Evans, R. J. (1980). Proc. Nutl. Acud. Sci. U.S.A. 77, 4895-4898. Mandel, J. L., and Chambop, P. (1979). Nucleic Acids Res. 7 , 2081-2090. Manes, C., and Menzel, P. (1981). Nature (London) 293, 589-590. Mann, B. M., and Smith, H. 0. (1977). Nucleic Acids Res. 4, 421 1-4221. McGhee, J. D., and Cinder, G . D. (1979). Nature (London) 280, 419-420. McKeon, C., Ohkubo, H., Pastan, I . , and de Crombrugghe, B. (1982). Cell 29, 203-210. Mermod, J. J., Bourgeois, S . , Defem, N., andCrepin, M. (1983). Proc. Nutl. Acud. Sci. U . S . A . 80, 110-1 14. Mohandes, T., Sparkes, R. S . , and Shapiro, L. J. (1981). Science 211, 393-396. Nakhasi, H . , Lynch, K. R., Dolan, K. P., Unterman, R. D., and Feigelson, P. (1981). Proc. Nail. Acad. Sci. U.S.A. 78, 834-837. Naveh-Many, T., and Cedar, H . (1981). Proc. Nutl. Acad. Sci. U.S.A. 78, 4246-4250. Nickol, J., Behe, M., and Felsenfeld, G. (1982). Proc. Nutl. Acud. Sci. U.S.A. 79, 1771-1775. Niwa, O., Yokota, Y., Ishida, H., and Sugahara, T. (1983). Cell 32, 1105-1113. Ott, M.-O., Sperling, L., Cassio, D., Levilliers, J., Sala-Trepat, J., and Weiss, M. C. (1982). Cell 30, 825-833. Pages, M., and Roizes, G. (1982). Nucleic Acids Res. 10, 565-576. Palmiter, R. D., Chen, H. Y., and Brinster, R. L. (1982). Cell 29, 701-710. Parker, M.-I., Judge, K., and Gevers, W. (1982). Nucleic Acids Res. 10, 5879-5891. Pellicer, A,, Wigler, M., Axel, R., and Silverstein, S. (1978). Cell 14, 133-141. Pollack, Y., Stein, R., Razin, A,, and Cedar, H. (1980). Proc. Nutl. Acud. Sci. U.S.A.77, 64636467. Pukkila, P., Peterson, J., Herman, G., Modrich, P., and Meselson, M. (1983). Genetics 104, 571582. Radman, M., van den Elsen, P., and Glickrnan, B. (1978). Muf. Gen. Genet. 163, 307-312. Razin, A., and Cedar, H. (1977). Proc. Nuti. Acud. Sci. U.S.A. 74, 2725-2728. Razin, A , , and Friedman, J. (1981). Prog. Nucleic AcidRes. Mol. Biol. 25, 33-52. Razin, A , , and Riggs, A . D. (1980). Science 210, 604-610. Razin, A , , Urieli, S . , Pollack, Y., Gruenbaum, Y., and Glaser, G. (1980). Nucleic Acids Res. 8, 1783- 1792. Razin, A,, Webb, C., Szyf, M., Yisraeli, J., Rosenthal, A., Naveh-Many, T., Sciaky-Gallili, N., and Cedar, H. (1983). Proc. Nutl. Acud. Sci. U.S.A., 81, 2275-2279. Reilly, J. G . , Thomas, C. A., Jr., and Lundell, M. J. (1982). DNA 1, 259-266. Riggs, A. D. (1975). Cytogenet. Cell Genet. 14, 9-1 1. Riggs, A. D., and Jones, P. (1983). Adv. Cancer Res. 40, 1-40. Roberts, R. J. (1976). CRC Crit. Rev. Biochem. 4, 123-164. Rogers, J., and Wall, R. (1981). Proc. Natl. Acud. Sci. U.S.A. 78, 7497-7501. Sanders-Haigh, L . , Blanchard-Owens, B., Hellewell, S . , and Ingram, V. M. (1982). Proc. Nutl. Acud. Sci. U.S.A. 79, 5332-5336. Schlagman, S . , Hattman, S . , May, M. S . , and Berger, L. (1976). J . Bacteriol. 126, 990-996.
184
AHARON RAZIN AND HOWARD CEDAR
Seidman, J. G . , and Leder, P. (1978). Nature (London) 276, 790-795. Shani, M., Admon, S . , Nudel, U., and Yaffe, D. (1983). Exp. Biol. Med. 9, 235-242. Shen, C. K. J., and Maniatis, T. (1980). Proc. Natl. Acad. Sci. U.S.A. 77, 6634-6638. Simon, D., Stuhlmann, H . , Jahner, D., Wagner, H . , Werner, E., and Jaenisch, R. (1983). Nature (London) 304, 275-277. Sims, M. A., Doering, J. L., and Hoyle, H. D. (1983). Nucleic Acids Res. 11, 277-290. Singer, J., Robert-Ems, J., and Riggs, A. D. (1979). Science 203, 1019-1023. Smith, H. 0. (1979). Science 205, 455-462. Solage, A,, and Cedar, H. (1978). Biochemistry 17, 2934-2938. Stein, R., Gruenbaum, Y., Pollack, Y., Razin, A , , and Cedar, H. (1982a). Proc. Natl. Acad. Sci. U.S.A. 79, 61-65. Stein, R., Razin, A., and Cedar, H. (1982b). Proc. Natl. Acud. Sci. U.S.A. 79, 3418-3422. Stein, R., Sciaky-Gallili, N., Razin, A,, and Cedar, H. (1983). Proc. Nut/. Acad. Sci. U.S.A. 80, 2422-2426. Stewart, C. L., Stuhlman, H . , Jahner, D . , and Jaenisch, R. (1982). Proc. Natl. Acud. Sci. U.S.A. 79,4098-4102. Stuhlmann, H., Jahner, D., and Jaenisch, R. (1981). Cell 26, 221-232. Sturm, K. S . , and Taylor, J. H. (1981). Nucleic Acids Res. 9, 4537-4546. Sutter, D., and Doerfler, W. (1980). Proc. Natl. Acad. Sci. U.S.A. 77, 253-256. Szyf, M., Gruenbaum, Y., Urieli-Shoval, S . , and Razin, A. (1982). Nucleic Acids Res. 10, 72477259. Szyf, M., Avraham-Haetzni, K., Reifman, A., Kaplan, F., Shlomai, J., Oppenheim, A , , and Razin, A. (1984). Proc. Natl. Acad. Sci. U.S.A., 81, 3278-3282. Tantravahi, U., Guntaka, R., Erlanger, B. F., and Miller, 0. J. (1981). Proc. Natl. Acud. Sci. U.S.A. 78, 489-493. Taylor, J. H. (1979). In “Molecular Genetics 111. Chromosome Structure” (J. H. Taylor, ed.), pp. 89-1 15. Academic Press, New York. Taylor, S. M., and Jones, P. A. (1979). Cell 17, 771-779. Taylor, S. M . , and Jones, P. A. (1982). J . Mol. Biol. 162, 679-692. Urieli-Shoval, S . , Gruenbaum, Y., Sedat, J . , and Razin, A. (1982). FEBS Len. 146, 148-152. Urieli-Shoval, S . , Gruenbaum, Y., and Razin, A. (1983). J . Bucteriol. 153, 274-280. van der Ploeg, L. H. T . , and Flavell, R. A. (1980). Cell 19, 947-958. Vardimon, L., Kressmann, A., Cedar, H., Maechler, M., and Doerfler, W. (1982). Proc. Natl. Acad. Sci. U . S . A . 79, 1073-1077. Vedel, M . , Gomez-Garcia, M . , Sala, M . , and Sala-Trepat, J . M. (1983). Nucleic Acids Res., 11, 4335-4354. Venolia, L., Gartler, S. M., Wassmann, E. R., Yen, P., Mohandas, T . , and Shapiro, L. J . (1982). Proc. Natl. Acad. Sci. U.S.A. 79, 2352-2354. Vovis, G . F., Horiuchi, K . , and Zinder, N. D. (1974). Proc. Natl. Acud. Sci. U.S.A. 71, 38103813. Waalwijk, C., and Flavell, R. A. (1978a). Nucleic Acids Res. 5, 3231-3236. Waalwijk, C., and Flavell, R. A. (1978b). Nucleic Acids Res. 5, 4631-4641. Weintraub, H. (1983). Cell 32, 1191-1203. Weintraub, H., and Groudine, M. (1976). Science 193, 848-856. Weintraub, H., Larsen, A , , and Groudine, M. (1981). Cell 24, 333-344. Weisbrod, S. (1982). Nature (London) 297, 289-295. Wigler, M., Sweet, R., Sim, G. K . , Wold, B., Pellicer, A , , Lacy, E., Maniatis, T., Silverstein, S . , and Axel, R. (1979). Cell 16, 777-785. Wigler, M., Levy, D . , and Perucho, M. (1981). Cell 24, 33-40.
DNA METHYLATION IN EUKARYOTIC CELLS
185
Wilks, A. F., Cozens, P. J., Mattaj, I. W., and Jost, J. P. (1982). Proc. Natl. Acad. Sci. U.S.A. 79, 4252-4255. Wilson, V. L., and Jones, P. A. (1983). Cell 32, 239-246. Woodcock, D. M., Adams, J. K . , and Cooper, I. A. (1982). Biochim. Biophys. Actu 696, 15-22. Woodcock, D. M., Adams, J. K . , Allan, R. G., and Cooper, 1. A. (1983). Nucleic Acids Res. 11, 489-499. Yagi, M., and Koshland, M. E. (1981). Proc. Null. Acad. Sci. U.S.A. 78, 4907-491 1.
This Page Intentionally Left Blank
Index
A Aging, evolution of, 105-107 Algae nitrate assimilation mutants of, 36-38 nitrate reductases of, 6 Arabidopsis thaliana, mutants, nitrate assimilation and, 35
B Barley, nitrate assimilation mutants of, 30-35
C Chromosome gene transfer and; 135 technical aspects of, 135, 137-138 host cell, transgenome stability and linkage to, 143-146 coated pits clustering in, 56-57 endocytosis from, vesicle formation in, 5966 structure in: clathrin, 58-59 C-shaped vesicles, endocytic vesicles and, 70
methylation and gene expression cause and effect, 170-172 correlation between, 166- 170 differentiation, 176-178 mechanisms of blocking gene activity, 172- 176 methylation pattern in eukaryotes, 160-163 clonal inheritance of Dattern, 163 methylases, 163-165 methylation at replication fork, 165-166 studies of recombination and ligation in mammalian cells, 152- 154 Differentiation, DNA methylation and, 176178
E Endocytic membrane traffic, functions of hormonal delivery, 83-84 other functions, 84-85 pathologic processes, 81-83 recycling of receptors and ligands, degradation, 79-80 traffic control of synthesized materials, 8081 Endocytic vesicles, characteristics of C-shaped vesicles, 70 lysosomes and prelysosomes, 69-70 macro-and micropinocytosis, 70-72 multivesicular bodies, 67-69 receptosomes, 66-67 tubular elements, 70 Endocytosis Golgi and, 72-76 kinetic classes of, 74 vesicle formation from coated pits and; 5966
D Darura, molybdenum cofactor mutants of, 28-
30 Deoxyribonucleic acid gene transfer and, 134 technical aspects of, 135- 137 methylation demethylation models, 179-180 main concepts, 180- 181
187
188
INDEX
Endoplasmic reticulum, materials from, Golgi and, 78 Eukaryotes DNA methylation pattern in, 160-163 clonal inheritance of pattern, 163 methylases, 163-165 methylation at replication fork, 165- 166 evidence for stability of translation process in fungi, 121-123 mammalian cells, 123-126 Exocytosis, Golgi and, 76-77 Exocytic membrane traffic functions of hormonal delivery, 83-84 other functions, 84-85 pathologic processes, 81-83 recycling of receptors and ligands, degradation, 79-80 traffic control of synthesized materials, 80-81
F Fungi genetic control of nitrogen assimilation in autogenous regulation, 18- 19 effect of carbon metabolism on, 25-26 evidence for turnover of nitrate reductase, 25 genetic analysis, 12- 18 nitrogen metabolism repression of assimilation, 19-23 possible translational control of, 23-25 nitrate reductases of, 5-6 stability of translation process in, I2 1 - 123 transformation systems, nitrate assimilation and, 42-43
G Gene(s), eukaryotic, isolation of, 149- 150 Gene activity, mechanisms of blocking of, 172-176 Gene expression DNA methylation and cause and effect, 170-172 correlation between, 166- 170 differentiation, 176- 178 mechanism of blocking of gene activity, 172- I76
eukaryotic, studies of regulation of, 150152 Gene mapping, intrachromosomal, 147- 149 Genetic materials, uptake and expression of donor materials in recipient cells cotransfer of nonselected genes, 146- 147 transgenome size and gene copy number, 139-143 transgenome stability and linkage to host cell chromosomes, 143-146 Gene transfer applications in genetic analysis of mammalian cells intrachromosomal gene mapping, 147149 isolation of eukaryotic genes, 149-150 studies of DNA recombination and ligation, 152-154 studies of regulation of eukaryotic gene expression, 150-152 choice of recipient cells in, 138-139 chromosome-mediated, 135 technical aspects of, 135, 137-138 DNA-mediated, 134 technical aspects of, 135- 137 Golgi, compartmentalization in endocytosis, 72-76 exocytosis: materials destined for cell surface, 76-77 lysosomal delivery, 77-78 materials from endoplasmic reticulum, 78
H Hormones, delivery of, 83-84
L Ligands, recycling of, 79-80 Lysosomal delivery, Golgi and, 77-78 Lysosomes, endocytic vesicles and, 69-70
M Macropinocytosis, endocytic vesicles and, 7072 Mammalian cells genetic analysis, applications of gene transfer in, 147-154 stability of translation process in, 123- 126 Messages, turnover of, 118-1 19
189
INDEX Methylases, eukaryotic, 163- 165 Micropinocytosis, neocytic vesicles and, 7072 Molybdenum cofactor, mutants and, 28-30 Molybdenum cofactor gene, from E. coli, cloning of, 4 1-42 Multivesicular bodies, endocytic vesicles and, 67-69
N Nicotinna, molybdenum cofactor mutants in, 28-30 Nicofiana plumbaginijolia, nitrate reductase mutants of, 27-28 Nicotiana robacurn. nitrate reductase mutants of, 27-28 Nitrate assimilation biochemical aspects of, 2-3 algal nitrate reductases, 6 assimilatory nitrate reductases, 7-9 biochemical regulation of, 9-1 1 fungal nitrate reductases, 5-6 general features of assimilatory nitrate reductases, 3-5 higher plant nitrate *cductases, 6-7 in v i m complemenistion studies on nitrate reductase, 11-12 cell and molecular biological advances in cloning nitrate assimilation genes-concluding remarks, 43-44 cloning of a molybdenum cofactor gene from E . coli, 41-42 fungal transformation systems, 42-43 plant transformation, 43 somatic hybridization studies, 39-41 genetic regulation in fungi autogenous regulation, 18- 19 effect of carbon metabolism on, 25-26 evidence for turnover of nitrate reductase, L3
genetic analysis, 12-18 nitrogen metabolism repression of assimilation, 19-23 possible translational control of, 23-25 genetics in plants conclusions, 38-39 genetic analysis, 26-38 Nitrate reductase, evidence for turnover of, 25 Nitrogen metabolism, repression of nitrate assimilaton in fungi, 19-23
P Pea, nitrate assimilation mutants of, 30-35 Physcomitrella patens, chlorate-resistant mutants of, 36 Plants genetics of nitrate assimilation in, 26-39 higher, nitrate reductase of, 6-7 transformation, nitrate assimilation and, 43 Plasma membrane, events at clustering in coated pits, 56-57 receptors in plasma membrane, 53-55 structure of coated pits: clathrin, 58-59 vesicle formation in endocytosis from coated pits, 59-66 Prelysosomes, endocytic vesicles and, 69-70 Prokaryotes, stability of translation process in, 119-121 Protein(s), turnover of, 118-1 19 Protein synthesis errors and, 1 1 1-1 15 models of error propagation and, 95-100
R Receptors in plasma membranes, 53-55 recycling of, 79-80 Receptosomes, endocytic vesicles and, 66-67 Replication fork, methylation at, 165- 166
S Soybean, nitrate reductase mutants of, 35-36 Synthesized materials, traffic control of, 8081 T
Transcription, errors and, 115-1 18 Translation process evidence for stability of from eukaryotes, 121-126 from prokaryotes, 119- 121 evolutionary considerations evolution of aging, 105-107 origins of translation, 103 selection of accuracy, 103- 105 measurement of error rates protein errors, 11 1-1 15 transcription errors, 1 15- 118
190
INDEX
nature of proof or disproof, 107- I I 1 models of error propagation generalized theory, 100- 103 protein synthesis, 95- 100 turnover of proteins and messages, I 18-1 19
Transport, mutants, nitrate assimilation and, 38 Tubular elements, endocytic vesicles and, 70
Contents of Recent Volumes Volume 70
Volume 72
Cycling Noncycling Cell Transitions in Tissue Aging, Immunological Surveillance, Transformation, and Tumor GrowthSEYMOUR GELFANT The Differentiated State of Normal and Malignant Cells or How to Define a “Normal” Cell in CUltUre-MINA J. BISSELL On the Nature of Oncogenic Transformation of C e l l d E R A L D L. CHAN Morphological and Biochemical Aspects of Adhesiveness and Dissociation of Cancer C e l l s AND YASUJI ISHIMARU HIDEOHAYASHI The Cells of the Gastric MUCOSa-HERBERT F. HELANDER Ultrastructure and Biology of Female Gametophyte in Flowering Plants-R. N. KAPILAND A. K. BHATNAGAR
Microtubule-Membrane Interactions in Cilia and Flagella-WtLLIAM L. DENTLER The Chloroplast Endoplasmic Reticulum: Structure, Function, and Evolutionary Significance-SARAH P. GIBBS R. LEHMANNAND P. DNA Repair-A. KARRAN Insulin Binding and Glucose Transport-RusSELL HILF, LAURIE K. SORGE,AND ROGER J. GAY Cell Interactions and the Control of Development in Myxobacteria POpUkitiOnS-DAVID WHITE Ultrastructure, Chemistry, and Function of the Bacterial Wall-T. J. BEVERIDCE INDEX
INDEX
Volume 73 Volume 71 Protoplasts of Eukaryotic Algae-MARTHA D. BERLINER Integration of Oncogenic Viruses in Mammalian Polytene Chromosomes of Plants-WALTER CellS
INDEX
191
192
CONTENTS OF RECENT VOLUMES
Volume 74
Organization and Expression of Viral Genes in Adenovirus-Transformed Cells-S. J. FLINT The Plasma Membrane as a Regulatory Site in Highly Repeated Sequences in Mammalian Genomes-MAXINE F. SINGER Growth and Differentiation of Neuroblastoma Ceh-sIEGFRIED w . DE LAAT Moderately Repetitive DNA in EvolutionROBERTA. BOUCHARD AND PAULT. V A N DER SAAG Mechanisms That Regulate the Structural and Structural Attributes of Membranous Organelles in BacteriaXHARLts C. REMSEN Functional Architecture of Cell SurfacesD. BERLIN Separated Anterior Pituitary Cells and Their ReJANETM. OLIVERA N D RICHARD sponse to Hypophysiotropic HormonesGenome Activity and Gene Expression in Avian A N D MARIA Erythroid CellS-KARLEN G . GASARYAN CARLDENEF,L u c SWENNEN, ANDRIES Morphological and Cytological Aspects of Algal Calcification-MICHAEL A. BOROWITZKAWhat Is the Role of Naturally Produced Electric Current in Vertebrate Regeneration and HealNaturally Occurring Neuron Death and Its Reging?-RICHARD B . BVRGENS ulation by Developing Neural PathwaysTIMVTHY J. CUNNINGHAM Metabolism of Ethylene by Plants-JOHN A. HALL DODDSA N D MICHAEL The Brown Fat Cell-JAN NEDERGAARD AND INDEX OLOV LINDBERG INDEX
Volume 75
Volume 77
Mitochondria1 Nuclei-TsuNEYosHI KUROIWA Calcium-Binding Proteins and the Molecular Slime Mold LectinS-JAMES R. BARTLES, Basis of Calcium ACtiofl-LINDA J. VANELWILLIAM A. FRAZIER,A N D STEVEN D. D I K , JOSEPH G. Z t N E D E G U I , DANIELR. MARSHAK, A N D D. MARTIN WATTERSON RVSEN Lectin-Resistant Cell Surface Variants of Eu- Genetic Predisposition to Cancer in Man: In Vitro Studies-LEVY KOPELOVICH karyotic Cells-EvE BARAKBRILES Cell Division: Key to Cellular Morphogenesis in Membrane Flow via the Golgi Apparatus of Higher Plant Cells-DAvrD G . ROBINSON the Fission Yeast, SchizosuccharomycesA N D UDO KRISTEN BYRVNF. JOHNSON,CODEB. CALLEJA, BONGY. Yoo, MICHAELZUKER,A N D IAN Cell Membranes in Sponges-WERNtR E. G . J. MCDONALD MULLER Microinjection of Fluorescently Labeled Pro- Plant Movements in the Space EnvironmentDAVIDG . HEATHCOTt teins into Living Cells, with Emphasis on E. KRHS Chloroplasts and Chloroplast DNA of AcefubuCytoskeletal PrOteinS-THOMAS laria rnedirerraneu: Facts and HypothesesA N D WALTERBIRCHMEIER ANGELALUTKE A N D SILVANO BONOTTV Evolutionary Aspects of Cell DifferentiationStructure and Cytochemistry of the Chemical R. A. FLICKINGER A N D WLADSynapSeS-STEPHEN MANALOV Structure and Function of Postovulatory FolliI M I R OVTSCHAROFP cles (Corpora Lutea) in the Ovaries of Nonmammalian VeItebrateS-SRINlVAS K. SAI- INDEX DAPUR INDEX
Volume 78 Volume 76 Cytological Hybridization to Mammalian ChromOSOmeS-ANN s. HENDERSVN
Bioenergetics and Kinetics of Microtubule and Actin Filament Assembly-DisassemblyTERRELLL. HILLAND MARCW. KIRSCHNER
CONTENTS OF RECENT VOLUMES Regulation of the Cell Cycle by SomatomedinS-HOWARD ROTHSTEIN Epidermal Growth Factor: Mechanisms of AcliOn-MANJUSRI DAS Recent Progress in the Structure, Origin, Composition, and Function of Cortical Granules in Animal Egg-SARDUL S. GURAYA
193
Immunofluorescence Studies on Plant Cells<. E. JEFFREE,M. M. YEOMAN,A N D D. C. KILPATRICK Biological Interactions Taking Place at a LiquidSolid hterfaCe-ALEXANDRE ROTHEN INDEX
INDEX
Volume 81 Volume 79 Oxidation of Carbon Monoxide by BacteriaYOUNGM. KIMA N D GEORGED. HEGEMAN The Formation, Structure, and Composition of the Mammalian Kinetochore and Ki- Sensory Transduction in Bacterial ChemotaxI S a E R A L D L. HAZELBAUER A N D SHIGEAKI netochore FiberXoNLu L. RIEDER HARAYAMA Motility during FertilizationaERALD SCHATThe Functional Significance of Leader and TrailTEN er Sequences in Eukaryotic mRNAs-F. E. Functional Organization in the NucleusRONALDHANCOCKA N D TENI BOULIKAS BARALLE C h r o m o s o m e ~ R A N T R. The Relation of Programmed Cell Death to De- The Fragile SUTHERLAND velopment and Reproduction: Comparative Psoriasis versus Cancer: Adaptive versus Studies and an Attempt at ClassificationIatrogenic Human Proliferative DiseasesJACQUES BEAULATON AND RICHARDA. SEYMOUR GELFANT LOCKSHIN Cryofixation: A Tool in Biological Ultrastruc- Cell Junctions in the Seminiferous Tubule and the Excurrent Duct of the Testes: Freezeturd Research-HELMUT PLATTNERA N D Fracture Studies-Toskilo NAGANOAND LUISBACHMANN FUME SUZUKI Stress Protein Formation: Gene Expression and Environmental Interaction with Evolutionary Geometrical Models for Cells in Tissues-HISSignificance-C. ADAMSAND R. W. R I N N ~ AO HONDA Growth of Cultured Cells Using Collagen as INDEX Substrate-JASON YANC AND s. NANDI
x
INDEX
Volume 80 DNA Replication Fork Movement Rates in Mammalian Cells-LEON N. KAPPA N D ROBERT B. PAINTER Interaction of Virsues with Cell Surface ReceptOrS-MARC TARDIEU,ROCHELLEL. EPSTEIN, A N D HOWARD L. WEINER The Molecular Basis of Crown Gall InductionW. P. ROBERTS The Molecular Cytology of Wheat-Rye Hybrids-R. APPELS Bioenergetic and Ultrastructural Changes Associated with Chloroplast Development-A. R. WELLBURN The Biosynthesis of Microbodies (Peroxisomes, G1yoxysomes)-H. KINDL
Volume 82 The Exon:Intron Structure of Some Mitochondrial Genes and Its Relation to Mitochondria1 EVohtiOn-HENRY R. MAHLER Marine Food-Borne Dinoflagellate ToxinsDANIELG. BADEN Ultrastructure of the Dinoflagellate AmphieSma-LENITA c. MORRILLA N D ALFREDR. LOEBLICH I11 The Structure and Function of Annulate Lamellae: Porous Cytoplasmic and Intranuclear Membranes-RICHARD G . KESSEL Morphological Diversity among Members of the
194 Gastrointestinal SAVAGE INDEX
Volume 83
CONTENTS OF RECENT VOLUMES Microflora-DWAYNE
c.
Cell Surface Receptors: Physical Chemistry and Cellular Regulation-DoucLAs LAUFFENBURGER AND CHARLES DELIS Kinetics of Inhibition of Transport Systems-R. M. KRUPKAAND R. DEVES INDEX
Transposable Elements in Yeast-VALERIE Mo- Volume 85 ROZ WILLIAMSON Techniques to Study Metabolic Changes at the Receptors for Insulin and CCK in the Acinar Cellular and Organ LeVd-ROBERT R. DEPancreas: Relationship to Hormone ActionFURIAAND MARYK. DYGERT IRA D. GOLDFINEAND JOHNA. WILLIAMS Mitochondria1 Form and Function Relationhips The Involvement of the Intracellular Redox State in Vivo: Their Potential in Toxicology and and pH in the Metabolic Control of StimPathology-ROBERT A. SMITH AND MURIEL ulus-Response COUpling-zYGMUND ROTH, J. ORD NAOMICHAYEN, A N D S H A B T A Y DIKSTEIN Heterogeneity and Territorial Organization of Regulation of DNA Synthesis in Cultured Rat the Nuclear Matrix and Related StructureHepatoma Cek-ROELAND V A N WllK M. BOUTEILLE,D. BOUVIER,AND A. P. Somatic Cell Genetics and Gene Mapping-FASEVE TENKAO Changes in Membrane Properties Associated Tubulin Isotypes and the Multigene Tubulin with Cellular Aging-A. MACIEIRA-COELHO Families-N. J . COWANA N D L. DUDLEY Retinal Pigment Epithelium: Proliferation and The Ultrastmcture of Plastids in ROOtS-JEAN Differentiation during Development and ReM. WHATLEY generation-OLcA G. STROEVAAND VICThe Confined Function Model of the Golgi TOR I. MITASHOV Complex. Center for Ordered Processing of INDEX Biosynthetic Products of the Rough Endoplasmic R e t i c u l u m - A ~M. ~ ~ TARTAKOFIProblems in Water Relations of Plants and Cells-PAUL J. KRAMER Volume 84 Phagocyte-Pathogenic Microbe InteractionsANTOINETTERYTER A N D CHANTALDE Controls to Plastid Division-J. V. POSINGHAM CASTELLIER AND M. E. LAWRENCE Morphology of Transcription at Cellular and INDEX Molecular hVeIS-FRANCINE PUVIONDUTILLEUL An Assessment of the Evidence for the Role of Volume 86 Ribonucleoprotein Particles in the Maturation Toward a Dynamic Helical Model for the Influof Eukaryote mRNA-I. T. KNOWLER ence of Microtubules on Wall Patterns in Degradative Plasmids-.I. M. PEMBERTON P l a n t s 4 ~ 1 W. v ~ LLOYD Regulation of Microtubule and Actin Filament Assembly-Disassembly by Associated Small Cellular Organization for SteroidogenesisPETERF. HALL and Large Molecule-TERRELL L. HILL Cellular Clocks and Oscillators-R. R. AND MARCW. KIRSCHNER KLEVECZ,S. A. KAUWMAN,AND R. M. Long-term Effects of Perinatal Exposure to Sex SHYMKO Steroids and Diethylstilbestrol on the ReMaturation and Fertilization in Starfish productive System of Male MammalsOOCYteS-LAURENT MEIJER A N D PIERRE YASUMASA ARAI,TAKAOMORI,YOSHIHIDE GUERRIER SUZUKI,AND HOWARDA. BERN
CONTENTS OF RECENT VOLUMES
195
Development Of the Cotton Fibe-AMARJIT s. BASRAAND C. P. MALIK The Neuronal Organization of the Outer Plex- Cytochemistry of Fat AbSorptiOtI-YUKARI iform Layer of the Primate Retina-ANDREW AND TOSHKMI MIZUNUMA TAKAHASHI P. MARIANI Electrical Activation of Arterial MuscleINDEX DAVIDR. HARDERAND ALANWATERS Aging of Cells in Culture-JAMES R. SMITH AND D. W. LINCOLN,I1 Chemotactic Factors Associated with Leukocyte Volume 87 Emigration in Immune Tissue Injury: Their Separation, Characterization, and Functional The Modeling Approach-BARBARA E. Specificity-HIDE0 HAYASHI, MITSUOHONWRIGHT AND MITSUOMI DA, YASUOSHIMOKAWA, Protein Diffusion in Cell Membranes: Some BiHIRASHIMA ological ~mp~ications-MlcHAEL McNeural Organization and Cellular Mechanisms CLOSKEYAND MU-MINGPo0 of Circadian Pacemakers-JON W. JACKLET ATPases in Mitotic Spindles-M. M. PRATT Cytobiology of the Ovulation-Neurohormone Nucleolar S t r u c t u r e 4 u u GOESSENS Producing Neuroendocrine Caudo-Dorsal Membrane Heterogeneity in the Mammalian Cells of Lymnaea stagnalis-E. W. ROUBOS Spermatozoon-W. V. HOLT INDEX Capping and the Cytoskeleton-Liuu Y. W. GERARD J. BOURG~JIGNONA N D BOURGLJIGNON The Muscle Satellite Cell: A Review-DENNIS Volume 90 R. CAMPION Cytology of the Secretion in Mammalian Sweat KUROSUMI, S U S U M U Electron Microscopic Study of Retrograde AxGlands-KnzuMAsA onal Transport of Horseradish PeroxidaseSHIBASAKI, AND TOSHIO ITO ERZSEBETFEHER INDEX DNA Sequence Amplification in Mammalian Celk-JoYcE L. HAMLIN, JEFFREY D. MILBRANDT,NICHOLASH. HEINTZ, AND Volume 88 JANEC. AZIZKHAN Computer Applications in Cell and NeuLysosornal Functions in Cellular Activation: robiology: A Review-R. RANNEYMIZE Propagation of the Actions of Hormones and Effect of Microtubule Inhibitors on Invasion and Other Effectors-CLARA M. SZEGO AND on Related Activities of Tumor CellS-MARC RICHARD J. PKETRAS M. MAREELAND MARCDE METS Neuronal Secretory SyStemS-MONA CASTEL, Membranes in the Mitotic Apparatus: Their H.-DIETER HAROLD GAINER, AND Structure and FUnCtiOn-hTER K. HEPLER DELLMANN AND STEPHEN M.WOLNIAK INDEX Pollen-Pistil Recognition: New Concepts from Electron Microscopy and Cytochemistry-C. DUMAS,R. B. KNOX,AND T. GAUDE Volume 89 Surface Topography of Suspended Tissue Cells-Yu. A. ROVENSKYAND Ju. M. Histochemistry of Zinc and C0pper-Em.R VASILIEV AND ETER KASA SZERDAHELYI Gastrointestinal Stem Cells and Their Role in Histochemistry of Adenylate Cyclase-G. AND Yu. D. Carcinogenesis-A. I. BYKOREZ AND POEGGEL,H. LUPPA,H.-G. BERNSTEIN, IVASHCHENKO J. WEISS INDEX Cell Biology of Trypanosoma Cruzi-WANDERLEY DE SOUZA
196
CONTENTS OF RECENT VOLUMES
Volume 91 Supramolecular Cytology of Coated VesiCkS-RICHARD E. FINE AND COLIN D. OCKLEFORD Structure and Function of the Crustacean Larval Salt Gland-FRANK P. CONTE
Pathways of Endocytosis in Thyroid Follicle CdIS-VOLKER HERZOG Transport of Proteins into MitochondriaSHAWN D O O N A N , ERSILIAMARRA, SALVATORE PASSARELLA, CECILIASACCONE, AND ERNESTO QUAGLIARIELLO INDEX
