International Review of Cytology: A Survey of Cell Biology, Volume 90

INTERNATIONAL REVIEW OF CYTOLOGY VOLUME90 ADVISORY EDITORS H. W. BEAMS DONALD G. MURPHY HOWARD A. BERN ROBERT G . E...

Author: Geoffrey H. Bourne

10 downloads 848 Views 16MB Size Report

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

DOWNLOAD PDF

INTERNATIONAL

REVIEW OF CYTOLOGY VOLUME90

ADVISORY EDITORS H. W. BEAMS DONALD G. MURPHY HOWARD A. BERN ROBERT G . E. MURRAY GARY G. BORISY RICHARD NOVICK PIET BORST ANDREAS OKSCHE BHARAT B. CHATTOO MURIEL J . ORD STANLEY COHEN VLADIMIR R. PANTIC RENE COUTEAUX W. J. PEACOCK MARIE A. DlBERARDlNO DARRYL C. REANNEY CHARLES J . FLICKlNGER LIONEL I . REBHUN JEAN-PAUL REVEL OLUF GAMBORG M. NELLY GOLARZ DE BOURNE JOAN SMITH-SONNEBORN WILFRED STEIN YUKIO HlRAMOTO YUKINORI HlROTA HEWSON SWlFT K. KUROSUMI K. TANAKA GIUSEPPE M ILLON IG DENNIS L. TAYLOR ARNOLD MITTELMAN TADASHI UTAKOJI AUDREY MUGGLETON-HARRIS ROY WIDDUS ALEXANDER YUDlN

INTERNATIONAL

Review of Cytology EDITED BY

G. H. BOURNE

J. F. DANIELLI

S t . George's University School oj Medicine St. George's, Greriadu West Itidies

Danielli Associates Worcester-, Massachusetts

ASSISTANT EDITOR K. W. JEON Department of Zoology University of Tennessee Knosville. Teniwmee

VOLUME90

1984

ACADEMIC PRESS, INC. (Harcourr Brace Jovonovich, 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 IN ANY FORM OR BY ANY MEANS, ELECTRONIC OR MECHANICAL, INCLUDING PHOTOCOPY, RECORDING, OR ANY INFORMATION STORAGE AND RETRIEVAL SYSTEM, WITHOUT PERMISSION IN WRlTlNO FROM THE PUBLISHER.

ACADEMIC PRESS,INC.

Orlando, Florida 3 2 8 8 7

United Kingdom Edition published by ACADEMIC PRESS, INC. ( L O N D O N ) LTD. 24/28 Oval Road, London N W I 7DX

LIBRARY OF CONGRESS CATALOG CARD NUMBER: 5 2 - 5 2 03

I S B N 0-12-364490-9 PRINTED IN THE UNITED STATES OF AMERICA 84 85 86 87

9 8 7 6 5 4 3 2 1

Contents

CONTRIBUTORS . . . . . . . . .

ix

Electron Microscopic Study of Retrograde Axonal Transport of Horseradish Peroxidase E R Z S L B ~ FEHER T 1.

I1 . 111.

IV . V.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Visualization of Horseradish Peroxidase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Morphology and Histology of the Reaction Product . . . . . . . . . . . . . . . . . . . . . . Uptake of Horseradish Peroxidase into the Nerve Terminals . . . . . . . . . . . . . . . Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

i 3 5

21 25 25

DNA Sequence Amplification in Mammalian Cells JoyCt

.

L . HAMLIN JEFFREYD . MII.BRANDT, NICHOLASH . HEINTZ. A N D JANL. c . AZIZKHAN

I. I1 .

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Occurrence of Amplification Phenomena . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111. Cytological Manifestations of Gcne Amplification . . . . . . . . . . . . . . . . . . . . . . . IV . Nature of Amplified Sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V . Agents That Increase the Frequency of Amplification . . . . . . . . . . . . . . . . . . . . VI . Proposed Mechanisms of Sequence Amplification . . . . . . . . . . . . . . . . . . . . . . . VII . Concluding Remark:, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

31

33

45 52 64 67 75 77

Computer Applications in Cell and Neurobiology: A Review R. RANNLYM i a 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II . The Microcomputer in the Research Laboretory . . . . . . . . . . . . . . . . . . . . . . Ill . Computer Systems for Microscope Control and Plotting . . . . . . . . . . . . . . . V

83 84 90

vi

CONTENTS

I V. V. VI . V11. V111.

IX . X.

93 98 103 107 Ill 117 117 119

Serial Section Reconstruction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Computer-Aided Morphonietric Measurement . . . . .................... Video Image Processing and Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Computer Uses in Photometry and Fluorescence Microscopy . . . . . . . . . . . . . . Computer-Automated Autoradiography and Immunocytochemistry . . . . . . . . . . Other Cell Biology Computer Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Effect of Microtubule Inhibitors on Invasion and on Related Activities of Tumor Cells MAHCM . M h H t u

I. I1 . Ill . I V. V.

VI . VII . VIII .

AND

MARC DE ME-IS

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Biochemistry of Microtubule AssemblylDisassembly . . . . . . . . . . . . . . . . . . . . . Analysis of Microtubules inside Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Antiinvasiveness of Microtubule Inhibitors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Antiproliferative and Cytotoxic Effect of Microtubule Inhibitors . . . . . . . . . . . . Directional Migration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Effect of Microtubule Inhibitors o n Plasma Membrane Functions . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

125 126 134

139 144 148 161 161

162

Membranes in the Mitotic Apparatus: Their Structure and Function

I. I1 . 111. I V. V.

VI . VII . VI11 .

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Early Studies o n Mitotic Membranes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ER in the MA of Higher Organisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Membranes in the MA of Lower Organisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . Golgi and Other Membranes in the MA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Membrane Function: Regulation of [ C a 2 + ] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Membrane Function: A Component in Chromosome Transport . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ....................

169

170 173 198 205 209 224 230 231

Pollen-Pistil Recognition: New Concepts from Electron Microscopy and Cytochemistry C . DUMAS.R . B . KNOX. A N D T . GAUIX

I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I1. The Mature Viable Pollen Grain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

239 241

CONTENTS

vii

I11. The Receptive Pistil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV . Male-Female Interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Surface Topography of Suspended Tissue Cells Y u . A . ROVENSKY A N D Ju . M . VASILIEV Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Morphology of Surface Microextensions of Suspended Cells . . . . . . . . . . . . . . . Surface Topography of Suspended Tissue Cells of Various Types . . . Mechanisms of Formation of Microextensions . . . . . . . . . . . . . . . . . . . . . . . . . . Effects of Previous Contacts with the Substrate on the Surface Topography of Suspended Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.

I1 . 111. IV . V.

273 274 285 290 299 303 304

Gastrointestinal Stem Cells and Their Role in Carcinogenesis A . I . BYKOREZA N D Yu . D . IVASHCHENKO

I. II . 111. IV .

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Stem Cells of the Small Intestine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Stem Cells of the Colon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Stem Cells of the Gastric Glands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V . Regulation of Proliferation and Differentiation in the Gastrointestinal Epithelium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI . Stem Cells in Carcinogenesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VII . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . INDEX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . CONTENTS OF PREVIOUS VOLUMES. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

309 311 318 323 332 344 363 364 375 379

This P a ge Intentionally Left Blank

Contributors

Numbers in parentheses indicate the pages on which the authors' contributions begin.

JANE C. AZIZKHAN(3 l), Department of Biochemistry, University of Virginia School of Medicine, Charlottesville, Virginia 22908 A. I. BYKOREZ (309), Department of Chemical Carcinogenesis, Kavetsky Institute for Oncology Problems, Academy of Science of the Ukrainian SSR, Kiev 252127, USSR MARCDE METS( 125), Laboratory of Experimental Cancerology , Department of Radiotherapy and Nuclear Medicine, University Hospital, B-9000 Ghent, Belgium C. DUMAS (239), Dkpartement de Biologie Vkgktale et C.M.E.A.B.G., Universite' Claude Bernard-Lyon I , Villeurbanne 69622 Cedex, France ERZSEBET FEHER( 1 ), First Institute of Anatomy, Semmelweis University Medical School, Budapest, Hungary T. GAUDE (239), De'partement de Biologie Vkgktale et C.M.E.A.B.G., Universite' Claude Bernard-Lyon I , Villeurbanne 69622 Cedex, France JOYCE L. HAMLIN(3 1), Department of Biochemistry, University of Virginia School of Medicine, Charlottesville, Virginia 22908 NICHOLAS H . H E I N T Z(3 ~ l ) , Department of Biochemistry, University of Virginia School of Medicine, Charlottesville, Virginia 22908 'Present address: Department of Biology, The Johns Hopkins University, Baltimore, Maryland 21218. 2Present address: Department of Pathology, University of Vermont College of Medicine, Burlington, Vermont 05401. ix

CONTRIBUTORS

X

PETERK. HEPLER( 169), Department of Botany, University of Massachusetts, Amherst, Massachusetts 01003 Yu. D. IVASHCHENKO (309), Department of Chemical Carcinogenesis, Kavetsky Institute for Oncology Problems, Academy of Science of the Ukrainian SSR, Kiev 252127, USSR R. B. KNOX (239), School of Botany, University of Melbourne, Parkville, Victoria 3052, Australia MARC M. MAREEL (125), Laboratory of Experimental Cancerology, Department of Radiotherapy and Nuclear Medicine, University Hospital, B-9000 Ghent, Belgium JEFFREYD. MILBRANDT3 (3I), Department of Biochemistry, University of Virginia School of Medicine, Charlottesville, Virginia 22908 R. RANNEYMIZE (83), Department of Anatomy and Division of Neuroscience, University of Tennessee Center for the Health Sciences, Memphis, Tennessee 38163 Yu. A . ROVENSKY(273), Laboratory of Mechanisms of Carcinogenesis, Cancer Research Center of the USSR Academy of Medical Sciences, Moscow, USSR Ju. M . VASILIEV(273), Laboratory of Mechanisms of Carcinogeriesis, Cancer Research Center of the USSR Academy of Medical Sciences, Moscow, USSR STEPHEN M. WOLNIAK (169), Department of Botany, University of Maryland, College Park, Muryland 20742

7Present address: Division of Laboratory Medicine. Washington University School of Medicinc. St. Louis, Missouri 631 10.

INTERNATIONAL

REVIEW OF CYTOLOGY VOLUME90


INTERNATIONAL REVIEW OF CYTOLOGY. VOL 90

Electron Microscopic Study of Retrograde Axonal Transport of Horseradish Peroxidase ERZSEBET FEHER First Institute

of Anatomy, Semmelweis

University Medictil School, Budapest, Hungary

I.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Visualization of Horseradish Peroxidase . . . . . . . . . . . . . . . . . . . . . . . Morphology and Histology of the Reaction Product . . . . . . . . . . . . . . A . Localization of Horseradish Peroxidase in the Nerve Cell Bodies.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B . Localization of Horseradish Peroxidase within Nerve Processes 1V. Uptake of Horseradish Peroxidase into the Nerve Terminals . . . . . . . V . Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. 111.

I 3 5 5 17

21 25 25

I. Introduction One of the earliest reports on retrograde transport of materials within axons, i.e., toward the cell body, was the in vitro observation of Matsumoto (1920), who followed the rapid movement of vesicles stained with neutral red within outgrowing sympathetic fibers. Experiments carried out by Kerkut et al. (1967) and by Watson ( 1968) indicated that radioactive labeled materials injected into muscles appeared in the perikarya of nerve cells innervating them. The retrograde axonal transport of horseradish peroxidase (HRP), which was originally demonstrated in peripheral motorneurons by Kristensson and Olsson (1971) and by Kristensson et ul. (197 I ) , has arroused great interest among neuroanatomists. It was shown that HRP is transported intraaxonally from the terminal region of an axon retrogradely to the parent cell body (Kristensson et al., 1971; LaVail and LaVail, 1972; Hanson, 1973; Jones and Leavitt, 1973; Kristensson and Olsson, 1973a,b; Ralston and Sharp, 1973; Warr, 1973; Graybiel and Devor, 1974; Kuypers et al., 1974; Nauta et a/., 1974; Ito ef a / ., 1981; Carlson and Mesulani, 1982a). This technique has now become widely used as an experimental tool for demonstrating neuronal connections both in the central and in the peripheral nervous systems (i.e., Kristensson, 1975; LaVail, 1975; Cullheim and Kellerth, 1976; Kitai et ul., 1976; Adams, 1977; Hedreen and McGrath, 1977; Hunt et d., 1977; Luiten and van der Pers, 1977; Keefer, 1978; Malmgren and Olsson, 1978; Kalia and Davies, 1978; Vanegas et al., 1978; Satomi et a l . , 1979; Contreras et I Copyright 0 1Y84 by Academic Prew. Inc All rights of rcpraluclion In any form rexrvcd ISBN 0-12-364490-9

2

ERZSEBET FEHER

a/., 1980; Panneton and Loewy, 1980; Arvidsson and Gobel, 1981; Kuo et

a/. ,

1981; Nicholson and Severin, 1981; Nomura and Mizuno, 1981, 1982; Ross et a/., 1981; Stuesse, 1982). A major advantage of the use of the enzyme HRP is elucidating the connections of the nervous system, that neuronal cell somata are labeled in a way which enables the determination of the cells inducing a particular fiber pathway (LaVail et al., 1973; Ralston and Sharp, 1973; Sherlock et a l ., 1975; Price and Fisher, 1978). For example, LaVail and LaVail (1974) observed retrograde transport of HRP from the region of retinal ganglion cell bodies after injection of HRP into the chick optic tectum. The enzyme HRP has been shown in electron microscopic studies to flow in an orthograde as well as in a retrograde direction (Hanson, 1973; Lynch et al., 1973, 1974; Reperant, 1975; Winfield e t a l . , 1975; Jones and Hartman, 1978). Some studies have demonstrated that orthograde transport of HRP can be used to reveal the central distribution of afferent fibers of peripheral nerves both at the light microscopic (Reperant, 1975; Scalia and Colman, 1974; Luiten, 1975; Gwyn et a / ., 1979; Mesulam and Brushart, 1979; Kalia and Mesulam, 1980) and the electron microscopic level (Muller and McMahan, 1976; Proshansky and Egger, 1977; Rastad et ul., 1977; Bettie et a/., 1978; Gobel and Falls, 1979; Ohara and Lieberman, I98 I ) . The cells projecting to regions injected with HRP are identifiable with histochemical procedures (Graham and Karnovsky, 1966) due to the accumulation of vesicular packets of reaction product, which presumably represent the pinocytotic incorporation of the protein at axon terminals (LaVail and LaVail, 1972; LaVail, 1975; Hedreen and McGrath, 1977). The distribution of reaction product helps to clarify under the light microscope the type of neuron, on the basis of its shape, size (Wilson and Groves, 1981), and dendritic and axonal arborization pattern. In a subsequent step the synaptic connections of this identified neuron can be examined by electron microscopes (Jankowska et a/., 1976; Muller and McMahan, 1976; Cullheini et a / . , 1977; Ralston et a/., 1978, 1980; Rethelyi et a / . , 1979; Robson and Mason, 1979; Langerback et a / . , 1981). The reaction product of the retrograde transport of HRP is usually filling the soma and proximal dendrites of the neuron; this means, first, that the type of neuron-that is labeled-can be determined, and second, that boutons undergoing anterograde degeneration following lesion of an afferent neuron can be traced to the soma and proximal dendrite. HRP is particularly useful for demonstrating the arbors of axons and should be applicable to the study of the intrinsic organization of any region with well defined afferent connections. The intracellularly applied HRP is an invaluable marker for tracing neuronal projections, to resolve the detailed morphology of individual neurons, and for marking cells in synaptological analysis with the electron microscope

EM STUDY OF HORSERADISH PEROXIDASE

3

(Cowan and Cuenod, 1975; Winer, 1977; Kristensson, 1975; LaVail, 1978; Eckert and Boshek, 1980; Elekes and Szabo, 1982).

11. Visualization of Horseradish Peroxidase

The method for visualizing the activity of HRP at the electron microscopic level was introduced by Graham and Karnovsky (1966) who have used it to study membrane recycling and to follow the path of the retrograde transport from the synaptic region. HRP being a protein of molecular weight of about 40,000 will not pass across cell membranes unless an invagination of the membranes does not occur. Visualization of the reaction product of the enzyme inside the terminal should therefore imply a membrane event of this kind. Membrane infoldings supposedly resulting from the release of transmitters have been described by several groups of workers (Holtzman et al., 197 1; Ceccarelly et ul., 1973). The Sigma Type VI HRP used contains mainly the basic isoenzyme. According to Giorgi and Zahnd (1978) it is only this isoenzyme that is taken up and transported retrogradely at detectable levels by undamaged nerve cells (Bunt et al., 1976; Bunt and Haschke, 1978; Malmgren rt al., 1978). The sections of materials used were processed to demonstrate the presence of HRP using tetramethylbenzidine (TMB) or 3,3’-diaminobenzidine (DAB) according to the method of Mesulam (1978) and to Graham and Karnovsky (1966), respectively. The distribution of reaction product was much greater in the TMB incubated tissue than in the DAB incubated tissue under the light microscope. This is consistent with the previous observations on the greater sensitivity of the TMB method (Mesulam and Rosene, 1979; Dietrichs et al., 1981; Carlson and Mesulam, 1982a). In the dorsal horn according to Carlson and Mesulam (1982b) DAB reaction product was localized within membrane-bound bodies located in synaptic terminals. These labeled bodies were generally larger than synaptic vesicles and some were elongated rather than circular in profile (Figs. 1 and 2). In contrast to the DAB reaction product, the crystalloid TMB reaction product was not confined to membrane-bound organelles and frequently filled significant portions of the entire synaptic terminal. It has been reported by several authors (Beattie et al., 1978; Gobel and Falls, 1979) that application of HRP to the proximal ends of dorsal roots and subsequent ultrastructural examination of DAB reaction product in the spinal cord showed labeling on the cytoplasmic side of the axolemma and on the external surface of synaptic vesicles and mitochondria. They concluded that this suggests such labeling occurs mostly by diffusion within the cytoplasm rather than by membrane-bound transport. According to Somogyi et al. (1979) a highly electron-dense reaction product


5

was formed when 3,3'-diaminobenzidine was used as substrate at pH 7.4. Only slightly electron dense, but of a characteristic appearance reaction product formed when 0.02% o-tolidine is used at the same pH. The reaction from otolidine at pH 7.4 is found in membrane-limited particles, including multivesicular bodies. The cobalt-glucose oxidase method is also used for HRP reaction by Itoh et al. ( 1979) and by Nakamura el ul. ( 198 1). In our observations to obtain information on the ultrastructural localization of HRP, materials were processed by the method of Somogyi et a / . (1979). In each cat 0.3-0.04 (1.1 at 20% solution of HRP (Sigma Type VI) in 0.05 M phosphate buffer was injected into the mesenteric nerves between the layers of the mesothelium under semisterile conditions over a period of 13-20 minutes. After 2 or 3 days survival the animals were perfused through the left ventricle with I % glutaraldehyde and I % paraformaldehyde in 0. I M phosphate buffer (pH 7.4) according to Benedeczky and Somogyi (1975). Small pieces of the intestine (the middle part of the intestine innervated by the injected nerves) were excised and then cut by a Vibratome in 30-km sections. The slices were washed for several hours in phosphate buffer and placed for 30 minutes in a medium containing 0.05% 3,3'-diaminobenzidine and 0.03% hydrogen peroxide in 0.1 M phosphate buffer for 1 hour. The slices were then postfixed in osmium acid, dehydrated, arid embedded in Araldite. Ultrathin sections were mounted on single-hole grids, contrasted with uranylacetate and lead citrate. At the control examination-processed in a similar wayy-of these sections no labeled cells and processes were found either on light or on electron microscopy.

111. Morphology and Histology of the Reaction Product

A. LOCALIZATION OF HORSERADISH PEROXIDASE I N THE NERVECELL BODIES The usefulness of HRP as a neuronal marker at the electron microscopic level has already been demonstrated via the use of intracellularly applied HRP by several authors (Cullheim and Kellerth, 1976; Jankowska et id., 1976; Snow et al., 1976; Rastad, 1978; Rastad et al., 1977; RCthelyi et a l . , 1982). In the labeled neurons large (300-700 nm in diameter), highly electron-dense profiles, identified earlier as residual bodies or secondary lysosomes (Broadwell et al.,

FIG. I . Labeled nerve processes in the myenteric plexus. Bar scale= 1 pm. x42,OOO. FIG.2. Arrows show the labeled membrane-bound bodies in the nerve terminal. Bar scale= I bm. X72.000.

6

ERZSEBET FEHER

1980) were found. Multivesicular bodies were also common and had a variable morphology; a portion of their limiting membrane was often coated on its cytoplasmic surface. Electron-lucent vesicles (40-80 nm in diameter), HRPlabeled vesicles with or without an external coat, and vacuoles (100-300 nm in diameter) of various shape were apparent in all the preparations, usually accumulated closely to the Golgi zones but also at other cytoplasmic sites (Fig. 3). The peroxidase reaction product eventually filled many of the lysosomal residual bodies in the perikarya (Colman et al., 1976; Takeuchi e f al., 1982). In vesicles, smooth endoplasmic reticulum and membrane-limited granules, the end product fills the space right up to the limiting membrane; in contrast, dense-core vesicles which are not labeled and occur in all neurons have a granular matrix, usually of higher electron density than the HRP reaction product, and there is a translucent zone between the matrix and the limiting membrane (Figs. 4 and 5 ) . Lipofuscin pigment is normally found in ganglion cells and appears to increase significantly with age. However, an accumulation of pigment that may be misinterpreted as HRP vesicles could be ruled out since these animals were young and the control materials showed the absence of these pigments. The distribution and cytological features of the labeled neurons were carefully examined and compared with those of the unlabeled neurons. The labeled neurons were seen in both the myenteric and the submucosal plexuses. They were medium-sized (30-50 p,m) and spindle-shaped, multipolar, triangular, or oval (Fig. 6). These data are similar to those obtained by light microscopy (FehCr and Vajda, 1982b). The shape and distribution of the labeled neurons resembled the medium-size cells stained by silver impregnation (FehCr and Vajda, 1972). According to Dogie1 (1895) and Type I1 nerve cells in the wall of the intestine once were believed to be sensory in nature. Later, Kadanoff and Spassowa (1959) described the sensory function of the bipolar and unipolar neurons in the gut. Kuntz (1922) traced nerve fibers from the submucosal plexus into the mucous membrane and suggested that some of the fibers were likely to originate from afferent neurons in the submucous ganglia. It has also been proved with degeneration methods (Schofield, 1960, 1968; Fehtr and Vajda, 1974) that some of the enteric neurons project centripetally along mesenteric neurovascular bundles. The combined anatomical and physiological studies by Bulbring et al. (1958) proved the presence of afferent neurons that innervate the mucous membrane. It has also been shown that with regard to ultrastructural features the small intestine contains different types of nerve cells (Fehtr and Csinyi, 1974; Cook and Burnstock, 1976). Physiological studies have also demonstrated that the intrinsic nerve plexus of the small intestine is composed of at least three types of neurons (Milton and Smith, 1956; Wood, 1975; Furness and Costa, 1980). The labeled neurons have oval nuclei, contain the usual cytoplasmic orga-


7

FIG. 3 . Labeled nerve cell in the submucous plexus. Arrows point to the HRP-labeled veticles close to thc Golgi zones. Bar scale= I pm. XS4.000.


9

FIG. 6. A medium-size oval-shaped neuron. Peroxidase is evident in a variety of sizes of vesicles and tubules of the neuron soma. Bar scale= I pm. X 18,000.

FIG. 4. Cytoplasm of the labeled nerve cell. Arrows show the dense-core vesicles. Bar scale= 1 pm. x42.000. FIG.5 . Cytoplasm of the labeled nerve cell in the myenteric plexus. Note the abundant densecore vesicles (arrows) occurring in the cytoplasm. Bar scale= I pm. X30.000.


I1

nelles, such as mitochondria, Golgi apparatus, smooth and rough endoplasmic reticulum, lysosomes, polysomes, and multivesicular bodies (Figs. 7 and 8). The retrogradely transported HRP has very high electron density and is distributed in the perikaryon. This observation was in good accordance with that found at light microscopic levels. The granules varied in size from 10 to 100 nm and many of the larger ones showed definite evidence of being, by fusion of smaller individual granules, of varying density. In an earlier study of neurons of the central nervous system, LaVail and LaVail (1 974) attempted to classify the organelles involved in the retrograde axonal transport of HRP. In the nerve cell bodies the labeled organelles were characterized as ( 1) large, approximately 100 nm vesicles, (2) multivesicular bodies, (3) cup-shaped organelles, and (4) tubules of agranular reticulum. Most of the HRP-filled vesicles were ovoid or somewhat elongated in shape and were bound by a single smooth membrane (Figs. 9 and 10) (Brownson et al., 1977), but the spherical 30-50 nm synaptic vesicles were not labeled. HRPcontaining vesicles and lysosomes were found throughout the cytoplasm of the ganglion cells. However, there is a noticeable tendency for HRP vesicles to occupy a perinuclear position (Kristensson and Olsson, 1971; LaVail et al., 1973; Sotelo and Riche, 1974; Ellison and Clark, 1975). The cross-sectional diameter of the individual vesicles ranged from 65 nm to 0.50 km, with many about 0.3 p m (Fig. 11). In many instances the HRP-positive structures were concentrated near the inner aspect of the Golgi complexes. The Golgi sacs and vacuoles themselves were usually free of HRP product, but some of them contained a small amount of reaction product (Brownson et a/., 1977). Broadwell and Brightman (1979) described the uptake of HRP by hypothalamic neurons under osmotic stress and its subsequent orthograde transport to the neurohypophysis without passage through the Golgi complex. The size of the vesicles containing HRP in neuronal somata increased with time. According to LaVail and LaVail (1974) the HRP was found within organelles in ganglion cell bodies of the retina contralateral to the tectal injection and the most of these labeled vesicles were larger, i.e., about 0.5 p m with some almost 1 p m in diameter by 24 hours. Most dense bodies were clearly membrane bound (Hanson, 1973; LaVail and LaVail, 1974; RepCrant, 1975; Weldon, 1975; Schwab, 1977) and within the larger ones there were often relatively clear vacuoles of varying size ( Al-Khafai el a / . , 198 1) (Fig. 12). N o diffuse HRP product was found in any ganglion cells. FIG. 7. Vesicles containing HRP accumulate near the Golgi complex. Cup-shaped organelles are showed by the arrow. Bar scale= I p.m. X42.000. FIG. 8. Large multivesicular bodies (arrows) are in the labeled neurons. Bar acale= I pin. x30,000.

FIG.9. Most of the HRP-filled vesicles are ovoid and irregular shape and are bound by a single smooth membrane (arrows). Bar scale= 1 pm. X54,OOO. FIG. 10. Irregularly shaped HRP positive structures are seen to be membrane bound (arrow). Bar scale= 1 pm. X42.000. 12

FIG. I I . The cross-sectional diameter of the individual vesicles ranged from 0.3 to 0.5 pm. Bar scale= 1 pin. X96.000. FIG. 12. Most of the multivesicular bodies contain relatively clear vacuoles of varying size (arrows), Bar scale= I Km. X54.000. 13

14

ERZSEBET FEHER

In several electron micrographs it was noted that typical HRP vesicles had fibrillary bridges between the smooth endoplasmic reticulum and the vesicles (Fig. 13). Processes originating from the labeled neurons also contained HRP granules (Holstege and Dekker, 1979; Robson and Mason, 1979). These granules were larger (300 to 500 nm) and disseminated throughout the cytoplasm (Fig. 14). Occasionally, granular vesicles, 80 to 120 nm in diameter, were present in the HRP-labeled neurons. Moreover, the labeled cells contained abundant rough endoplasmic reticulum, which was oriented parallel to periphery of the perikarya. The morphological features of the labeled neurons were apparently different from those of unlabeled neurons in the small intestine. The HRP-labeled neurons were covered with nerve processes forming synapses with the soma (Figs. 15 and 16). The fact that cell somata containing electron-dense granules of the reaction product were of one single type of cell, and most of the unlabeled cells showed different morphological features, yields further evidence for the identification of the sensory nature of some neurons in the small intestine. Recent studies by means of retrograde axonal transport of HRP have demonstrated that neurons in the wall of the small intestine project toward the celiac ganglion in the cat (Feher and Vajda, 1982b). It appears that the peristaltic reflex is mediated by an intrinsic reflex arc, in which the afferent neurons were labeled by HRP. Synapses on their surfaces suggest that these neurons collect information from other local, possibly interneuronal nerve cell processes and thus influence prevertebral ganglion cells. It is also possible that the intrinsic afferent neurons converge and establish synapses on the HRP-labeled neurons conveying the information to the prevertebral ganglion. Hence, some of the labeled neurons may be considered as interneurons. If our assumption is correct, such labeled neurons with their synapses might, in fact, be units of integration. The ultrastructural features of the labeled neurons are similar to that described as Type I by FehCr and Csinyi (1974) and some of these cells were seen to degenerate after capsaicin treatment (FehCr and Vajda, I982a). The peroxidase reaction product eventually filled many of the lysosomal residual bodies in the perikarya. The small, HRP-containing vesicles enter the soma from the axon and are supposed to coalesce to form large structures, since most of the vesicles in the soma are larger than those in the axons, and coalescing profiles are frequently seen near the Golgi region (Sellinger and Petiet, 1973). Nauta et a / . (1975) observed that the HRP-filled tubular profiles sometimes FIG. 13. At the arrow the HRP-containing multivesicular body is continouos with smoothsurfaced endoplasmic reticulum. Bar scale= 1 pm. X96.000. FIG. 14. Labeled nerve process originating from the cell of the subrnucosal plexus. Bar scale= 1 pm. ~18,000.


17

appeared to be branched and not infrequently appeared to be in continuity with HRP-filled lysosomes. Several investigators have described the early incorporation and segregation of extracellular markers within cup-shaped organelles and multivesicular bodies (Brightman, 1965; Birks et d . , 1972; Bunge, 1973; Holtzman et al., 1973). Additional internal vesicles and entire multivesicular bodies may be formed in the cell soma near the Golgi apparatus (Hirsch et al., 1968; Friend, 1969). Holtzman et al. (1967) have stressed the relationship between the agranular reticulum and lysosomal system of neurons. We have also noted that many of the lysosomes containing retrogradely transported HRP are associated with the agranular reticulum, and in some cases these lysosomes appear to have an associated tail containing HRP. Thus, the appearance of HRP in the agranular reticulum and lysosomes is consistent with the comparmentalization, transport, and degradation of any exogenous protein taken up by the neuron. The HRP disappears from the cell in 3-4 days and many of the residual lysosomes in these nerve cells are more electron dense than those in the control nerve cells. The structures with more loosely packed HRP reaction product may represent various stages of degradation of HRP. HRP can also reach the neuron’s soma and dendrites as a diffuse label, however, and both diffuse and agranular labeling may coexist in the same cell cluster or even in the same neuron (Adams and Warr, 1976). In the vicinity of the injection site of HRP into the central nervous system, many of the oligodendrocytes, superficial glial cells contained accumulated HRP diffusely within their cytoplasm, particularly after longer intervals (Krishnan and Singer, 1973; LaVail and LaVail, 1974).

B . LOCALIZATION OF HORSERADISH PEROXIDASE WITHIN NERVEPROCESSES The intraaxonal retrograde transport of exogenous protein in the nervous system is an established phenomenon. Multivesicular bodies are regularly present in presynaptic terminals and axons. It is noted that vesicles contain not only dense granules but also a variety of membranous elements, including multivesicular substructures. When endocytosed tracers such as HRP are present, the multivesicular bodies become labeled among the synaptic vesicles (Villegos, and Fernandez, 1966; Holtzman et a/., 1971 ; Teichberg et a/., 1975). According to Theodosis ( 1982) up to 4 hours after the tracer injections, large vacuoles and cup-shaped figures FIG. 15. Small clear vesicles containing nerve terminal (arrow) synapse with the soma. Bar scale= I pm. X72.000. FIG. 16. The HRP-labeled neurons are covered with the nerve process forming synapse with the soma (arrow). Bar scale= I pm. X30.000.

18

ERZSEBET FEHER

(from 100 to 300 nm in diameter) are predominant in intraaxonal organelles labeled with HRP. Both in longitudinally and cross-sectioned axons, the HRPlabeled vesicles were often found interspersed between neurotubules and neurofilaments, centrally located in the axoplasm. Associated with these organelles were smaller vesicles and tubules, with or without reaction product. Profiles of intraaxonal HRP vesicles were identified in several configurations. The fine structure of all profiles of HRP vesicles had in common a single outer limiting membrane of smooth endoplasmic reticulum. The dense core type of HRP vesicles found in the cytoplasm was also observed in axons. However, high resolution micrographs of intraaxonal HRP vesicles revealed a variety of subunits in vesicles. Other investigators (Winfield et al., 1975; Colman et al., 1976; Mizuno et al., 1978) found that the HRP product appeared as small dense membranebound bodies. HRP-labeled organelles were present in both axons and terminals (Figs. 17 and 18). Multivesicular bodies, or vacuoles enclosing smaller vesicles, were also labeled by the enzyme its reaction product usually filling their matrix but not the dense-cored synaptic vesicles. Nauta et al. (1975) observed that HRP was localized in tubular profiles and vesicles of varying size but they were always clearly much larger than 25 nm microtubules and no HRP could be found in association with the 25 nm microtubules. It is also of interest to note that the lysosome-like HRP vesicles observed in axons had frequently varying amounts of subcapsular clear or lucid areas and an occasional contact with neurotubules as previously reported by LaVail et al. ( 1973). Labeled multivesicular bodies were more frequent in preparations fixed 8 hours after the peroxidase injections, as were vacuoles completely filled with reaction product. In addition to those labeled organelles, numerous vacuoles, cup-shaped profiles, and multivesicular bodies, devoid of tracer, were apparent in the axonal cytoplasm at all survival periods studied by Theodosis (1982). The labeling of multivesicular bodies in terminals is enhanced by conditions that promote active transmission by the terminals and thus produce increased uptake of tracers into the synaptic vesicles (Teichberg et al., 1975; Schacher et a/., 1976). The amount of endocytized peroxidase that eventually undergoes retrograde transport is markedly increased in preparations whose synaptic vesicles have become labeled through synaptic activity (Teichberg et a / . , 1975). However, according to Gobel and Falls (1979) the HRP reaction product binds to the neurofilaments, neurotubules, and the cytoplasmic surface of the axolemma in the primary axons of the substantia gelationosa of Rolando. Although the terminal is densly labeled, the size and shape of the synaptic vesicles, and the two synaptic contacts remain clearly visible (Rastad, 1981). FIG. 17. Labeled terminal in the submucosal plexus. Bar scale= I pm. X96,OOO. FIG. 18. Multivesicular bodies, vacuoles are present in the labeled nerve processes in the myenteric plexus (arrows). Bar scale= 1 pm. X 18,000.

20

ERZSEBET FEHER

According to Beattie et al. (1978) the reaction product can be observed adjacent to mitochondria1 membranes and appears to surround synaptic vesicles. However, according to Nauta e f al. (1975) and Robson and Mason (1979) no HRP was found within the synaptic vesicles. It is frequently claimed that the peroxidase reaction product is found within synaptic vesicles but a close inspection of the micrographs serving as the basis for such claims suggests that it is on the external surface, and the cores of the vesicles are electron lucid (Ceccarelli et af., 1973; Heuser and Reese, 1973; Ripps et af., 1976). In axons, reaction product can be seen adjacent to the plasma membrane and aggregated in the cytoplasm, occasionally over microtubules (Holstege and Dekker, 1979). Egger et al. (1981) using the intraaxonal injection of HRP were able to stain the functionally identified afferent fibers in the cat spinal cord. The labeled terminals proved to be predominantly axodendritic asymmetric synapses containing round, clear vesicles. Multiple synapses on a single dendrite were also observed, at a 900 nm distance from each other. However, when a bouton was making synaptic contact with an HRP-labeled dendrite, postsynaptic densities were not clearly distinguished (Langerback et al., 1981). The cerebellar-olivary axon terminals were detected by Mizuno et al. (1980) anterogradely with HRP injected into the lateral cerebellar nucleus. In the principal olive contralateral to the HRP injection, electron-dense HRP granules were found in axon terminals contacting dendritic profiles. In these HRP-labeled axon terminals the synaptic vesicles were spherical. Morphometric analysis in the supraoptic nucleus of the rat was made by Theodosis (1982) who found that the estimated mean volume density of the peroxidase-containing profiles was relatively small and tended to increase only slightly up to 8 hours. On the other hand, the proportion of these organelles was much higher in the nuclei of dehydrated animals. At 4 hours after administration of HRP, their volume density was twice as that of controls, and at 8 hours it had increased further to over three times the control value. Kistler and Schwartz (1982) used HRP-conjugated to the lectin, wheat germ agglutinin, which binds with high affinity to cell surfaces, and examined the retrograde transport in a single invertebrate neuron. There was apparently more reaction product in the cells after the transsection than after the ligature of the axon. By electron microscopy, there was a marked difference between cells with transsected and ligated axons. While the cells with ligated axons contained no labeled organelles, some organelles in the cells with cut axons contained reaction product. It is noted that the increased HRP uptake by injured neurons resulted in a heavy homogeneous staining of neuron processes (Kristensson and Olsson, 1974). Labeled synaptic vesicles seem not to participate in large numbers in the retrograde transport (Kristensson et al., 1971; LaVail and LaVail, 1974), but


21

multivesicular bodies and other structures that may be precursors or contributors to multivesicular bodies or other lysosomes (elongated sacs, tubules) are prominent among the bodies, that carry endocytozed tracers in a retrograde direction (LaVail and LaVail, 1974; Teichberg et al., 1975). Ligature experiments originally indicated that HRP is transported proximaldistally. According to Gwyn et al. (1982) in the labeled axon the electron-dense reaction product was associated with the microtubules and, in the axoplasm, it was found between the microtubules. The appearance of the HRP reaction product described in vagal terminals by Gwyn et al. (1982) resembled the somewhat dispersed appearance of HRP labeling reported in dorsal root terminals by Beattie er al. (1978) and in mamillothalamic terminals by Holstege and Dekker (1979). In contrast to these findings a report of labeling of cerebello-olivary terminals (Mizuno et al., 1980) showed the HRP reaction product as a small number of electron-dense aggregation. Some HRP-labeled terminals showed degenerating features like shrinkage, glia reactions, etc. (Figs. 19 and 20) (Dekker and Kuypers, 1976). A number of regulatory reactions such as chromatolysis after axon injury (Cragg, 1970), retrograde transsynaptic changes (Cowan, 1970), glial reactions (Sjostrand, 1965), and growth regulation (Prestige, 1970) also exist, especially in the central nervous system. No evidence has been provided for either the extrusion of HRP into the extracellular space, or for its transsynaptic transport of HRP in the small intestine. It has been demonstrated by De Olmos and Heimer, (1977) and Mesulam and Brushart (1979) that neurons labeled retrogradely through one of their long axon collaterals and their other long axon collateral contain also HRP. It is even more likely that short local axon collaterals will become labeled after retrograde transport of HRP along the main axon.

IV. Uptake of Horseradish Peroxidase into the Nerve Terminals Findings considered are pertinent to questions concerning the localization, intracellular movement, and degradation of exogenous protein molecules taken up by neurons from the extracellular space. However, numerous reports suggest that HRP can be taken up from the extracellular space by pinocytosis along the cell surface (Teichberg and Bloom, 1976). The uptake of HRP into isolated nerve terminals (synaptosomes) has been studied by Marchbarks ( 1982), using a spectrophotometric method to determine the enzyme activity. The uptake was not affected by metabolic poisons, while it was reduced at lower temperatures and was not associated with any significant release of cytoplasmic lactate dehydrogenase suggesting an endocytotic mechanism.


23

Chan et ul. (198 I ) found that in the in vitro uptake phase (neurit terminal region), as many vacuoles as vesicles and tubules were labeled. In the transport phase (along neurit), labeled vacuoles were predominant. It appeared that these neurons utilized mainly vacuoles as a means to transport endocytozed protein to soma1 lysosomes, reminiscent of cultured neuroblastoma cells (Chan et a/., 1980). It is widely accepted that neuronal perikarya synthesize most of the proteins and many of the other substances required for the maintenance and function of their axons. In the case of local uptake by the cell body, HRP is found in pinocytotic vesicles and ultimately in lysosomes and multivesicular bodies (Becker et al., 1968; Holtzman, 1971; Holtzman et ul., 1967; Holtzman and Peterson, 1969; Nagasawa et al., 197 I ; Sellinger and Petiet, 1973). In the case of local uptake by axon terminals HRP has been found in coated vesicles, cisternae of the agranular reticulum, and synaptic vesicles (Brightman, 1968; Zacks and Saito, 1969; Brightman et ul., 1970; Nagasawa et ul., 197 1 ; Ceccarelli et ul., 1973; Heuser and Reese, 1973). In the nerve terminals the peroxidase-containing sacs and vesicles can be distinguished from most of the large dense-cored vesicles since the latter usually have a translucent zone just inside the membrane (Somogyi et al., 1979). The incorporation of the tracer into the small vesicles is thought by Pysh and Wiley (1974) to be formed at least in part from the loaded plasma membrane. Multivesicular bodies have been implicated in the sequestration of endocytotically derived membrane in numerous systems, including neurons, where they were found to increase with increasing exposure time to the tracer (Theodosis, 1982). It is not unlikely that a first step in the sequestration of the endocytotically derived membrane occurs in multivesicular bodies, either within the terminals, after transformation from the vacuolar and cup-shaped figures, or within the perikarya (Holtzman et a/. , 1977). Since multivesicular bodies can move retrogradely in axon cytoplasm (LaVail e t a / ., 1980; Tsukita and Ishikawa, 1980) they could also have served to transport the endocytotically derived membrane to the perikarya. Ceccarelli et al. (1973) believe the vesicles membrane remain discrete during the process of exocytosis and is withdrawn immediately afterward, so that the vesicle is reformed with a complement of HRP that entered from the extracellular fluid during the release of transmitter. This also appears to be the view of Zimmerman and Denston (1977). Several laboratories have obtained results consistent with the possible involvement of the smooth endoplasmic reticulum in retrograde axonal transport (Sotelo FIG. 19. Arrow shows the degenerated labeled nerve terminal in the myenteric plexus. Bar scale= I pm. X42.000. FIG. 20. Degenerated labeled nerve process in the submucosal plexus. Bar scale= I pm. X 42,000.

24

ERZSEBET FEHER

and Riche, 1974; Nauta et al., 1975; Repirant, 1975; Price and Fisher, 1978). Once taken up, HRP seems to be transported retrogradely along the axon inside vesicles or smooth endoplasmic reticulum (Turner and Harris, 1974; Nauta et al., 1975; Teichberg et al., 1975). Results from other in vitro studies using single or serial sections by Birks ef al. (1972), Wessels et al. (1974), Weldon (1973, and Bunge ( 1977) have consistently pointed to the fact that tracer containing tubules in the neurons may be morphologically distinguished from the agranular reticulum. Sotelo and Riche (1974) identified tubules filled with HRP that extended for several microns within the axon of the pars reticulata of the substantia nigra after neostrial injections of HRP. Uptake of HRP by intact neurons (especially at their terminals) takes place normally by endocytosis (Becker et al., 1968; Bunt, 1969; Zacks and Saito, 1969; Nagasawa et a / ., 197 1; Heuser and Reese, 1973; Turner and Harris, 1973; Teichberg et al., 1975) or pinocytosis in regions removed from the synaptic complex as were observed in a number of other neuronal systems (Krishnan and Singer, 1973; AlKhagai et al., 1980). LaVail and LaVail (1974) calculated a 9.2 mm/day rate of movement of the HRP vesicles. The amount of HRP transported in retrograde direction may be even greater than in anterograde direction. This estimate would indicate that the terminal region of only a fraction of HRP retinal cells had access to the HRP (LaVail and LaVail, 1974). Brownson et al. (1977) found that by 48 hours following injection the ciliary processes contained large numbers of HRPpositive vesicles in the nonmyelinated axons. However, by 6 days HRP-labeled axons could not be found in the ciliary processes. The anterograde movement of the HRP label from the postganglionic neuron perikarya in the superior cervical ganglion to terminals in the ciliary body may be as rapid as 4 hours and as prolonged as 48 hours, according to Brownson et al. (1977). In the peripheral nervous system axoplasmic components are transported at different rates (Litchy , 1973; Iversen et al., 1975; Jacobson et al., 1975; McEwan and Grafstein, 1968; Sjostrand, 1970). The observations by Brownson et al. (1977) reported that the retrograde flow rate of HRP in the sympathetic nervous system is 500-700 mm/day. Holtzman et al. (1973) have suggested that multivesicular bodies may be involved in the turnover of cell surface materials as well as in the degradation of secretion granule contents and membranes related to the Golgi apparatus and other structures. In view of the work of Droz et af. (1975), it seemed plausible that HRPlabeled elongated tubular structures are involved in retrograde axonal transport. LaVail et al. (1980) observed further that there might be a portion of the smooth endoplasmic reticulum a part of a continous channel. LaVail and LaVail (1974) suggested that microtubules may be involved in the


25

mechanism of retrograde transport. Further support for the involvement of microtubules comes from the findings that low doses of vinblastine block retrograde transport of HRP in retinal ganglion cells in adult rats (Bunt and Lund, 1974). Microtubules have also been associated with the rapid, anterograde axonal transport of proteins (Schmitt, 1968; Samson, 1971; Ochs, 1972) and of synaptic vesicles (Smith, 1971; Droz et af., 1973). In summary, the enzyme appeared to be removed or inactivated from the injection site within 4-5 days of injection, and began to disappear from ganglion cells by the third day after injection. This agrees with the finding of Kristensson and Olsson (1973) and LaVail and LaVail (1974).

V. Concluding Remarks A major advantage of the use of the enzyme HRP in elucidating the connections of the nervous system is that neuronal cell somata are labeled in a way which enables the determination of the cells of origin of a particular fiber pathway. The reaction product can be studied under the electron microscope, to clarify the type of neuron, on the basis of the shape, size, and ultrastructural features of the soma, and its dendritic and axonal morphology. In a further step, the synaptic connections with other unlabeled nerve terminals can be studied. The cytoplasm of the labeled nerve cells was identified as round to irregular shaped vesicles bound by a single membrane. The content of the vesicles is dense and homogeneous. The HRP-labeled vesicles appeared in large number around the nucleus than elsewhere in the cytoplasm. Labeled neurons were also seen in the myenteric and in the submucosal plexuses. The granules varied in size from 10 to 100 nm and many of the larger ones showed definite evidence of being by the fusion of smaller individual granules, of varying density. In several electron micrographs it was noted that typical dense core type HRP vesicles showed fibrillary bridges extending between the smooth endoplasmic reticulum and the vesicle membrane. Processes originating from the labeled neurons also contained HRP granules. The fact that the cell somata containing electron-dense granules of the reaction product were of one type of cell, and most of those which were unlabeled showed different morphological features, yields further evidence for the identification of the neurons in the peripheral and the central nervous systems.

REFERENCES A d a m , J . C. (1977). Neuroscience 2, 141-145. A d a m , J. C . , and Warr, W . B . (1976). J . Comp. Neurol. 170, 107-122.

26

ERZSEBET FEHER

AIKhafai, F. A. H., Anderson, P. N., Mitchell, J . , and Mayor, D. (1980). J . Anat. 130, 883-889. AIKhafai, F. A. H., Anderson, P. N., and Mitchell, J . (1981). J . Neurocytol. 10, 353-362. Arvidsson, J . , and Gobel, S . (1981). Brain Res. 210, 1-16. Beattie, M. S., Bresnahan, J . C., and King, J . S . (1978). Brain Res. 153, 127-134. Becker, N. H., Hirano, A,, and Zimmerman, H. M. (1978). J . Neuroputhol. Exp. Neurol. 27,439452. Benedeczky, J . , and Somogyi, P. (1975). Cell Tissue Res. 162, 541-550. Birks, R. I . , Mackey, M. C., and Weldon, P. R. (1972). J . Neurocytol. I , 31 1-340. Brightman, M. W. (1965). Am. J . Anat. 117, 193-200. Brightman, M. W. (1968). In “Brain Barrier Systems, Progress in Brain Research” (A. Lajtha and D. H. Ford, eds.), pp. 19-37. Elsevier, Amsterdam. Brightman, M. W., Klatzo, I . , Olsson, Y., and Reese, T. S . (1970). J . Neurol. Sci. 10, 215-239. Broadwell, R . D., and Brightman, M. W. (1979). J . Comp. Neurol. 185, 31-74. Broadwell, R. D., Oliver, C., and Brightman, M. W. (1980). J . Comp. Neurol. 190, 519-532. Brownson, R. H., Uusitalo, R., and Palkama, A. (1977). Brain Res. 120, 407-422. Bunge, M. B. (1973). Anut. Rec. 175, 280-281. Bunge, M. B. (1977). J . Neurocytol. 6, 407-439. Bunt, A. H. (1969). J . Ultrustruct. Res. 28, 41 1-421. Bunt, A. H., and Haschke, R. H. (1978). J . Neurocytol. 7 , 665-678. Bunt, A. H., and Lund, R. D. (1974). Anat. Rec. 178, 507-508. Bunt, A. H . , Haschke, R. H., Lund, R. D., and Calkins, D. F. (1976). Brain Res. 102, 152-155. Bulbring, E., Lin, R. C. Y., and Schofield, G. (1958). Q.J . Exp. Physiol. 43, 26-37. Carlson, K . A,, and Mesulam, M. M. (1982a). I n “Tracing Neural Connections with Horseradish Peroxidase. IBRO Handbook. Series: Methods in Neurosciences” (M. M. Mesulam, ed.), pp. 153-184. Wiley, New York. Carlson, K. A,, and Mesulam, M. M. (1982b). Neurosci. Lett. 29, 201-206. Ceccarelli, B., Hurlbut, W . P., and Mauro, A. (1973). J . Cell Biol. 57, 499-524. Chan, K. Y., Bunt, A. H., and Haschke, R. H. (1980). J . Neurocyrol. 9, 381-403. Chan, K. Y . , Bunt, A. H . , and Haschke, R . H. (1981). Neuroscience 6, 59-69. Colman, D. R., Scalia, F., and Cabrdles, E. (1976). Brain Res. 102, 156-163. Contreras, R . J . , Comez, M. M.. and Norgren, R. (1980). J . Comp. Neurol. 190, 373-394. Cook, R. D., and Burnstock, G . (1976). J . Neurocytob 5, 171-194. Cowan, W. M. (1970). I n “Contemporary Research Methods in Neuroanatomy” (W. J . H. Nauta and S . 0. E. Ebbenson, eds.), pp. 217-151. Springer-Verlag. Berlin and New York. Cowan, W . M., and Cuenod, M. (1975). In “The Use of Axonal Transport for Studies of Neuronal Connectivity” (W. M. Cowan and M. Cuenod, eds.), pp. 2-24. Elsevier, Amsterdam. Cragg, B. G . (1970). Brain Res. 23, 1-21. Cullheim, S . , and Kellerth, J . 0. (1976). Neurosci. Left. 2, 307-313. Cullheim, S . , Kellerth, J . , and Couradi. S . (1977). Bruin Res. 132, 1-10. Dekker, J. J . , and Kuypers. H. G. J . M. (1976). Brain Res. 117, 387-398. DeOlmos,J . , and Heimer, L. (1977). Neurosci. Lett. 6 , 107-114. Dietrichs, E., Walberg, F., and Nordby, T. (1981). Brain Res. 204, 179-183. Dogiel, A. S . (1895). Anat. Anz. 10, 517-528. Droz, B., Koening, H.. and diGiamberardino, L. (1973). Bruin Res. 60, 93-127. Droz, B., Ranibourg, A , , and Koening, H . (1975). Brain Res. 93, 1-13. Eckert, H. E., and Boschek, C. B. (1980). I n “Experimental Entomology: Neuroanatomical Techniques” (N. J . Strausfeld and T. A . Miller, eds.), pp. 325-339. Springer-Verlag, Berlin and New York. Egger, M. D., Freeman, N. C. 0.. Malamed, S., Masarachia, P., and Proshansky, E. (1981). Bruin Res. 207, 157-163.


27

Elekes. K., and Szabo, T. (1982). Bruin Res. 237, 267-281. Ellison, F. P., and Clark, G. M. (1975). J . Comp. Neurol. 161, 103-1 13. Feher, E., and Csanyi, K . (1974).Actu Anut. 90, 617-628. Feher, E.. and Vajda, J . (1972). Actu Morphol. Acud. Sci. Hung. 20, 13-25. Feher, E., and Vajda, J. (1974). Actu Anut. 87, 97-109. Feher, E.. and Vajda, J . (1982a). Actu Morphol. Acud. Sci. Hung. 30, 57-63. Feher, E., and Vajda. J . (l982b). Z . Mikrosk. Anut. Forsch. 96, 2-9. Friend, D. S. (1969). J . Cell B i d . 41, 269-279. Furness, J . B.. and Costa, M. (1980). Neuroscience 5 , 1-20. Giorgi, P. P., and Zahnd, J. (1978). Neurosci. Lett. 10, 109-1 14. Gobel, S., and Falls, W. M. (1979). Bruin Res. 175, 335-340. Graham. R. C.. and Karnovsky, M. J. (1966). J . Hisrochem. C.ytochem. 14, 291-302. Graybiel, A. M.. and Devor. M. (1974). Bruin Res. 68, 167-173. Gwyn, D. G . , Lesie, R . A., and Hopkins, D. A. (1979). Neurosci. Lett. 14, 13-17. Gwyn, D. G . , Wilkinson, P. H., and Leslie, R. A. (1982). Neurosci. Lett. 28, 139-143. Hanson, H. A. (1973). Exp. Eye Res. 16, 377-388. Hedreen, J . C . , and McGrath, S. (1977). J . Comp. Neurol. 176, 225-246. Heuser, J. F . , and Reese, T. S. (1973). J . Cell Biol. 57, 315-344. Hirsch, J. G., Fedorko, M. E., and Cohn, J. A . (1968). J . Cell Biol. 38, 629-632. Holstege, J . C., and Dekker. J. J. (1979). Neurosci. Lett. 11, 129-135. Holtzman, E. (1971). Philos. Trans. Ser. B . 261, 401-421. Holtzman, E., and Peterson, 1. (1969). J . Cell B i d . 40, 863-869. Holtzman, E., Novikoff, A. B., and Villaverde, H. (1967). J . Cell B i d . 33, 419-434. Holtzman, E., Freedman. A. R . , and Kashner, L. A. (1971). Science. 173, 733-736. Holtzman, E., Teichberg, S., Abraham, S . J., Citkiwitz, E., Crain, S . M . , Kawai, N., and Peterson, E. (1973). J . Hisrochem. Cvtochem. 21, 349-385. Holtzman, E., Schacher, S . , Evans, J., and Teichberg, S. (1977). I n “Cell Surface Reviews” (G. Poste and G. L. Nicholson, eds.), pp. 165-246. Elsevier, Amsterdam. Hunt, S . P., Streit, P., Kiinzle, H., and Cuenod, M. (1977). Bruin Res. 129, 197-212. Ito, H., Tanaka, H., Sakamoto, N., and Morita, Y. (1981). Bruin Res. 207, 163-169. Itoh, K., Konishi, A , , Noniura, S . , Mizuno, N . , Nakamura, Y., and Sugimoto, T. (1979). Bruin Res. 175, 341-346. Iversen. L. L., Stockel, K . , and Thoenen, H. (1975). Bruin Res. 88, 37-43. Jacobson, S . , and Trojansowski, J. Q. (1975). Bruin Res. 85, 385-401. Jankowska, E., Rastad, J . , and Westman, J. (1976). Bruin Res. 105, 557-562. Jones, E. G., and Hartman, B. K. (1978). Annu. Rev. Neurosci. 1, 215-296. Jones, E. G . , and Leavitt, E. (1973). Bruin Res. 63, 414-418. Kadanoff, D., and Spassowa, I. (1959). Acfu Neuroveg. (Vienna) 20, 19-32. Kalia, M., and Davies, R. 0. (1978). Bruin Res. 149, 477-481. Kalia, M. and Mesulam. M. M. (1980). J . Comp. Neurol. 193, 435-465. Keefer, D. A. (1978). Bruin Res. 140, 15-32. Kerkut, G. A., Shapira, A., and Walker, R. J . (1967). Comp. Biochem. Physiol. 23, 729-748. Kistler, H. B., and Schwartz, J. H. (1982). Bruin Res. 244, 343-346. Kitai, S. T . , Kocsis, J. D., Preston, R. J., and Sugimori, M. (1976). Bruin Res. 109, 601-606. Krishnan, N . , and Singer, M. (1973). Am. J . Anut. 136, 1-14. Kristensson, K. (1975). I n “The Use of Neuronal Connectivity” (V. M. Cowan and M. CuCnod, eds.), pp. 69-81, Elsevier, Amsterdam. Kristensson, K., and Olsson, Y . (1971). Acru Neuropurhol. (Berlin) 19, 1-9. Kristensson, K., and Olsson, Y . (1973a). f r o g . Neurobiol. 1, 85-109. Kristensson, K., and Olsson, Y. (1973b), Aclu Neuroputhol. (Berlin) 23, 43-47.

28

ERZSEBET FEHER

Kristensson. K., and Olsson, Y. (1974). Bruin Res. 79, 101-109. Kristensson, K., Olsson, Y., and Sjostrand, J . (1971). Brain Res. 32, 399-406. Kuntz, A. (1922). Anut. Rec. 24, 193-210. Kuo, D. C . , Krauthamer, G. M., and Yamasaki, D. S. (1981). Bruin Res. 208, 187-191. Kuypers, H. G. J . M., Kievit, J . , and Groen-Klevant, A. C . (1974). Bruin Res. 67, 211-218. and Kellerth, I. 0. (1981). Bruin Res. 207, Langerback, P. A . , Ronnevi, L. O., Cullheim, J. 0.. 247 -266. LaVail, J . H. (1975). In “The Use of Axonal Transport for Studies of Neuronal Connectivity” (W. M. Cowan and M. Cuenod, eds.), pp. 217-248. Elsevier, Amsterdam. LaVail, J . H. (1978). In “Neuroanatomical Research Techniques” (R. T. Robertson, ed.), pp. 355384. Academic Press, New York. LaVail, J. H . , and LaVail, M. M. (1972). Science 176, 1416-1417. LaVail, J . H . , and LaVail, M. M. (1974). J . Comp. Neurol. 157, 303-358. LaVail, J . H . , Winston, K . R., and Tish. A. (1973). Bruin Res. 58, 470-477. LaVail, J. H., Rapisardi, S., and Sugino, I. K. (1980). Bruin Res. 191, 3-20. Litchy, W. J . (1973). Bruin Res. 56, 377-381. Luiten, P. G. M. (1975). Bruin Res. 89, 181-186. Luiten, P. G . M., and Van der Pers, J . N. C. (1977). J . Comp. Neurot. 174, 575-590. Lynch, G., Smith, R. L., Mensak, P., and Cotman, C . (1973). Exp. Neurol. 40, 516-524. Lynch, G., Gall, C., Mensak, C. W., and Cotman, C. W. (1974). Bruin Res. 65, 373-380. Malmgren, L., and Olsson, Y. (1978). Bruin Res. 148, 279-294. Malmgren, L., Olsson, Y . , Olsson, T., and Kristensson, K. (1978). Bruin Res. 153, 477-493. Marchbarks, R. M. (1982). Bruin Res. 244, 243-258. Matsumoto, T. (1920). Bull. Johns. Hopkins Hosp. 31, 91-93. McEwen, B . S., and Grafstein, B. (1968). J . Cell Biol. 38, 494-508. Mesulam, M. M. (1978). J . Histochem. Cytochem. 26, 106-117. Mesulam, M. M., and Brushart, T. M. (1979). Neuroscience 4, 1107-1 117. Mesulam, M. M., and Rosene, D. L. (1979). J . Histochem. Cytochem. 27, 763-774. Milton, G. W., and Smith, A. W. M. (1956). J . Physiol. (London) 132, 100-114. Mizuno, N., Nomura, S . , Itoh, K., Nakamura, Y., and Konishi, A. (1978). Exp. Neurol. 59, 254262. Mizuno, N . , Konishi, A,, Itoh, K., Iwahovi, N . , and Nakamura, Y. (1980). Neurosci. Lett. 20, 1114. Morrell, J . I., Greenberger, L. M., and Pfaff, D. W. (1981). J . Histochem. Cyrochem. 29,903-916. Muller, K. J . , and McMahan, U. J . (1976). Proc. R . SOC.(London) Ser. B 194, 481-499. Nagasawa, J . , Douglas, W. W., and Schultz, R. A. (1971). Nuture (London) 232, 341-342. Nakamura, Y., Mizuno, N., and Konishi, A. (1981). Bruin Res. 212, 127-130. Nauta, H . J. W., Pritz, M. B., and Lasek, R. J. (1974). Bruin Res. 67, 219-238. Nauta, H. J . W., Kaiserman-Abramof, I. R., and Lasek, R. J . (1975). Bruin Res. 85, 373-384. Nicholson, J. E., and Severin, C. M. (1981). Neurosci. Lett. 21, 149-154. Nomura, S., and Mizuno, N. (1981). Bruin Res. 214, 229-237. Nomura, S., and Mizuno, N. (1982). Bruin Res. 236, 1-13. Ochs, S. (1972). Science 176, 252-260. Ohara, P. T., and Lieberman, A. D. (1981). Bruin Res. 207, 153-156. Panneton, W. M., and Loewy, A. D. (1980). Bruin Res. 191, 239-244. Prestige, M. C. (1970). In “The Neurosciences, Second Study Program” (F. 0. Schmitt, ed.), pp. 73-82. Rockefeller Univ. Press, New York. Price, P., and Fisher, A. W. F. (1978). J . Anut. (London) 125, 137-147. Proshansky, E., and Egger, M. D. (1977). Neurosci. Lett. 5, 103-110. Pysh, J . J., and Wiley, R. G . (1974). J . Cell Biol. 60, 365-374. Ralston, H. J . , 111, and Sharp, P. V. (1973). Bruin Res. 62, 273-278.


29

Ralston, H. J., 111, Light, A. R., and Perl, E. R. (1978). Neurosci. Absrr. 4, 570. Ralston, H. J., 111, Light. A. R., Perl, E. R., and Ralston, D. D. (1980). Anut. Rec. 196, 152. A. Rastad, J. (1978). In “Amino Acids as Chemical Transmitters” (F. Fonnum, ed.), pp. 39-48. Plenum, New York. Rastad, J . (1981). Bruin Res. 223, 397-401. Rastad, J., Jankowska, E., and Westman, J. (1977). Bruin Res. 135, 1-10. Repirant, J. (1975). Bruin Res. 85, 307-312. RCthelyi, M., Light, A. R., and Perl, E. R. (1979). Neurosci. Absrru. 5, 728. RCthelyi, M., Light, A. R., and Perl, E. R. (1982). J. Comp. Neurol. 207, 381-393. Ripps, H., Shakib, M., and McDonald, F. D. (1976). J. Cell Biol. 70, 86-96. Robson, J. A., and Mason, C. A. ( 1979). Neuroscience 4, 99- I I 1 , Ross, Ch. A., Ruggiero, D. A., and Reis, D. J. (1981). Bruin Res. 223, 402-408. Samson. F. E., Jr. (1971). J. Neurobiol. 2, 347-360. Satomi, H., Yamamoto, T., Ise, H., and Takahashi, K. (1979). Neurosci. Lerr. 11, 259-263. Scalia, F., and Colman, D. R. (1974). Bruin Res. 79, 496-504. Schacher, S . , Holtzman, E., and Hood, D. C. (1976). J. Cell Biol. 70, 178-192. Schmitt, F. 0. (1968). Proc. Nurl. Acud. Sri. U.S.A. 60, 1052-1100. Schofield, G. C. (1960). Bruin 83, 490-514. Schofield, G. C. (1968). In “Handbook of Physiology. Section on the Alimentary Canal” (F. C. Code and W. Heidel, eds.), pp. 1579-1628. American Physiological Soc., Washington, D.C. Schwab, M. E. (1977). Bruin Res. 130, 190-196. Sellinger, 0. Z., and Petiet, P. D. (1973). Exp. Neurol. 38, 370-385. Sherlock, D. A., Field, P. M., and Raisman, G. (1975). Brain Res. 88, 403-414. Sjostrand, J. (1965). Z. Zellforsch. 68, 481-493. Sjostrand, J. (1970). Bruin Res. 18, 461-467. Smith, D. S. (1971). Philos. Trans. R. Soc. London Ser. B 261, 395-405. Snow, P. J., Rose, P. K.. and Brown, A. G. (1976). Science 191, 312-313. Somogyi, P., Hodgson, A. J., and Smith, A. D. (1979). Neuroscience 4, 1805-1852, Sotelo, C.. and Riche, D. (1974). Anat. Histol. Embryo/. 146, 209-218. Stuesse, S. L. (1982). Bruin Res. 236, 15-25. Takeuchi, Y., McLean, J. H., and Hopkins. D. A. (1982). Bruin Res. 239, 583-588. Teichberg, S . , and Bloom, D. (1976). J. Cell Biol. 70, 285-286. Teichberg. S., Holtzman, E., Crain, S . M., and Peterson, E. R. (1975). 1. CellBiol. 67, 215-230. Theodosis, D. T. (1982). Bruin Res. 233, 3-16. Tsukita, S.. and Ishikawa, H. (1980). J. Cell B i d . 84, 513-530. Turner, P. T . , and Harris, A. B. (1973). Nurure (London) 242, 57-59. Turner, P. T., and Harris, A. B. (1974). Bruin Res. 74, 305-326. Vanegas, H . , Hollander, H., and Distal, H. (1978). J. Comp. Neurol. 177, 193-212. Villegos, G. M., and Fernandez, J. 11966). Exp. Neurol. 15, 18-36. Warr, W. €4. (1973). Anur. Rec. 175, 464-465. Watson, W. E. (1968). J. Physiol. (London) 196, 122-123. Weldon, P. R. (1975). J. Neurocyrol. 4, 341-356. Wessels, N. K., Ludvena, M. A., Letournean, M. A . , Wrenn, P. C., Spooner, B. S . , and Thorotrast, C. (1974). Tissue Cell 4, 757-776. Wilson, Ch. J.. and Groves, P. M. (1981). Bruin Res. 220, 67-80. Winer, J. A. (1977). Biohehuv. Rev. I , 45-54. Winfield, D. A., Gatter, K . C., and Powell, T. P. S. (1975). Bruin Res. 92, 462-467. Wood, J. D. (1975). Phvsiol. Rev. 55, 307-324. Zacks, S. I., and Saito, A. (1969). J. Histochem. Cytochem. 17, 161-170. Zimmerman, H., and Denston, C. R. (1977). Neuroscience 2, 715-730.


INTERNATIONAL REVIEW OF CYTOLOGY. VOI. YO

DNA Sequence Amplification in Mammalian Cells JOYCEL. HAMLIN,JEFFREYD. MILBRANDT,' NICHOLASH. HEINTZ,~ AND JANE C. AZIZKHAN~ Department of Biochemistry, University of Virginia School of Medicine. Charlottesville, Virginia I. Introduction . . . . . .................... 11. Occurrence of Am .................... A. Phylogenetic Range . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Amplified Loci in Mammalian Cells . . . . . . . ..... C. Amplification of Transfected Genes . . . . . . . . D. Evidence for Amplification during Evolution . . . . . . . . . . E. Known and Probable Sequence Amplifications in Maligna 111. Cytological Manifestations of Gene Amplification . . . . . A. Homogeneously-Staining Regions (HSRs) . . . . . . . . , . . . . . . . . . ....................... B. Double Minutes (DMs) . . . . . . C. Relationship between DMS and H S R s . . . . . . . . . . . . . . . . . . . . . IV. Nature of Amplified Sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...... A. Amplified Endogenous Genes. . . . . . . . B . Amplified Transfected Genes . . . . . . . . . . . . . . . . . . . . . . . . . . . . V. Agents That Increase the Frequency of Amplification . . . . . . . . . . . . A. Agents That Interfere with DNA Metabolism . . . . . . . . . . . . . . . B. Growth-Promoting Substances . . . . . . . . . . . . . . . . . . . . . . . . . . . VI. Proposed Mechanisms of Sequence Amplifi A. Unequal Sister Chromatid Exchange . . . . . . . . . . . . . . . . . . . . . . B. Rereplication . . . . . . . . . . . . . . . . . . . . VII. Concluding Remarks . . . ....................... ....................... References . . . . . . . . . . .

31 33 34 38 40 42 43 45 45 50 52 57 58 63 64 64 67 67 68 71 75 77

I. Introduction In the typical eukaryotic somatic cell cycle, each chromosome is precisely duplicated during the DNA synthetic (S) period. The synthesis of a chromosomal DNA fiber occurs through the agency of thousands of tandemly arranged replicons (Huberman and Riggs, 1968), each of which usually functions only once 'Present address: Division of Laboratory Medicine, Washington University School of Medicine, St. Louis, Missouri 631 10. 'Present address: Department of Pathology, University of Vermont College of Medicine, Burlington, Vermont 05401. 'Present address: Department of Biology, The Johns Hopkins University, Baltimore, Maryland 21218. 31 Copyright 0 1Y84 by Academic Prew. Inc All rights of rcpraluclion In any form rexrvcd ISBN 0-12-364490-9

32

JOYCE L. HAMLIN ET AL.

in a given S period to ensure that the fiber is exactly duplicated along its entire length. The result of this process is that the two identical daughter chromatids lie side-by-side, connected by a centromere, until separation of chromatids occurs at mitosis and the ploidy of each daughter cell is restored to the original configuration. There are exceptions to this mode of replication, however, in which parts of chromosomes or the entire chromosomal complement are rereplicated prior to a cell division event, with the consequence that the genetic constitution of the cell can be increased in total, or only at selected loci. During polytenization in the larval stages of certain Dipteran species, the chromosomes replicate over and over again without intervening mitosis, resulting in as many as 8000 copies of the genome in a single cell (Daneholt and Estrom, 1967). The result of this process is that the multiple identical DNA fibers lie together in colinear bundles. A related phenomenon known as endoreduplication occurs infrequently in a variety of cell types, and results in the precise reduplication of the entire chromosomal complement without intervening cytokinesis (Herreros and Gianelli, 1967; Gatti et al., 1973). The frequency of endoreduplication can be increased by several agents, including the mitotic spindle inhibitor colchicine (Weber and Hoegerman, 1980; Sutou, 1981). Both of the above processes seem to affect all chromosomal replicons in the same way, and may reflect the overriding of controls that usually prevent initiation of the S period until after mitosis. During development, there are situations in which a single genetic locus can be preferentially rereplicated during the cell cycle by a process known as amplification. The preferential replication of ribosomal genes during oogenesis is a widespread phenomenon in both animals and plants, and presumably allows the organism to cope with the great demand for protein synthesis during development of the oocyte (See Long and Dawid, 1980, for review). In the follicle cells of certain insects, a developmentally controlled process results in the amplification of selected chorion genes whose products are utilized in egg shell formation (Spradling and Mahowald, 1980). Another type of selective amplification has received a great deal of attention in recent years, and is the major focus of this review. When cultured mammalian cells are selected for resistance to increasingly higher concentrations of certain drugs over the period of many months, cell lines can eventually be isolated that greatly overproduce the target protein (usually an enzyme) for the corresponding drug. In almost every case of extremely high levels of resistance, a DNA sequence containing the gene coding for the target protein has been shown to be selectively amplified. The amplified sequences are located in expanded chromosomal regions or on extrachromosomal double minutes. This latter mode of amplification has generated considerable interest for several reasons. Since drug

DNA SEQUENCE AMPLIFICATION

33

resistance of this type has been observed in patients and cultured cell systems that were treated with anticancer agents such as methotrexate, gene amplification is thought to be a major problem in cancer chemotherapy. In addition, evidence has accumulated for many years that tumor cells taken from patients who have apparently not been subjected to drug treatment often display chromosomal anomalies such as the double minutes and expanded chromosomal regions that are characteristic of gene amplification in experimental systems. With the recent discovery that the overproduction of certain normal gene products from cellular oncogenes can lead to the transformed phenotype (see Bishop, 1983, for review), it is clear that gene amplification could be one underlying mechanism in malignancy. Since amplification has been observed at virtually every genetic locus for which there is a suitable selective agent, the phenomenon is apparently widespread and possibly random. This notion is supported by the vast literature on existing gene duplications in mammalian cells. Through the workings of evolution, the extra copy or copies of a gene can be conserved and can function to produce larger amounts of their gene product than could the diploid complement (as in the case of ribosomal genes); or the extra copies can be mutated, leading eventually to new functions in the cell or to inactivation in the case of pseudogenes. Thus, the amplification process observed during the acute development of drug resistance in cultured cells may be a telescoped version of the duplication and amplification mechanisms that have taken place over millions of years to produce the complex genomes of higher organisms. In this article, we will begin by citing several examples of selective DNA amplification in both prokaryotic and eukaryotic systems. We will then discuss the cytological manifestations of gene amplification, and the stability of the process. We will cite recent studies on the molecular nature of the amplified sequences in selected systems, and the types of agents that might provoke gene amplification. We will then attempt to put the major observations related to this interesting chromosomal phenomenon into perspective by discussing certain models for the mechanisms involved in gene amplification.

11. Occurrence of Amplification Phenomena

We will discuss here specific examples of gene duplication and amplification phenomena that occur throughout the evolutionary spectrum, in order to tabulate the possible mechanisms that may be available to mammalian cells for the type of amplification associated with drug resistance. We will then focus on the various genetic loci that are known to be duplicated or amplified in mammalian cells, in order to indicate the broad spectrum of this phenomenon.

34


A. PHYLOCENETIC RANGE Gene duplication and tandem amplification in bacteria are frequent occurrences that are detected in drug selection protocols, and have been observed in several bacterial species and at several genetic loci (Anderson and Roth, 1977). Amplification of a single locus can proceed to the point at which the multiple copies can represent 20% of the bacterial chromosome (Anderson et a/., 1976). There are numerous examples of the amplification of drug resistance genes carried on bacterial plasmids (Clowes, 1972). Exposure to the particular antibiotic selects for cells carrying plasmids with tandemly repeated resistance markers (Rownd, 1982). In the F1 plasmid, PRSDl (Schmitt et al., 1979; Mattes et al., 1979), each gene is flanked by insertion sequences, forming a transposon, which is apparently the unit of amplification. The direct repeats flanking the gene are required for amplification, and the process is apparently dependent on host recombination systems. Phage p is itself a transposable element, and transposition from one genomic site to another probably occurs by a process involving replication of the element (Galas and Chandler, 1981; Harshey and Bukhari, 1981). In the process, the phage can be inserted in multiple tandem copies by a mechanism that may involve breakage of the host chromosome and attachment to a nicked end of the element, followed by continuous replication into the chromosome. The bacteriophages A (Edlund et al., 1980), T4 (Kozinski et al., 1980), and PI (Meyer and lida, 1979) have also been shown to undergo selective and multiple reduplications of particular genetic markers. In the case of T4, the amplified loci lie in the vicinity of the known origins of replication (Kozinski et al., 1980). Gene amplification has also been reported in the chromosomes of yeast. In particular, the gene for metallothionein is amplified in tandem after stepwise selection in cadmium (Fogel and Welch, 1982). In petite mutants of yeast, amplification of mitochondria1 DNA sequences occurs by a process resulting in the formation of multiple, tandem copies of fragments of the original mitochondrial genome arrayed in a circle (Gaillard et ul., 1980). The genetic constitution of the repeated unit varies in different mutants, and each repeated unit contains an origin of DNA synthesis (de Zamaroczy et al., 1981; Hyman et a / . , 1982). In insects, certain forms of resistance to insecticides display several of the properties characteristic of drug resistance in mammalian cells. Resistance appears to result from high levels of the target protein for the insecticide, and can be stable or unstable when the chemical agent is removed from the environment (Goldstein, 1974). The natural developmental process of ribosomal DNA amplification during oogenesis (reviewed in Long and Dawid, 1980) occurs in a large variety of organisms, including hypotrichs, echiuroid worms, clams, insects, fish, amphibia, mammals, and plants. The amplified ribosomal DNA (rDNA) copies are


35

usually extrachromosomal, and are arrayed in tandem as linear or circular elements. In the case of Xenopus laevis, the repeated units appear to be homogeneous in length (Wellauer et al., 1976). Since there are multiple integrated rDNA copies of variable length in the Xenopus genome prior to amplification, this finding suggests that amplification occurs from a single integrated unit in a selective way (Wellauer et al., 1976). Furthermore, different oocytes from the same animal can amplify different integrated copies (Wellauer et al., 1976; Bird, 1978). Electron microscopic observations have suggested that rDNA amplification in Xenopus occurs via a rolling circle mechanism analogous to the mode of replication of the single-stranded DNA bacteriophage, XI74 (Hourcade et u l . , 1973; Rochaix er al., 1974; Buongiorno-Nardelli et a!., 1976). A related developmental phenomenon occurs in the ovarian follicle cells of certain insects during eggshell formation. Members of the chorion multigene family are organized into two clusters on different chromosomes in Drosophila (Spradling er a l . , 1980; Griffin-Shea et a/., 1980). During oogenesis, each cluster is amplified in situ (Spradling and Mahowald, 1980). Bidirectional replication proceeds from a fixed origin of DNA synthesis located in a central position in each locus (Spradling, 1981). After multiple initiations, the copy number of sequences flanking the origin decreases with distance from the origin, suggesting that replication forks terminate at random positions. Since the amplified DNA sequences are recovered in high-molecular-weight DNA fractions, it appears that the multiple daughter strands do not separate after replication. This suggestion has recently been confirmed in electron microscopic studies on the follicle cells of Drosophila. Multiple replication bubbles lying side-by-side in an onion skin array were seen to terminate at random positions relative to the center of the replication bubble (Oscheim and Miller, 1983). Another example of selective amplification occurs in the Dipteran, Sciaraidae. Several loci are selectively overreplicated during polytenization in the late larval stage of development, apparently as a result of hormonal stimulation (see Bostock and Sumner, 1978, for review). The extra copies remain associated with the giant polytene chromosomes as DNA puffs, and may result from a process similar to that observed in chorion gene amplification in Drosophila. Considerable information exists about the amplification of several viruses that integrate into the chromosomes of higher eukaryotic cells. In the papova virus group, SV40, Polyoma, and BK viruses can integrate as single copies or as headto-tail tandem repeats in transformed host cells (e.g., Botchan et al., 1980; Birg et a/., 1979; Pater et al., 1982). The sites of insertion into the host genome appear to be nonspecific (Gelb et al., 1971; Sambrook et al., 1975; Kutcherlapati et a / . , 1978), and the inserted copies can give rise to free viral DNA (e.g., Hiscott et a / . , 1981; Bullock and Botchan, 1982; Gattoni et al., 1980; Pater et al., 1982). A single integrated copy can apparently also amplify in situ, provided that it contains an origin of replication (Botchan et al., 1979; Colantuoni et al.,

36

JOYCE L. HAMLIN ET AL

1980; Baran et af.,1983). Integration, amplification, and possibly excision seem to require a functional T antigen, a protein required for viral replication (Della Valle et al., 1981; Colantuoni et al., 1982; Botchan et al., 1979). The size of the integrated, tandemly repeated viral unit is conserved during the excision of polyoma; i.e., if the repeated unit is 1.3 equivalents long, then the extrachromosomal element is also 1.3 equivalents long (Gattoni et af., 1980). This suggests that integration and excision events are mediated by some common feature related to homologous recombination. However, models involving concomitant replication and excision of the integrated virus have also been proposed (see Section VI). Studies on BK virus indicate that transformation of semipermissive mouse cells results in the tandem integration of viral DNA into the host chromosome, but in nonpermissive hamster cells, the virus integrates as a single copy (Meneguzzi et af., 1981). However, when the virus is linearized with a restriction enzyme that generates staggered cohesive ends, and hamster cells are subsequently transformed with this DNA, the virus is inserted in tandem arrays (Grossi et al., 1981). This result suggests that the formation of tandem arrays may occur prior to integration in the semipermissive situation, possibly through polymeric DNA replication intermediates. It has also been shown that the DNA damaging agent, mitomycin C, provokes onion skin replication of the integrated polyoma (Baran et al., 1983) and SV40 (Bullock and Botchan, 1982) genomes in inducible cell lines. In the former case, replication was shown to proceed from the viral origin into flanking cellular sequences, and appeared to terminate at fixed positions in the host DNA. The phenomenon of amplification of chromosomal DNA sequences in mammalian cells is a relatively recent discovery, and began with studies on drug resistance in cultured cells. Several years ago, both Fischer (1961) and Hakala et af. (1961) selected mouse cell lines that were resistant to the antifolate drug, methotrexate (MTX), and showed that resistance could be accounted for by an increase in the intracellular level of the target enzyme for this drug, dihydrofolate reductase (DHFR). Littlefield ( 1969) subsequently proposed that overproduction of DHFR in MTX-resistant cells could be due to constitutive expression of a normally repressible gene, if the cell had suffered a mutational loss of the repressor itself. Evidence against this hypothesis was obtained by showing that hybrids between resistant and sensitive cells expressed an intermediate level of drug resistance (Littlefield, 1969), and DHFR synthesis was therefore not turned off by the postulated active repressor supplied by the normal, sensitive cell in the hybrids. As an alternative mechanism for DHFR overproduction, Littlefield suggested that the gene coding for DHFR could be amplified in these cell lines. In 1976, Biedler and Spengler made the critical observation that in cytological preparations of mitotic chromosomes from a series of near-diploid, antifolateresistant Chinese hamster lung cells, a very unusual chromosomal anomaly was


37

consistently correlated with very high levels of drug resistance (Biedler and Spengler, 1976a,b). In these cell lines, expanded chromosomal regions were observed that stained uniformly to an intermediate degree in the G-banding protocol (contrasted to the usual alternating light and dark banding pattern seen on most chromosomes), and stained darkly when subjected to C-banding. These homogeneously staining regions (HSRs) were not present in drug-sensitive Chinese hamster cells. They suggested that HSRs could be a manifestation of the in situ amplification of the gene coding for dihydrofolate reductase, thus accounting for increased levels of the enzyme. Schimke and co-workers were then able to show that DHFR messenger RNA (mRNA) was greatly overproduced in MTX-resistant murine S 180 cells (Kellems et al., 1976), and they subsequently cloned cDNAs (DNA complementary to mRNA) representing DHFR mRNA species overproduced by this cell line (Ah et a/. , 1976). Using these cDNAs in solution hybridization studies, they showed that genomic DNA isolated from resistant mouse cells contained several hundred copies of the DHFR gene (Ah et a / . , 1976). Schimke and collaborators also showed that multiple copies of the DHFR gene could be localized to HSRs in MTX-resistant Chinese hamster ovary (CHO) and murine cell lines by demonstrating the selective hybridization of radioactive DHFR cDNA to the HSRs in mitotic chromosomes (Nunberg et al., 1978; Dolnick et al., 1979). However, in some MTX-resistant mouse cell lines, the extra DHFR genes were shown to reside on extrachromosomal double minutes (DMs)-small paired, acentromeric bodies dispersed among the mitotic chromosomes. It was demonstrated that the number of DMs per metaphase spread correlated roughly with the number of amplified DHFR genes, and, further, that the DHFR genes cosedimented with the DM fraction on sucrose gradients (Kaufman et al., 1979; Brown et a / . , 1981). These structures were only seen in unstable variants of murine S180 and L1578Y (i.e., those that quickly lose resistance to MTX after removal of drug from the culture medium). Stable variants of the same origin manifested HSRs, and did not contain DMs (Dolnick et al., 1979). Thus, by early 1980, it was reasonably clear that the predominant mechanism for resistance to MTX in cultured mammalian cells was the amplification of the gene coding for DHFR. The amplification of DHFR genes during development of MTX resistance has now been shown to occur in a variety of mammalian cells. DHFR gene amplification has been observed in murine 3T6 (Brown et a / . , 1981), Sarcoma 180 (Kaufman et al., 1979), the EL4 and L1578Y lymphomas (Bostock and TylerSmith, 1982; Kaufman et al., 1979), the SEWA ascites tumor (Martinsson et al., 1982), and the PG193 melanoma (Bostock and Clark, 1980). MTX-resistant variants of the karyologically stable CHO (aneuploid) and the Chinese hamster lung (near-diploid) cell lines have been extensively characterized (Biedler et a/., 1980; Nunberg et al., 1978; Milbrandt et al., 1981; Flintoff et al., 1982). In

38

JOYCE L. HAMLIN ET A L

addition, human leukemias (Srimatkandada et al., 1983; Horns et al., 1984), ovarian adenocarcinomas (Trent et al., 1984), breast cancer cells (Cowan et a l . , 1982), and HeLa cells (Wolman et a l . , 1983), as well as a series of rat hepatomas (Fougere-Deschatrette et a l . , 1982) have all been shown to amplify the DHFR gene in response to treatment with increasing concentrations of MTX. Recently, the gene coding for a bifunctional thymidylate synthetase-dihydrofolate reductase in the parasite, Leishmania tropica, has also been found to be amplified in response to MTX treatment (Coderre et ul., 1983).

B. AMPLIFIED LOCIIN MAMMALIAN CELLS To date, the amplification of DHFR genes in the establishment of MTX resistance is the most extensively studied example of gene amplification in mammalian cells, primarily because it was the first to be described. However, the list of genetic loci that can undergo endoreduplication in mammalian cells is growing, and, indeed, seems limited only by the availability of suitable drugs for isolating resistant variants at any given locus. Stark and co-workers have determined that resistance to the antimetabolite, N-(phosphonacety1)-L-aspartate (PALA), is developed in Syrian hamster cells by the overproduction of the multifunctional CAD protein that catalyzes the first three steps in uridine biosynthesis (Wahl et al., 1979). They have shown that the CAD gene is amplified as much as 100-fold in some cell lines, and the multiple genes can be localized to an HSR in each PALA-resistant cell line (Wahl et al., 1982). The metallothioneins are proteins that sequester toxic metals such as cadmium and zinc, and their synthesis is regulated by both heavy metals and by glucocorticoids (Kagi and Nordberg, 1979; Karin et al., 1980). When cultured murine Hepa 1A (hepatoma), S180 (sarcoma), or Friend erythroleukemia cells are subjected to stepwise increases in cadmium concentration over the period of several months, resistant variants are recovered that overproduce metallothionein and its mRNA, and contain 10-60 copies of the metallothionein gene (Beach and Palmiter, 1981; Beach et al., 1981). The Hepa 1A cells displayed numerous DMs, but neither the S 180 nor erythroleukemia cells displayed karyotypic anomalies that could be related to DMs or HSRs (Beach et al., 1981). Amplification of metallothionein genes also occurs in Chinese hamster ovary cells without apparent karyotypic changes (Gick and McCarty, 1982). The amplification of adenosine deaminase in mouse C 1 fibroblasts is responsible for the resistance developed to stepwise increases in coformycin (Yeung et al., 1983a). Increased levels of adenosine deaminase were observed in this system, resistance was lost upon removal of the drug, and DMs were detected in metaphase spreads (Yeung et a l . , 1983a). Cloned adenosine deaminase cDNA probes were used to show amplification of the cognate gene (Yeung et a l . ,


39

1984). The gene for adenosine deaminase is also amplified in human choriocarcinoma cells (Yeung et a / . , 1983b) and in CHO cells (Debatisse et af., 1982) after stepwise increases in selective drug. The X-linked gene for HPRT (hypoxanthine/guanine phosphoribosyltransferase) is amplified in both mouse neuroblastomas (Brennand et al., 1982) and in Chinese hamster cell lines (Fuscoe et af., 1983). In the former case, three copies of the X chromosome are fused in a large, rearranged marker chromosome, but this obvious threefold amplification cannot account for the apparent 50-fold amplification of the gene relative to normal cells (Melton et al., 1981). No HSRs or double minutes were observed in this system. Amplification of the HPRT locus was detected in the Chinese hamster cell system by selecting for revertants of a thermo-sensitive HPRT mutant by growth at 39°C (Fuscoe et a/., 1983). This approach was also used by Chasin and colleagues to isolate amplified, mutant DHFR genes without selection in MTX (Chasin et al., 1982), and seems to work in both of the above cases by the overproduction of a partially active protein. Several other drug treatment regimens have been used to select mammalian cells that overproduce the target protein, and many have been shown to have amplified the corresponding gene. 5-Fluorodeoxyuridine selects for the overproduction of thymidylate synthetase (Baskin et al., 1975; Rossana et ul., 1982), hydroxyurea for the M2 subunit of ribonucleotide reductase (Akerblom et af., 1981), albiizin for asparagine synthetase (Andrulis et al., 1983), compactin for 3-hydroxy-3-methyl glutaryl CoA reductase (Chin et al., 1982), tunicamycin for N-acetylglucosaminyltransferase (Criscuolo and Krag, 1982), and colchicine and vincristine for unidentified (microtubular?) proteins (Kopnin, 198 1 ; Biedler, 1982). In many of these cases, some karyological manifestation of amplification is observed in the form of DMs or HSRs. Multidrug cross-resistance has been described by Baskin and colleagues in uptake mutants of murine neuroblastoma cells that were treated with either maytansine, vincristine, adriamycin, or Baker's Antifol (Baskin et al., 1981). These cell lines contain numerous DMs and have been shown to contain elevated levels of alkaline phosphatase. Biedler and Riehm (1970) have also described multidrug cross-resistance in mouse and hamster cell lines that were selected for resistance to actinomycin D. There are, in addition, numerous examples of the nondevelopmental amplification of rDNA genes in mammalian tumor cell lines, including several human and rat neoplasms (Henderson and Megraw-Ripley, 1982; Murao et al., 1982; Tantravahi et d., 1981, 1982). In these instances, there is no obvious selection pressure that maintains the extra copies in the genome. All of these amplifications appear as HSRs, and many are located at the known positions of nucleolar organizer regions in these cell lines (Henderson and Megraw-Ripley, 1982).

40


Interestingly, the sarcoma line, XC, contains an additional unidentified HSR that does not contain rDNA genes, suggesting a possible predisposition for amplification in these cells (Tantravahi et al., 1982).

c. AMPLIFICATION OF TRANSFECTED GENES There is a new class of gene amplification that has arisen as a result of the molecular cloning of selectable genes [e.g., DHFR, CAD, and thymidine kinase (TK)] and the advent of gene transfer techniques for introducing DNA into mammalian cells. Cloned DNA is usually delivered to cells either as a CaPO, coprecipitate which is ingested by pinocytosis (Graham and van der Eb, 1973), or by fusion of cells to bacterial spheroplasts harboring recombinant plasmids (Schaffner, 1980). After long-term culture of the transformed derivatives (obtained by growth in a medium that selects for the function imparted by the transferred gene), the DNA is invariably integrated into the chromosome at one or a few sites, and, depending upon the mode of delivery, can often integrate as multicopy, tandem arrays. In the case of the CAD and DHFR genes, stepwise increases in the corresponding drug results in amplification of the transfected genes. The first such instance reported involved the introduction of the entire CAD gene (contained in a recombinant cosmid) into CAD-deficient or wild-type CHO cells (de Saint Vincent et al., 1981). The ura+ derivatives of the CAD-deficient cells were shown to contain more than 10 copies of the recombinant CAD gene, which were subsequently amplified further upon treatment with increasing concentrations of PALA. Transfectants of wild-type CHO cells were selected in a high concentration of PALA directly, and contained multiple copies of the CAD gene, accounting for PALA resistance. The PALA-resistant clones were shown to contain HSRs on a chromosome distinct from the probable location of the CAD gene in wild-type cells. The entire 25 kb CHO DHFR gene has been cloned in a cosmid, and has been used to rescue a DHFR- CHO cell line to the DHFR phenotype (Milbrandt et al., 1983a). Upon amplification after selection with increasing concentrations of MTX, the extra DHFR gene copies (500-700 per diploid genome) were shown to reside in an HSR on a chromosome distinct from the parental gene location, as shown by the binding of radiolabeled DHFR probes to mitotic chromosomes (Milbrandt et al., 1983b). It was also demonstrated that the amplified sequence included more than 90 kb of DNA in addition to the transfected gene itself. This extra DNA probably represents genomic sequences flanking the site of insertion of the recombinant cosmid into the chromosome. The very large (31 kb) murine DHFR gene has been truncated to smaller versions by a variety of experimental approaches, and these minigenes have been used to study DHFR expression and amplification after introduction into CHO +


41

and murine cell lines. Crouse and co-workers have constructed chimaeras of DHFR genomic and cDNA sequences that convert DHFR- CHO cells to the DHFR+ phenotype after transfection (Crouse et a / . , 1983). In all instances, multiple copies of the DHFR plasmids were inserted into the genome of the recipient cell. DHFR transformants with approximately fiva: copies of the recombinant plasmid could be shown to amplify the inserted DNA further when subjected to incremental increases in MTX, and flanking cellular sequences were amplified as well. Interestingly, if carrier mouse embryo DNA was added to the DHFR plasmids during the transfection step, clones were obtained that had integrated very large numbers of the minigene (sometimes more than 400 copies per genome), apparently in tandem arrays. These clones were selected under conditions that presumably required only one or a few copies of the plasmid (i.e., were not subjected to MTX selection). This raises the possibility that certain sequences in the carrier mouse embryo genomic DNA were picked up by the plasmids as a result of intracellular ligation, and facilitated amplification. Gasser et al. (1982) and Kaufman and Sharp (1983) have also used DHFR minigenes to transform DHFR- CHO cells to the wild-type phenotype. They observed that most transformants had integrated multiple copies of the plasmids after selection for wild-type levels of DHFR enzyme. In several transfected clones, pBR322 sequences in the vector portion of the recombinants were lost, possibly through homologous recombination events. Furthermore, in both studies, significant numbers of rescued cell lines showed rearrangements of the input DNA, suggesting the relative instability of inserted DNA in this system. Kaufman and Sharp (1983) showed that a DHFR transformant with multiple inserts at different chromosomal locations amplified only one of the inserts after MTX selection, and that flanking (presumably chromosomal) sequences were coamplified with the DHFR minigene. From one cell line that was transfected with a DHFR minigene that included a large part of the SV40 t antigen gene, subsequent amplification resulted in overproduction of both DHFR and a polypeptide related to t antigen. This important result indicates that selectable and amplifiable cloned genes such as DHFR can be used to amplify any nonselectable colinear gene. Kaufman and Sharp also showed that the amplified copies of this chimaeric plasmid were located in the chromosomes as HSRs after amplification, and were usually at terminal positions or in dicentric chromosomes. These investigators suggest that telomeric regions are favored sites of integration, and that the integration event destabilizes the telomeres, inducing the formation of dicentric chromosomes. This is an interesting suggestion, since many endogenous amplified genes also reside at terminal positions on chromosomes, and dicentric chromosomes have been observed in at least two other cases involving amplification of the DHFR gene (Hamlin, unpublished observations; Fougere-Deschatrette et al., 1982). Murray et al. (1983) have constructed a vector that uses the LTR from Harvey +

+

42


sarcoma virus to provide 5’ regulatory sequences required for the expression of a colinear DHFR cDNA. When this DNA is delivered to NIH3T3 cells, colonies are recovered that are resistant to MTX as a result of the insertion of multiple copies of the chimaeric recombinant plasmid. When subjected to further increases in MTX, the plasmid copy number per genomic equivalent goes up, and DMs are detected in those cell lines with high copy numbers. The viral sequences are also expressed in these constructions, as evidenced by the ability of the plasmid to produce transformed foci on 3T3 cells. These foci all exhibit elevated levels of DHFR and are MTX resistant. Axel and Roberts (1982) have carried out a series of studies in which they have transfected APRT-/TK -L cells with a chimaeric plasmid containing a wildtype APRT gene and a truncated, promoter-less TK gene. The initial APRT+/ TK- transformants gave rise to TK+ revertants at a very high frequency ( as a consequence of amplification of the chimaeric plasmid and the resulting overproduction of the partially functional TK gene. A unique finding in a subsequent study was that the tandemly arranged amplicons in these cell lines are integrated into chromosomes, as opposed to the DMS that are usually observed when mouse cells have amplified endogenous or transfected genes (Roberts et al., 1983). It remains to be seen whether or not the amplification processes that occur after DNA transfection exactly mimic those observed during amplification of endogenous genes. We will consider the nature of this process in Section VI. Thus, the range of mammalian cell types that have been shown to selectively amplify specific genes after drug treatment regimens is extremely broad. The number of drugs that can select for variants of this type is also large, and the list of genes that can be amplified will probably be lengthened as cloned probes for individual genes become available, allowing direct quantitation of gene copy number. D. EVIDENCE FOR AMPLIFICATION DURING EVOLUTION Most of the above examples of selective endoreduplication involve the rather extensive amplification of the gene in question, resulting in tens or even thousands of copies of the gene per diploid cell. However, it is appropriate to point out here that there exists in the genomes of virtually all higher organisms compelling evidence for many duplications and amplifications throughout the course of evolution. The reason for maintenance of reduplicated copies of genes is undoubtedly a combination of tolerance, mutation, and selection on the part of the affected organism. The examples of the multicopy rDNA and chorion gene families have been cited, and to these should be added the multimembered families of tRNA (reviewed in Long and Dawid, 1980), histone (Kedes, 1979), and immunoglobulin (Gottlieb, 1980) genes, as well as globins (Maniatis et al.,


43

1980), actins (Fyrburg et al., 19801, myosins (Robbins et al., 1982; Epstein et al., 1974), collagens (Solomon, 1980), a2-globulin (Kurtz, 1981), vitellogenin (Wahli et al., 1979), interferons (Allen and Fantes, 1980), histocompatibility antigens (Nathanson er al., 1981), and ovalbumin (Royal et al., 1979), all of which have at least three members per haploid genomic equivalent. A partial list of gene duplications in higher eukaryotes includes hexokinase (McLachlan, 1979), ferrodoxin (Wakabayashi ef al., 1980), ricin (Villafranca and Robertus, 1981), calcitonin (Perez et al., 1982), renin (Piccini et al., 1982), isocitrate dehydrogenase (Sattler and Mecham, 1979), lens y-crystallin (Moorman et al., 1982), alcohol dehydrogenase (Oakeshott et al., 1982), salivary amylase (Pronk ef al., 1982), cytosolic malate dehydrogenase (McMillin and Scandalios, 1980), 6-phosphogluconate dehydrogenase (Rao and Rao, 1980), and phosphoglucose isomerase (Gottlieb and Weeden, 1979). However, in many of these instances, the duplicated genes are juxtaposed to one another without significant amounts of flanking non-gene DNA sequence between them. This contrasts with the DNA arrangement found in the development of drug resistance in experimental systems, in which the unit repeated sequences (amplicons) are usually much larger than the gene itself (see Section IV). This difference may indicate that the duplicated genes themselves are retained because they confer some selective advantage throughout evolution, but any extra DNA amplified along with the gene in the initial event is lost by recombination or deletion mechanisms. Alternatively, the original duplication events that led to tandem arrays of genes without intervening DNA may have occurred via mechanisms different than those observed during the acute development of drug resistance observed in experimental systems.

E. KNOWNAND PROBABLE SEQUENCE AMPLIFICATIONS IN MALIGNANCY A most important recent addition to the list of sequence amplifications relates to cellular oncogenes, the counterparts of viral oncogenes whose overexpression has been implicated in the genesis of cellular transformation (Bishop, 1983). In the human myeloid leukemia cell line, HL60, and in primary leukemia cells of the same patient, an 8- to 16-fold amplification of a cellular oncogene (c-myc) has been demonstrated directly with the use of cloned viral oncogene probes (Dalla Favera er a / . , 1982), and overexpression of the corresponding gene was also shown (Westin et al., 1982). The human neuroendocrine tumor lines, Colo 320 DM and Colo 320 HSR, have also been shown to amplify the cellular oncogene, c-myc (Alitalo et a/., 1983). Both lines exhibit enhanced expression of this gene relative to normal cells, and a radiolabeled c-myc genomic fragment hybridizes to the HSR regions in Colo 320 HSR. The cellular oncogene c-Ki-ras, has also been shown to be amplified 30- to 60-fold in cells of the murine adrenocortical tumor lines, YI-DM and Yl-HSR (Schwab et a/.. 1983). In this

44


case, the amplified oncogene has suffered rearrangements, but still expresses inordinately large amounts of the c-Ki-ras mRNA and protein. Finally, the oncogene, c-abl, is amplified 4- to 8-fold in a human myelogenous leukemia cell line (K-562), and may involve a translocation that positions the c-abl sequence next to the gene coding for the immunoglobulin K light chain on the Philadelphia (#22) chromosome (Collins and Groudine, 1983). A causal relationship between oncogene amplification per se and transformation has not been demonstrated in any system as yet, however, and it is important to point out that oncogenes occur naturally in multiple copies in certain species of mouse and hamster (Chattopadhyay et al., 1982). It is possible that amplification of cellular oncogenes is a natural developmental phenomenon characteristic of cells in particular stages of differentiation, and that the cultured tumor cell lines used in the above studies happened to be trapped in these developmental stages. In addition to these examples of known oncogene amplifications in malignancy, there is an extensive list of probable amplifications in various neoplasms in mammals, most of which have not been treated with anticancer drugs. The argument for amplification in these cases derives from the frequent occurrence of either DMs or HSRs in the chromosomal complement. The list of examples is extensive and has recently been reviewed thoroughly by Cowell (1982) and by Barker (1982). We refer to selected reports here in order to indicate the diverse nature of neoplasms in which these chromosomal abnormalities have been observed. In no case has the amplification been shown to be the cause of cellular transformation, and it should be remembered that a hallmark of neoplasms is the vast range of chromosomal abnormalities that they contain. Double minutes were first observed by Spriggs and co-workers in cells from a pleural effusion of a malignant lung tumor (Spriggs et al., 1962). Several cases of double minutes were then reported in tumors of neurogenic origin, particularly those of children. It was soon appreciated, however, that several different kinds of human tumors contained double minutes. Many of the tumors were derived from patients that had never been exposed to chemotherapy or to overt radiation. The list (taken from Barker, 1982) includes carcinomas of the breast, cervix, colon, stomach, bladder, lung, and thyroid; chondro- and osteosarcomas; leukemias and lymphomas; gliomas, medulloblastomas, retinoblastomas, and neuroblastomas; and ovarian and testicular tumors. DMs occur in rodents in a similarly broad range of tumor types, including sarcomas, lymphomas, and neuroblastomas. Several of the murine sarcomas have been induced by Rous Sarcoma Virus or by Polyoma. DMs have been detected in both rats and mice after chemical carcinogenesis or radiation (Barker, 1982). Thus, the presence of DMs is ubiquitous in mammalian tumor cells, and as yet has not been observed in normal cells. However, since DMs are difficult to detect because of their small size and lack of G-banding, it is possible that a renewed effort may detect them in normal tissues.


45

HSRs have also been observed in tumor cells of man and rodents. They were first described in G-banded preparations of cultured human neuroblastomas (Biedler and Spengler, 1976b). Since then, they have been observed in human breast and colon carcinoma cell lines, solid human tumors, transformed mouse salivary epithelial cells, mouse adrenocarcinomas, rat hepatomas, mouse lymphomas, sarcomas, and melanomas (see Barker, 1982, for references). The probability that DMs and HSRs are different (but sometimes interchangeable) configurations of the same basic chromosomal phenomenon will be discussed in Section Ill.

111. Cytological Manifestations of Gene Amplification

While the existence of double minutes and abnormally long chromosomes has been recognized for years in the metaphase spreads of mammalian tumor cells, their association with gene amplification was not recognized until 1976, when Biedler and Spengler observed a strong correlation between high levels of MTX resistance in Chinese hamster lung fibroblasts and the presence of HSRs on long marker chromosomes. When cloned DHFR sequences became available, it was possible to show that the HSR in a MTX-resistant Chinese hamster ovary cell line was the site of amplification of the DHFR gene (Nunberg et a / . , 1978). However, most MTX-resistant murine cell lines did not display obvious HSRs when subjected to G-banding protocols. Instead, fluorescent staining with acridine orange clearly showed large numbers of small, paired chromatin bodies known as double minutes, whose number per cell correlated roughly with DHFR gene copy number in the particular MTX-resistant cell line (Brown e t a / ., 1981). In this section, we will consider HSRs and DMs separately with regard to occurrence, staining properties, size, stability, and replication pattern. We will then consider situations in which the two forms appear to interconvert in some cell lines under experimental manipulation or long-term culture. The reader is referred to the excellent recent review by Cowell ( 1982) on the subject of DMs and HSRs. We present here an overview of the critical features of the karyology and behavior of these interesting chromosomal anomalies. A. HOMOGENEOUSLY-STAINING REGIONS(HSRs)

HSRs are detected in G-banded preparations as expanded chromosomal regions that do not exhibit the characteristic irregularly spaced dark vertical bands observed in most chromosomes. Instead, the HSRs stain either uniformly lightly, uniformly darkly, or exhibit very fine bands at regular intervals against a background of lighter uniform staining (see Fig. 1A). HSRs have been observed in tumor cells of hamster, mouse, rat, and human

FIG. 1. Cytological manifestations of sequence amplification. (A) Homogeneously staining regions. Methotrexate-resistant Chinese hamster lung fibroblasts (MQ19, Biedler and Spengler, 1976a) were subjected to colcemid treatment. the mitotic chromosomes were spread on microslides, and were stained by the standard G-banding protocol. The HSR is located on chromosome 2 in this cell line and is indicated by a bracket. The normal 2 homolog is indicated with an m w . Note the relatively uniform, intermediate staining of the HSR when compared to other chromosomal regions. (Picture courtesy of J. L. Biedler.) (B) Double minutes. A G-banded preparation of the human neuroendocrine tumor cell line, Colo 320DM (George and Francke, 1980). Note the relatively uniform size of the lightly staining extrachromosomal bodies (double minutes). Note also that a few of these elements appear to be single. (Picture courtesy of D. George.)


47

origin (see previous section), and can range in size from barely detectable to 1520% of the condensed mitotic genome length. There can be one or more than one HSR per cell, often located terminally on chromosomes. In a given stable cell line, virtually every cell in the population displays the same number and position of HSRs. In drug-resistant cells, it can be shown that HSRs often reside on the chromosome in which the original unamplified gene is located. Alternatively, they can be located on other identifiable chromosomes. In some cases, the HSRs are located on unidentified marker chromosomes that could arise by breakage of an elongated HSR from its original site, with subsequent attainment of a centromere. In a MTX-resistant CHO cell line derived by Chasin and associates, the amplified DHFR genes are located on the long arm of chromosome 2 (Nunberg et al., 1978). The DHFR gene in parental drug-sensitive CHO cells has also been shown to reside on chromosome 2 in cell fusionlchromosome mapping studies (Roberts er al., 1980). Flintoff et al. (1982) observed HSRs on chromosomes 2 2 (a rearranged 2 ) , 5, and Z5a in a series of MTX-resistant CHO cells developed in their laboratory. In some of these highly resistant cell lines with as many as 75 copies of the DHFR gene, no obvious HSRs were detected, although one line had suffered a rearrangement of chromosome 2 . Biedler and colleagues have extensively characterized a large number of independently isolated Chinese hamster lung fibroblasts that are highly resistant to MTX. Each of these lines displays a single HSR whose length varies approximately in proportion to the level of drug resistance in each line (Biedler et a l ., 1980). While the HSRs in these cell lines can be located at a variety of positions (e.g., chromosomes 2 , 4, 9, and unidentified marker chromosomes), they are often located terminally on the long arm of chromosome 2 , but not necessarily next to the same G-band on the original 2q in each case. These investigators also examined several cell lines with low DHFR gene copy numbers that exhibited highly rearranged, abnormally banded chromosomes (often chromosome 2 ) (Biedler et al., 1980). In some cell lines, small uniformly stained regions are interspersed with segments that band normally, and in situ hybridization with DHFR probes results in separated clusters of grains on a single chromosome. Biedler and colleagues have interpreted their results to mean that amplification of the DHFR gene may occur extrachromosonially in the initial stages, and that the extra copies then insert randomly into chromosomes (and often back into chromosome 2 ) . However, most of the abnormal chromosomes pictured in this study could also be explained by supposing that amplification occurs in situ at the original location of the DHFR gene in chromosome 2 , but that the resulting tandem sequences are unstable and often break at random positions within the array. The two free ends could then provoke translocations and intrachroniosomal recombinations, leading to dispersion of the amplified sequences and the resulting complex karyotypes observed in these cell lines.

48

JOYCE L. HAMLlN ET AL.

A highly MTX-resistant CHO line derived by Hamlin and colleagues contains two detectable HSRs, neither of which is on chromosome 2 (Milbrandt et al., 1981). Instead, studies on a cloned series of increasingly resistant cell lines from which this line was derived show that even at the lowest level of resistance, the original HSR was located on the long arm of chromosome 1 (J. L. Hamlin, unpublished observations). As amplification increased, this original HSR lengthened and eventually fragmented to yield a second HSR-bearing marker chromosome, which itself lengthens with increasing resistance. In addition, in situ hybridization studies with a cloned DHFR genomic fragment detects a third site of amplified genes on chromosome 2 4 that is undetectable as an HSR at any drug level. In those MTX-resistant murine lines that exhibit HSRs (as opposed to DMs), the amplified DHFR genes are sometimes on chromosome 2, which may be the site of the parental gene (Dolnick er al., 1979). The amplified DHFR genes in human cells have been observed on chromosomes 4 , 5 , 6 , 10, and 19 (Wolman et al., 1983; Trent et al., 1984; Srimatkandada et al., 1983). The site of the endogenous gene in normal, MTX-sensitive human cells is presently not known. Among the various neoplasms that display HSRs, there is as yet no consistent correlation between the chromosomal positions of HSRs and the type of tumor in which they are observed (e.g., see Gilbert and Balaban, 1982). However, it is possible that the same oncogene is amplified in a given class of tumors, but the resulting HSRs are subjected to translocations and rearrangements that mask the original chromosomal location of the oncogene. Balaban-Malenbaum and Gilbert (1982) have also made the interesting observation that an HSR occurs at chromosomal position Ip34 in both a human retinoblastoma (Y79) and in a neuroblastoma (IMR32), suggesting that the same sequence (oncogene?) may be amplified in both cases. The expanded chromosomal regions originally detected by Biedler and Spengler (1976a) in mitotic preparations of MTX-resistant Chinese hamster lung cells and in certain neuroblastomas were termed HSRs because of their uniform (unstriated), intermediate staining with Giemsa. However, a variety of HSRs have now been described that deviate from this euchromatic appearance. In PALA-resistant Syrian hamster cells, Wahl and co-workers have shown that the amplified sequence includes not only the gene coding for the CAD protein complex, but ribosomal DNA as well (Wahl et al., 1983). Most of the amplified sequences are located at the terminus of chromosome 9, which was also shown to be the location of some of the rDNA copies in parental Syrian hamster cells. The expanded chromosomal region on chromosome 9 is characterized by finely apposed dark bands. In rat erythroleukemia (Murao e? al., 1982), H4 hepatoma (Tantravahi el al., 198 l ) , and XC sarcoma (Tantravahi et al., 1982) cell lines, rDNA genes can be extensively amplified, and are observed as HSRs or finely banding regions at known nucleolar organizer regions in this species. Regularly


49

spaced alternating light and dark bands have also been observed in a variety of other cell lines, including human neuroblastomas and melanomas (see Cowell, 1982). As pointed out by Cowell, the appearance of expanded chromosomal regions may be critically dependent on the fine points of the staining technique. Indeed, the HSRs in different stained preparations of the same MTX-resistant CHO cell can appear either finely banded or homogeneously euchromatic (Milbrandt et al., 1981). At the present time, very little is known about the composition of amplified sequences in most cell lines. For example, it is not known how many of the HSRs in different cell lines actually do contain ribosomal DNA, satellite DNA, or other repetitive sequences that could confer the finely banding property often observed. The critical observation is that HSRs appear unusual in Giemsa or Qbanding studies because they exhibit either uniform staining or regularly spaced bands, and can usually be detected even by the untrained eye. Since many of the amplified sequences are now being cloned and characterized with respect to sequence composition, it will soon become possible to relate DNA sequence arrangement to chromatin banding properties. This is an especially promising area of cytogenetics that has been difficult to study on single copy genes in the past. Regardless of cell type, HSRs are relatively stable entities that behave in most respects like typical chromosomal segments. Biedler and colleagues have shown that the amplified DHFR genes that reside in HSRs in some MTX-resistant Chinese hamster lung cell lines can be maintained for years in the absence of MTX selection (Biedler el al., 1983). However, in most cell lines, a gradual decline in DHFR activity was observed over the period of several years in culture in the absence of drug, accompanied by a gradual decline in the length of the HSRs. One cell line that exhibits a prominent HSR on chromosome 2 was observed to lose more than 90% of DHFR activity within about 50 cell doublings (Biedler et nl., 1983). Thus, the stability of HSRs can vary, although the mechanism for such variation is unknown. In general, amplified drug resistance markers contained in HSRs are much more stable than those camed on DMs, which can be completely lost from the population in 20-30 cell doublings (e.g., Biedler et nl., 1983; Brown et al., 1981). HSRs seem to replicate by the same mechanisms that govern replication of the remainder of the karyotype, and since they are carried on chromosomes with legitimate centromeres, the daughter HSRs are distributed equally to the two daughter cells at mitosis. The HSRs that contain amplified DHFR genes have been shown to be early replicating in Chinese hamster cells (Harnlin and Biedler, 1981; Milbrandt et al., 1981) and in one MTX-resistant murine cell line (Kellems er al., 1982). In addition, the HSRs in several human neuroblastomas appear to replicate early in the S period (Biedler and Spengler, 1976b). In MTX-

50


resistant CHO cells, it has been shown that each unit repeated sequence (amplicon) contains a functional origin of DNA synthesis (Heintz and Hamlin, 1982), with the result that initiation of DNA synthesis at the beginning of S begins synchronously at multiple loci within the HSR (Milbrandt et al., 1981). Since there is probably only one origin within each unit, an amplicon may be equivalent to a replicon. It follows that amplification may proceed via the agency of replicons. This idea will be discussed in Section V1. It has been suggested that all expressed genes are replicated early during the S period, since early replication has long been correlated with euchromatic (presumably active) elements in mammalian chromosomes (see Bostock and Sumner, 1978, for review). However, Hamlin and Biedler (1981) have shown that the HSRs in two MTX-resistant Chinese hamster lung cell lines are early replicating, even though these HSRs are clearly C-band positive. These HSRs therefore probably contain at least some constitutive heterochromatin, which is normally late replicating (Bostock and Sumner, 1978). Interestingly, in Chinese hamster cells that have been transfected with a cloned DHFR gene (J. L. Hamlin, unpublished observations) or with cloned DHFR cDNA (Kaufman et al., 1983), and subsequently have been subjected to incremental increases in MTX concentration, the resulting HSRs have also been shown to be early replicating. This finding adds strength to the argument that expressed genes are always early replicating, but raises questions as to how the transfected gene becomes subject to this control in a new chromosomal location. It is likely that the cloned DHFR sequences have integrated next to an earlyreplicating origin in a new chromosomal location, possibly because euchromatin is a better substrate for recombination than is late-replicating heterochromatin. Alternatively, sequences that integrate into heterochromatin may not be expressed. It is also possible that early-replicating transformants are preferentially selected, since the gene is normally transcribed in late G,/early S (Mariani et al., 19811, and may require early replication for its expression. In any case, studies on these transformants should shed light on the important questions of amplification itself and time-ordered DNA synthesis in mammalian cells. B. DOUBLEMINUTES (DMs) Typically, double minutes are seen in metaphase spreads as small, paired chromatin bodies that usually stain poorly with G, Q, C, R, and Cd banding protocols, and are therefore believed to be euchromatic in nature (see Fig. 1B). They have been observed in tumor cells and cell lines of mouse, human, rat, and hamster origin (Barker, 1982). They have been shown to contain amplified DHFR genes in several MTX-resistant murine cell lines (e.g., Kaufman et al., 1979; Martinsson et al., 1982), and their presence has been correlated with amplification of many other genes in drug resistance, among them vincristine


51

(Kuo et al., 1982), metallothionein (Beach et al., 1981), and adenosine deaminase (Yeung er al., 1983a). In addition, they have frequently been observed in a variety of tumor cells in which the amplified material has not as yet been identified (see Barker, 1982, for review). DMs can be distinguished from very small chromosomes by their lack of Cbanding and unusual behavior during mitosis. DMs can range in size from barely detectable at the light microscopy level to the size of the smallest mammalian chromosomes. Large DMs can, in unusual cases, assume the shapes of rods or rings (e.g., Bostock and Clark, 1980). In a given cell, the DMs are usually approximately the same size. Electron microscopic studies on DMs isolated by sucrose gradient fractionation support the concept that the chromatin structure of DMs is closely related to that of normal chromosomes, with the exception that they lack centromeres (Barker and Stubblefield, 1979). DMs vary in number in different cell lines, or even within a single clonal population. They are sometimes observed only in a small subpopulation of a clonal cell line (e.g., Baskin et al., 1981; Levan et al., 1977). In some cell lines, there may be only one or two DMs per cell, and in others, as many as a thousand per metaphase spread (Cowell, 1982). In the latter case, the DMs are very small. In the case of MTX resistance, it has been possible to relate the number of DMs in murine cell lines with different levels of drug resistance (Brown et al., 1981). In this case, there appears to be a rough correlation between the number of DMs and the gene copy number, and the lines seem to maintain a relatively steady-state level of DMs at a given MTX concentration. DMs lack centromeres, and distribute themselves to daughter cells during mitosis by attaching themselves randomly (and usually in groups) to the ends of other chromosomes at the metaphase plate during chromosome segregation (Levan et al., 1976). They are thus carried along adventitiously to daughter cells in a random fashion, with the result that unequal distribution can occur. In cases where the DMs offer a selective advantage to the cell (e.g., drug resistance), the presence of the selective agent would tend to continuously select for the daughter cell that had received the most DMs. However, balanced against this process is the frequent loss of DMs, due to a failure to attach to chromosomes during mitosis, leading to encapsidation by the reforming nuclear envelope, and eventual extrusion from the cell (Levan and Levan, 1978). Brown et al. (1981) and Biedler et al. (1980) have shown that more than 90% of the DHFR genes carried on DMs can be lost in a matter of weeks in the absence of the selective agent, MTX. It has also been observed that cells with large numbers of DMs grow more slowly than do related cell lines with smaller numbers (Kaufman et al., 1981). Thus, the number of DMs observed in any steady-state condition (e.g., constant selective pressure) must be a product of all these factors. This suggests that the DMs observed in tumor cells that have not been treated with drugs must somehow confer a selective advantage on the host cell; otherwise,

52


unequal distribution, loss of DMs by enucleation, and slower growth rate would eventually eliminate cells containing DMs. Since DMs are not integrated in a stable fashion into chromosomes, do they replicate, and, if so, how? It is possible that once they accumulate by whatever means (e.g., HSR fragmentation), they do not replicate, but rather are maintained in the population by unequal segregation. Several observations suggest that DMs do indeed replicate once per cell cycle, and apparently during the early S period, These conclusions derive from three types of experiment: (1) bromodeoxyuridine (BUdR) incorporation followed by Hoescht 33258 staining yields the typical harlequin staining pattern characteristic of a single cycle of semiconservative replication during one S period, and shows that each half of a double minute represents the counterpart of a chromatid (Quinn et al., 1979; Barker and Hsu, 1979); (2) after a brief pulse of [3H]thymidine, autoradiography demonstrates that DMs are labeled only in cells that show early replication patterns on the rest of the chromosomes (Barker et al., 1980); ( 3 ) when premature chromosome condensation is provoked in G , phase cells containing DMs, many single minute structures are also observed; however, in G , phase, double minutes predominate, suggesting that they were replicated during the S phase; once again, the BUdR/Hoescht technique displays the harlequin staining pattern expected after a round of semiconservative replication (Barker et al., 1980). In total, these observations support the concept that DMs behave as minichromosomes with respect to replication, but, as stated before, do not divide at mitosis. Separation of the two chromatids must occur sometime between the end of one S period and the beginning of the next, to account for the observation that DMs do not increase in size in stable cell populations during prolonged culture. However, the occasional failure to separate during one cell cycle could lead to larger DMs, and could explain the observed variability in the size of DMs in some cell types. The question of replication of DMs cannot be completely understood at the present time because the basic structure of DMs is not known. They are thought to be composed of linear or circular tandem arrays of amplicons. By analogy to HSRs, each amplicon in a DM could contain an origin of replication. DNA synthesis could then proceed by a mechanism analogous to bidirectional chromosomal replication from multiple origins, as has been suggested for the amplicons contained in HSRs (Heintz and Hamlin, 1982).

C. RELATIONSHIP BETWEEN DMs

AND

HSRs

If DMs and HSRs are different manifestations of the same phenomenon (amplification), then what determines whether a given amplified sequence will take one form or another? This is a complex question for which there is presently no


53

satisfactory answer. HSRs have been observed in cells of murine, hamster, rat, and human origin. DMs have been observed in murine, human, and rat cells, but rarely in hamster cells. However, DMs are a property of tumor cells, and very few cultured hamster tumor lines are available for study. DMs have been detected in vincristine-resistant Chinese hamster embryo fibroblasts (Kuo et al., 1982). However, in this case, there is some question whether the DMs observed actually contain the gene that imparts resistance to vincristine, since the number of DMs does not decrease when the drug is removed, even though vincristine resistance decreases. However, this result does show that hamster cells are capable of generating and maintaining such structures. It is likely, therefore, that taxonomic differences per se do not explain the propensity of a given cell line to maintain amplified DNA sequences in either intra- or extrachromosomal forms. The question then arises whether the particular amplified locus in a cell line is constrained to assume one form or another by unknown mechanisms related to DNA sequence. There is no clear-cut answer to this question either, primarily because few systematic studies have been performed in which a single cloned parental cell line has been treated with a variety of drugs to isolate variants that have amplified different genes. However, Biedler and colleagues have shown that a near-diploid cloned Chinese hamster lung cell line (DC3F) gives rise to HSRs when subjected to increasing concentrations of either MTX or vincristine (Biedler and Spengler, 1976a; Biedler, 1982). Hence, two different loci manifest the same chromosomal form in this case. It would be of interest to subject the parental DC3F cell line to stepwise increments of other drugs to determine whether any locus in this cell line could give rise to DMs. It is clear that certain cell lines display a propensity toward the formation of either DMs or HSRs. In highly MTX-resistant Chinese hamster cells (ovary and lung), amplified DHFR genes invariably reside on chromosomal HSRs (Biedler and Spengler, 1976a; Nunberg et a/., 1978; Flintoff et a/.., 1982; Milbrandt et al., 1981). Syrian hamster cells selected for PALA resistance exhibit HSRs, and DMs have apparently never been observed in this system (Wahl et a / . , 1983). In cultured MTX-resistant human cells, the amplified DHFR genes are usually detected as HSRs (Trent, 1982; Wolman el a / . , 1983; Srimatkandada el a / . , 1983), but Trent and co-workers have recently reported the occurrence of one or two DMs in a small percentage of cells obtained from a MTX-resistant human ovarian adenocarcinoma (Trent et a / ., 1984). In most MTX-resistant murine lines, the amplified DHFR genes are located on DMs (Kaufman et a / . , 1979). However, after long-term culture of cell lines bearing DMs (e.g., murine S180 cells), stably amplified lines can occasionally be derived in which the genes are located on HSRs (Dolnick et a/., 1979). The observation that the same cell line can give rise to sublines that display either DMs or HSRs therefore argues that the two forms are interconvertible in some instances. There are several examples among neoplasms that support this

54

JOYCE L . HAMLIN ET AL

suggestion. Cell lines derived from the murine Y 1 adrenocortical tumor can have either DMs or HSRs, but not both in the same cell line (George and Franke, 1980). Utilizing clones from an amplified genomic sequence derived from partially purified DMs, George and Powers (1982) have demonstrated that the same sequence is amplified in the HSR of another Y 1 subline. Thus, in this case, the DMs and HSRs are most likely different manifestations of the same amplification phenomenon in the same genetic background. The human neuroendocrine tumor, Colorado 32 1, can also manifest one or the other chromosome abnormality, and sublines can interconvert in culture (Quinn et al., 1979). The RVP-3 mouse tumor displays DMs early after establishment in culture, but these gave rise to microchromosomes after long-term maintainence, some of which C-band and seem to have centromeres (Sainerova and Svoboda, 1981). A provocative finding by Levan and co-workers is the observation that the amplified sequences in the murine tumor line, SEWAIR, exist in alternate states, depending upon whether the cells are cultured as an ascites tumor in vivo or are maintained in tissue culture (Levan and Levan, 1982). The effect of culture conditions points out that subtle cellular differences may determine the propensity of genetic amplifications to assume one or the other form in a given cell type. Since, in the SEWAIR tumor, the HSRs disappear and are replaced by DMs in vivo, the possibility exists that DMs somehow provide a selective advantage to cells in vivo. This suggestion is supported by their finding that subcutaneous injection of a SEWAIR subline displaying DMs provoked tumors in animals, but sublines without DMs did not form tumors (Martinsson, Dahloff, Sandberg, and Levan, unpublished observations). Martinsson et al. (1 982) made the additional very important observation that a given cell can maintain both DMs and HSRs simultaneously, albeit not at the same genetic locus. When the SEWAIR tumor displaying HSRs (amplified oncogenes'?) was subjected to incremental increases in MTX for several months in vitro, multiple DMs developed which ostensibly arose from the amplification of the DHFR gene, but the original HSRs were maintained. This finding argues against the possibility that individual cells in a population can maintain only one of the two configurations due to ambient intracellular conditions. Another approach to understanding the genesis and/or stability of the alternate chromosomal manifestations of gene amplification involves fusion between cell lines. The MTX-resistant murine tumor line, SEWAIR TC13, when cultured in vitro, maintains a mixed but relatively constant proportion of cells showing either DMs (60% of cells) or HSRs (40% of cells) (Jakobsson et al., 1984). However, upon fusion with Chinese hamster V79 cells, all MTX-resistant hybrids exhibited only HSRs. Thus, in this particular combination of cells, DMs do not appear to be transferred and/or maintained in the hybrids. It could be argued that DMs are easily lost at mitosis during the undoubtedly complex


55

adjustments that the hybrids must make to the newly formed karyotype. Hybrids with a large enough number of DMs (therefore DHFR genes) to survive selection in MTX would therefore be very few in number. Alternatively, the milieu contributed by the V79 cell in the hybrid may not be able to support the presence of DMs for unknown reasons, and, thus, only cells with HSRs would be selected in MTX . Contrasted with these results are the observations of Kano-Tanaka et ul. (1982). Mouse neuroblastoma cells showing only DMs were fused with either rat liver, rat glioma, or Chinese hamster brain cells. All hybrids manifested DMs, indicating that under these circumstances, DMs are able to be transferred and maintained in the new hybrid cells, even though in one instance, one partner in the hybrid (the Chinese hamster cell) rarely displays DMs. Another variation involves a human neuroblastoma cell line that displays only HSRs, but when fused to mouse fibroblasts, yields cells displaying only DMs (Balaban-Malenbaum and Gilbert, 1980). The interpretation here would be that the resulting hybrids are not able to support HSRs, or that fragmentation of the HSR occurs during fusion, and for unknown reasons, the hybrid cell is not able to reaggregate and/or integrate the DM sequences into HSRs. The results of all these experiments presently do not allow the formulation of any unifying rules for the maintenance of amplified genes in hybrid cells. Clearly, the DMs or HSRs bearing amplified genes cannot by themselves determine their eventual configuration in hybrid cells. Additional factors related to the new cellular environment created by fusion of two disparate cell types must also be involved. As discussed in an earlier section, cloned amplifiable genes have been transferred to cultured cells, and after stepwise selection with the appropriate drug, the cytological properties of resistant cells have been determined. In these cases, the transfected, amplified genes seem to reside in the same kind of chromosomal structure characteristic of amplified endogenous genes. Both the CAD and DHFR genes are located in stable HSRs after transfection and amplification in hamster cells (Wahl et al., 1983; Milbrandt et al., 1983b). The most telling results were obtained in experiments utilizing a variety of cloned murine DHFR minigenes to transform either murine or Chinese hamster cells. After selection for MTX resistance, murine transformants invariably contained large numbers of DMs (Murray et al., 1983), whereas MTX-resistant Chinese hamster transformants displayed only HSRs (Kaufman et al., 1983; Gasser ef nl., 1982). Either the sequences flanking the site of integration must somehow determine the subsequent chromosomal state of the amplified material, or the cell line itself determines it. By the former argument, it might have been expected that independent transformants of a given recipient cell type would have displayed both alternative chromosomal manifestations, depending on the sequences surrounding the site of integration. However, all independent isolates of a given cell line (e.g., mouse)

56


consistently displayed the same abnormality. It is more likely that these studies reflect the general observation that cultured Chinese and Syrian hamster cells most often display HSRs as a corrollary to drug resistance and gene arnplification, whereas the formation of DMs is the more usual manifestation of gene amplification in murine cell lines. Schimke and collaborators have attempted to understand the process of gene amplification in greater detail by examining the early stages of amplification (i.e., single step selections in low levels of drug) and the characteristics of loss and fixation of genes as drug resistance is lost or gained. They have utilized a fluorescent MTX derivative that binds tightly to DHFR as an indirect indicator of the number of DHFR genes in a cell. Because the bound MTX fluoresces, cells incubated with the compound can be individually analyzed and/or separated on a fluorescence-activated cell sorter. In addition, their results, in most cases, have been confirmed by determining gene copy number in sorted cells. They first studied MTX-resistant murine lines derived from 3T6 cells that display numerous DMs. When cultured in the absence of MTX, these cell lines lost DHFR genes (fluorescence) with kinetics identical to the loss of DMs, as determined cytologically (Brown et izl., 1981). These results were confirmed by determining the number of DHFR genes in revertant cell lines by Southern blot analysis, using a radiolabeled cloned DHFR cDNA. These studies thus support the notion that the amplified DHFR genes that impart MTX resistance are located on DMs, and the DMs are unstable entities that are frequently lost or unequally distributed to cells during mitosis. These suggestions were confirmed in studies on DM-containing MTX-resistant murine S180 cells (Kaufman et al., 1981). When grown in the absence of MTX for many doublings, cells with progressively fewer DMs became dominant in the population. A most interesting result was obtained when the development of MTX resistance was studied in CHO cells, which invariably display stable HSRs at high levels of drug resistance after long-term culture. When cells were selected at a low drug level and were subsequently analyzed after relatively short intervals (e.g., 2 weeks), a heterodisperse population with variable DHFR gene copy numbers was observed (Kaufman and Schimke, 1981). Moreover, if sorted cells with a given number of DHFR genes were subsequently cloned and were grown in the absence of MTX for about 20 cell doublings, each clone behaved somewhat differently, indicating clonal variation in the stability of the amplified sequences. The progeny of some clones maintained the original gene copy number in the absence of drug, while the progeny of other clones lost all or a large fraction of the amplified genes. Other clones actually gave rise to cells that had amplified the DHFR gene still further, even in the absence of MTX. DMs were apparently not observed in these cells under any of the above manipulations. These data indicate that shortly after amplification, the initial extra DHFR gene copies need not be stably integrated into the chromosome, even in CHO


57

cells. However, it could be argued that only those copies that are stably integrated in the initial event give rise to the stable HSRs observed in highly resistant sublines obtained after incremental MTX selection and long-term culture. Furthermore, since cells can amplify the gene in the absence of drug selection, amplification must be a random event, and administration of drug must select those cells that contain enough DHFR genes to support growth at a particular drug level. In all cases where CHO cells were grown at a given drug concentration for 100 doublings or more, the amplified genes were shown to be maintained stably as HSRs after removal of drug for prolonged periods of time. This suggests that the extrachromosomal copies observed at low drug levels are lost from the population, or are eventually integrated in tandem into the chromosome. Since the initial DHFR gene amplification events observed in both mouse and Chinese hamster cells can be unstable, there must be inherent differences in the ability of these cell lines (S 180 and CHO) to fix and maintain the amplified genes as HSRs after long-term culture in MTX. It may be that at all stages of amplification, CHO cells integrate the duplicated segments covalently into the DNA fiber more efficiently than do the murine cell lines, owing to differences in the structural or functional organization of chromosomes. Alternatively, the murine lines may be more efficient at excising duplicated segments through homologous recombination events, leading to extrachromosomal DMs. The few reported cases of MTX-resistant murine cell lines that bear amplified DHFR genes on HSRs would then presumably represent instances in which the recombination process has somehow been suppressed. This argument could also explain the fact that individual sublines originally derived from a single cell (e.g., the murine Y 1 adrenocortical tumor cells) can display either HSRs or DMs (George and Franke, 1980).

IV. Nature of Amplified Sequences In order to understand the mechanisms involved in the amplification of DNA sequences, it will be necessary to define the nature of the amplified unit. One would like to know the answers to the following questions: 1. Are the multiple copies of an amplified sequence arrayed tandemly in a linear fashion, or do they lie side-by-side in an onion skin configuration'? 2 . How large are the unit repeated sequences? 3. Are the amplified sequences of equal size in a given cell line, i.e., is the unit of amplification precise or imprecise? At a given locus, are the boundaries of an amplicon fixed by some aspect of the nucleotide sequence that itself determines the mechanism? 4. Does the sequence continue to be amplified with precision, or is it trimmed

58


or rearranged during the process, i.e., do the characteristics of the amplicons observed in highly drug-resistant lines tell us anything about the initial unit of amplification? A . AMPLIFIED ENDOGENOUS GENES

The most compelling evidence that amplification results in linear tandem arrays of a unit repeated sequence (amplicon) is that the HSRs that contain amplified sequences are elongated rather than thickened, and that an HSR has the same diameter as the rest of the chromosomal complement. Furthermore, Biedler and colleagues have shown in a series of highly resistant Chinese hamster cell lines that the length of the HSR is roughly proportional to the level of drug resistance in each line (Biedler et al., 1980). This phenomenon has also been observed in a cloned series of increasingly MTX-resistant CHO cells by Hamlin and colleagues (unpublished observations). Evidence for a linear tandem arrangement of amplicons is less clear for the DNA in double minutes, since they can appear to be large or small in diameter, and since there is some question as to whether they are circular rather than linear (e.g., Bostock and Clark, 1980). However, since MTX-resistant mouse cells display HSRs and DMs as alternative forms of DHFR gene amplification (Kaufman et a/., 1979), it is reasonable to assume that the amplicons are arranged tandemly in both situations. In addition, in their studies on the nature of the nucleotide sequences in the CAD amplicon, Stark and co-workers have found unique junction fragments that are predicted by end-to-end joining of amplicons (see Ardeshir et a / . , 1983, and discussion below). Original estimates for the size of amplicons in several cell lines were made by determining the length of HSRs, and by determining the number of amplified genes in a given cell line. By assuming a figure for the number of residues in a given length of HSR, Nunberg et a / . , (1978) estimated that the size of the amplicon in a MTX-resistant CHO cell was 500-1000 kb in length. Using the same strategy, Bostock and Clark (1980) estimated that the amplicon could be as large as 3000 kb in a MTX-resistant mouse PG193T lymphoma cell line. The observations of Milbrandt et al. (1981) suggest that in some cell lines, the amplicon can be considerably smaller. The restriction fragments derived from the amplicons in highly MTX-resistant CHO cells can be visualized on ethidium bromide-stained agarose gels, and by summing the lengths of all amplified fragments, they estimated that the unit repeated sequence was approximately I35 kb in length. Since there are 1000 copies of the amplicon in this cell line, a total of 1.35x108 bp, or about 4% of the genome length, represents amplified sequences. However, the length of the HSRs in mitotic spreads totals about 8% of the condensed genome length. This discrepancy might be explained if it is assumed that the restriction fragments that can be visualized on gels represent


59

only a consensus sequence that is amplified in all amplicons, but that is flanked by more or less DNA in each repeated unit. The 135 kb estimate would then be a lower estimate for the size of the repeating unit. Alternatively, estimates for copy number and length of HSRs may be inaccurate enough to account for the twofold discrepancy. Bostock and Tyler-Smith ( 1982) studied MTX-resistant murine EL4 lymphoma cell lines that were cloned from a resistant population selected in a single, low concentration of MTX. After being subjected to increasing drug levels, the amplicons in the highly resistant derivatives were apparently very similar to one another, and appeared to be about 500 kb in length, as determined by direct visualization of amplified restriction fragments in ethidium bromide-stained agarose gels. This result suggests that in this system, the original amplified unit is maintained during amplifications to higher copy number. Furthermore, Bostock and Tyler-Smith used isolated DMs carrying DHFR genes to transfer MTX resistance to sensitive mouse L cells. After incremental increases in MTX to raise the level of resistance, the amplicons were examined in the derivatives, and were found to contain the same sequences as the amplicon in the donor cell. Since a single DM is apparently transferred in these experiments, Bostock and Tyler-Smith argue that every DM in the EL4 lymphoma must have the same array of amplicons. In order to understand more about the structure of amplified sequences contained in DMs or HSRs, it is obviously necessary to isolate the sequences in question. Molecular cloning of amplified sequences is greatly simplified owing to the much larger number of these sequences per diploid nucleus. The practical result of this is that many fewer bacterial clones from libraries containing recombinant genomic fragment have to be screened in order to isolate overlapping clones spanning several equivalents of the amplicon. Schimke and collaborators have isolated more than 200 kb of the DHFR amplicon from the MTX-resistant cell line, S180, by a procedure that involved isolation of a chromosome fraction enriched in double minutes on sucrose gradients (Schilling et af., 1982). By using cloned DHFR cDNA sequences as radioactive probes, they initially cloned fragments from the DHFR gene itself, and then used the endmost fragments from these clones to “walk” to the right and left of the gene. They have used these overlapping clones as radioactive hybridization probes on restriction digests of other MTX-resistant murine cell lines to ask whether the amplicon is the same or different in independently isolated cell lines. They found that these probes cross-hybridized with different lengths of DNA sequence in each cell line, ranging from 80 kb in a murine 3T6 cell line displaying double minutes, to as large as 200 kb in the S 180 cell line itself. This result could indicate that the actual size of DHFR amplicons in different murine cell lines can be different. However, many of these MTXresistant murine lines derive from different parental lines that have been main-

60


tained in culture for many years, and many of these cells are karyotypically unstable. It is therefore possible that the parental sequences flanking the DHFR gene are not identical, due to deletions and/or rearrangements, and, hence, cross-hydridization is not detected between the divergent sequences. A more convincing case for variable amplicon size arising from the same genetic background could be made if cell lines derived from exactly the same parental cell would be shown to have variable sized amplicons. Hamlin and colleagues have cloned approximately 110 kb of contiguous DNA sequence from the amplicon of a MTX-resistant CHO cell line, by first isolating the DHFR gene itself, followed by chromosomal walking (Milbrandt et al., 1983a). These clones have been used to probe independently isolated MTXresistant cell lines of the same or closely related [Chinese hamster lung (CHL)] parentage (Montoya-Zavala and Hamlin, unpublished observations). Both CHO and CHL cells are extremely stable karyotypically, and the CHL cells have a near diploid chromosome complement (Biedler and Spengler, 1976a). The cloned probes from the MTX-resistant CHO cell line cross-hybridize to virtually identical restriction fragments in the DNA of three other MTX-resistant cell lines (two MTX-resistant CHL lines and another MTX-resistant CHO cell). The major differences so far detected between cell lines can be accounted for by simple restriction site gains or losses, but no major deletions or insertions have been observed. It could be argued that sequences more distant from the DHFR gene that have not been cloned as yet will detect different sized amplicons in each cell line. However, direct visualization of the amplified restriction fragments in ethidium bromide-stained agarose gels indicates that the amplicons in all four of these cell lines are similar in size and composition (Hamlin et al., 1982). Together, these findings suggest that two resistant cell lines with exactly the same parentage (line DC3F in the case of two CHL cells, and CHO K1 in the case of the two CHO cells) are constrained to amplify the same sequence, if the parental cell lines are karyotypically stable. Stark and colleagues have prepared libraries of genomic clones from two different PALA-resistant Syrian hamster cell lines (Acdeshir et al. 1983). Clones derived from the CAD amplicons were identified by using total genomic DNA from PALA-resistant cells as radioactive probes (after removing nonspecific, highly repeated sequence elements). These clones represent 162 and 68 kb of each amplicon, although most of the fragments have not been ordered or shown to overlap with one another as yet, and together probably represent a small part of the CAD amplicon in each cell line (estimated to be approximately 500 kb in length). Nevertheless, they have found several fragments in each library that appear to represent junctions fragments between amplicons. From this result, they argue that amplification proceeds by an imperfect mechanism, and that the amplicons within a given cell line end at different places, but usually straddle the CAD gene (see Fig. 2). This argument is supported by their finding that different

ABCOEFGHIJKLMNOP~qRSTUWWXYZ

AB CDE FGH I JKL( NNOPQ RS ) [ MNOPORS) TUUU XY 2

1

A I

4

AB CDE FGH I JKL IMNOPqRS ) ( HNOPQRS ) TUVW XY 2

further amp1 i f i c a t i o n

ABCDEFGH! JKL(f'WIPQRS)nTUVWXY Z

or -

duplication

ABCDEFWI JKL(MN0PqRS) (MNOPQIU ) TUYW X Y Z

+ 4

further

amplification

i+

ABCDE FGH I JKLH( NOPPRST) ( NOPQRST1 uvw X Y z

or

ABCDE FGHI JKL MNOPQRS ) ( PQRS ) ( MNOPQRS ) [ NOPQR) TUWXYz

duplication

ABCDEFGHIJK( LMElOP R)(LMNOPQR)SfUWXYZ

5,

1

J.

further amplification

ABCDEFGHIJK( LMNOPI?RI,,STUWYZ

FIG.2. Different mcdcs uf hrquence amplificaticin. The linear sequence of a chromosomal DNA Lher is represented by thc lcttcrs of the alphdbet. whcrc P might reprcscnt a SclCctdhk gcnc. In the precisc mechanism illustrated ahove (1 in figure), the boundaries of [he amplified unit are p r e d e t e m n e d hy sonic aspcct (if the parcntal DNA eequence ( c . g unit of replication, or Iocatiun of highly rcwated e q u e n c e elements), so that all cells derived from the samc parental ccIl must amplify MNOFQRS in the first duplication step. and in all succeeding amplifications at this locus. Only one new junction fragment. S M . I!. formed in the unpindl duplication. and its copy nurnbcr is equivalent to the n u m k r of repeatcd units in all subscquent steps. In the rnechanicm outlined in verf~on2. (he unit of amplification is not fixcd, and the initial duplicakd sequence can be positioned diffcrcntly m u n d P in dlfferent ceH lincs dcrived from thc same parent. However, in subsequent amplificaticm, the duplicated sequencc i s amplified faithfully. Different junction fragments are therefurc amplified in individual cell lincs dcrived fmm the varne parental cell line. In the impcrfcct mechanism rmtlincd in verhion 3. the initial duplicatcd scqucncr can hc positioned differently around P in different cells from the came parent (as in mechanism Z), and in huhsequent amplificatiwns. diffcrcnt sequences arc ampllficd each time. In this case, w m e fragments are prescnt more oflen than othcrs. and severdl uniquc junction fragmunts rcsull fe.g.. LIB. F A . HI) whose copy number is less than thc copy numbcr of P.

.

62


fragments are amplified to different extents in a single PALA-resistant cell line. This is an important result that is somewhat disturbing, since it indicates that determining the exact nature of the amplified unit may be difficult, if not impossible. By using these clones as probes on restriction digests of genomic DNA, they also showed that no two independently isolated, highly PALA-resistant clones had exactly the same amplified sequences flanking the gene, although a consensus sequence of about 44 kb (including the CAD gene) was amplified in all cell lines (Ardeshir el al., 1983). This result suggests that major rearrangements have occurred during the amplification process to account for the fact that the CAD gene is flanked by different sequences in each cell line. It is important to point out that all of the clones examined derived from two different parental cell lines (a cloned derivative of BHK 21/13 and an SV40-transformed derivative of this cell line), both of which were heteroploid to begin with, indicating inherent chromosomal instability. Thus, it is possible that some of the DNA sequence rearrangements observed are not involved in the amplification process per se, and could be masking a more simple underlying mechanism. When the parental BHK cell line was subjected to a single-step selection in a low level of PALA, most of the resulting PALA-resistant clones amplified the same large sequence (about 68 kb in length) (Zieg et al., 1983). This result agrees with the observations of Bostock and Tyler-Smith ( 1 982) on MTX-resistant mouse lines that were subjected to single-step selections in drug. However, Zieg el al. also observed that different fragments were amplified to different extents in a given PALAresistant cell line, suggesting that the units of amplification are not all identical in length. George and Powers (1981) have also cloned fragments from the amplicon in the mouse adrenocarcinoma, Y 1 . In this case, random fragments cloned from a fraction enriched in DMs were tested for their presence in the amplicon by assessing their genomic copy number on Southern blots or by in situ hybridization to related Y 1 cell lines displaying HSRs (1982). Several clones behaved in both tests as if they were amplified. These findings confirm their suggestion that the DMs and HSRs observed in different Y 1 sublines are alternate forms of the same amplified sequence. Most recently, Kanda et al. (1983) have isolated a large HSR-bearing chromosome from the 1MR32 human neuroblastoma by fluorescence-activated flow sorting of mitotic chromosomes stained with the fluorescent dye, 33258 Hoescht. This approach is feasible because the HSR resides on the largest chromosome in human cells (chromosome l ) , and the length is greatly increased by the expanded HSR. The HSR in this cell line was apparently translocated from its original position on chromosome 2. Genomic DNA fragments from this sorted fraction were cloned into A phage, and approximately 20% of the recombinant clones were shown by Southern blotting and in situ hybridization to derive from the


63

HSR in the neuroblastoma. These workers have so far isolated approximately 40 kb of noncontiguous DNA from the amplicon, and one of these clones was found to be amplified in several other neuroblastoma cell lines. Genomic DNAs from all these cell lines, including IMR32, were then screened with a v-myc probe by Southern blotting procedures, and the probe was found to have weak homology to a 2.0 kb EcoRI restriction fragment in all cell lines. This 2.0 kb genomic DNA fragment was then cloned, and was shown to be amplified in these and several other neuroblastomas. Moreover, the EcoRI fragment was shown by somatic cell hybridization/Southern blotting techniques to map to chromosome 2, as did all the cloned probes obtained from the HSR in 1MR32 (Kohl et al., 1983). These studies present compelling evidence that independent neuroblastomas may arise by the amplification of similar cellular sequences that include genes related to v-myc. B. AMPLIFIED TRANSFECTED GENES In all cell lines in which transfected genes such as CAD, DHFR, or TK have been amplified, the unit of amplification is much larger than the gene itself. Milbrandt et al. (1983b) showed that the amplified DHFR gene was flanked by approximately 90 kb of additional (presumably genomic) DNA sequence in transfected MTX-resistant CHO cells. After amplification of the transfected CAD gene in CHO cells, de Saint Vincent et al. (1981) showed that flanking DNA was included in the amplicon, although it was not demonstrated how much additional DNA was amplified in this case. When DHFR minigenes are amplified after transfection into either mouse or Chinese hamster cells, it is also clear in most instances that DNA in addition to the cloned gene is amplified (Crouse et al., 1983; Kaufman and Sharp, 1982; Murray et al., 1983). This result was obtained in the case of minigenes regardless of whether or not carrier DNA was included in the CaPO, precipitation step. Thus, in some cases, the amplified extra DNA could represent sequences flanking the site of insertion into the genome, and in others, the extra sequences could come from carrier DNA that was ligated to the DHFR genes prior to integration into the chromosomes or aggregation into DMs. It therefore appears that a cloned amplifiable gene is probably not capable of amplifying itself in situ in the absence of other genomic sequences. These other sequences could contain origins of replication or repetitive sequence elements that promote high frequency recombination. In none of the above studies is it possible to study the nature of the flanking amplified DNA without actually cloning the entire amplicons in each case. Roberts and Axel (1982) have attempted to obviate this problem by inducing amplification of defined sequences. By utilizing a pool of 20 cloned human genomic DNA sequences as carrier for a chimaeric plasmid containing the APRT

64


gene and a promoter-less thymidine kinase gene, they transfected mouse APRT - / TK- L cells, and selected APRT+/TK- transformants. Each cell line had integrated the chimaeric plasmid along with variable numbers of the carrier clones. From these cell lines, they isolated TK+ revertants that had amplified the chimaeric plasmid and flanking DNA, and therefore overexpressed the partially functional TK gene. By utilizing radiolabeled plasmid (pBR322) DNA to probe digests of genomic DNA from these revertants, they were able to examine the arrangements of the chimaeric plasmid and the carrier clones in the amplicons (Roberts et al., 1983). They observed that the amplified DNA consisted of at least 20 repeating units that ranged in length from 40 to 200 kb, depending on the cell line. Furthermore, by examining cloned DNA fragments from the amplicons, they were able to show that the units were contiguous, and were joined to each other apparently by recombination between homologous repetitive elements in each recombinant clone (often between the pBR322 elements themselves). In a given cell line, the amplicons varied in size, as evidenced by the fact that some fragments are present in larger copy numbers than others. Their data suggest that the chimaeric plasmid integrated into a chromosome along with a concatamer of the carrier plasmids, and that amplification occurred as a result of multiple rounds of replication of this unit. Subsequent homologous recombination events then linked the extra units together and to the chromosome. No genomic DNA flanking the site of the original insertion event seems to be included in the amplicon. Roberts et al. suggest the interesting possibility that one of the carrier plasmids contains an origin of DNA synthesis that is responsible for amplification in this system.

V. Agents That Increase the Frequency of Amplification From the large number of amplified genetic loci that have so far been observed in mammalian cells, it would appear that almost any locus can be amplified, and suggests that it could be an unprovoked, random event. Howeve1 many of the agents used to select for amplification are drugs that interfere with DNA metabolism (e.g., MTX, and hydroxyurea), and could, in fact, directly interfere with DNA replication, recombination, or repair processes whose malfunctiw could be responsible for amplification. ,

A. AGENTSTHATINTERFEREWITH DNA METABOLISM Schimke and colleagues have tested a variety of agents for their ability to increase the initial rate of amplification of the DHFR gene in mouse 3T6 ai. CHO cells (Tlsty et al., 1982; Brown et al., 1983). The experimental protocoi involved pretreatment with hydroxyurea, cytosine arabinoside, or MTX (metabolic inhibitors of DNA synthesis), UV light or carcinogens (e.g., N-acetoxy-N-


65

acetoylaminofluorene), and 12-0-tetradecanoyl-phorbol13-acetate (TPA, a tumor promoter). The latter agent was used alone or in combination with the other listed agents. Cells were exposed to the agent at several levels and for various time periods, and were then allowed a recovery time interval in the absence of agent. They were then challenged with various concentrations of the selective drug, MTX, and plating efficiencies were determined. Alternatively, in several experiments, DHFR enzyme levels in individual cells in the resulting populations were determined using the fluorescent MTX derivative and the fluorescenceactivated cell sorter. In some cases, the DNA of the resulting MTX-resistant cloned cell lines was analyzed by quantitive Southern blotting for DHFR gene copy number. The results of these experiments can be summarized as follows: (1) all of the above agents tested in pretreatment regimens caused marked enhancement of the frequency of MTX-resistant colonies subsequently selected (in some cases as much as 1000-fold); in fact, in recent studies, this group has been able to induce amplification in a majority of cells in the population (R. T. Schimke, personal communication); ( 2 ) in general, the effect of the pretreatment was greatest at high selective concentrations of MTX; (3) the length of the recovery time after pretreatment can determine the frequency of MTX-resistant colonies subsequently selected, implying that the inductive effect of the agent can be repaired in some cases; (4) pretreatment with the tumor promoter, TPA, by itself was not able to increase the frequency of occurrence of MTX-resistant colonies; however, in combination with UV light, hydroxyurea, or MTX itslef, TPA markedly increased the frequency of MTX-resistant colonies; (5) after all such treatments, many of the MTX-resistant colonies were shown not to have amplified the DHFR gene, and therefore must have sustained other mutations such as decreased transport of MTX or changes in MTX affinity in the enzyme itself; however, the proportion due to amplification per se increases when higher MTX concentrations were used during selection. As pointed out by Brown et al. (1983), there are several variables in these experiments that are difficult to adequately control (e.g., plating efficiencies due to pretreatment, metabolic coupling, clonal variation, etc.), but there appears to be no doubt that agents which inhibit DNA replication and/or damage DNA markedly enhance the frequency of DHFR gene amplification under these experimental conditions. These studies are supported by the observations of Lavi (198I ) on the carcinogen-mediated amplification of integrated SV40 virus in Chinese hamster embryo cells. After exposure to a variety of carcinogens, including 7,12-dimethylbenz(a)anthracene, a heterogeneous collection of extrachromosomal DNA fragments was observed, each of which contains part of, but usually not all of, the SV40 genome. She found that a functional origin of replication was required for amplification in this system, suggesting that the process is somehow mediated through the normal origin of the virus. Another possible example of induced amplification is the observation that N -

66


methyl-N’-nitro-N’-nitrosoguanidine provokes resistance to mycophenolic acid in Chinese hamster V79 cells by the overproduction of IMP dehydrogenase (Huberman e t a / ., 1981). While it is possible that a regulatory locus has been mutagenized in this instance, causing uncontrolled or constitutive expression of this gene, it is also possible that the IMP dyhydrogenase gene has been amplified. Hanawalt (1982) has suggested a variety of routes by which agents that interfere with DNA metabolism might provoke or enhance amplification. DNA damage in bacteria stimulates daughter strand gap repair and the SOS repair system (Hanawalt et a / ., 1979). Both systems are error prone, and could cause mutations leading to loss of negative control over chromosomal origins of DNA synthesis. This could result in extra rounds of DNA synthesis (amplification) at a single locus, and would guarantee that unscheduled synthesis would occur over and over again at the same locus. SOS repair in bacteria has also been shown to short-circuit the usual DNA synthetic control mechanisms by causing reinitiation at the legitimate chromosomal origin of DNA synthesis, apparently without actually mutating the origin (Kogoma and Lark, 1975). Tatsumi and Straws (1979) have also shown that SOS repair allows initiation of replication at sequences other than the legitimate origin, resulting in extra rounds of DNA synthesis. In addition, thymidine starvation, as would be induced by MTX (acting on DHFR) or 5-fluorodeoxyuridine (acting on thymidylate synthetase), leads to the accumulation of breaks in DNA (Barclay et al., 1981). Thymidine starvation can also lead to the incorporation of dUTP into DNA, whose subsequent repair often leads to errors and mutation (Hanawalt, 1982). Hence, many of the agents used to select for amplification can actually function as mutagens that lead to loss of control through mutation or that induce unscheduled replication at legitimate or illegitimate sites in chromosomes. However, a host of selective agents have no obvious involvement in DNA metabolism, and probably do not by themselves cause or abet amplification. Among these agents are compactin (for HMG CoA reductase), vincristine (microtubular proteins), and cadmium (metallothionein). However, it could be argued that the intricate control mechanisms that coordinate DNA synthesis with mitosis, doubling of cell mass (including membrane structures), and eventual cell division can be interrupted at several points with a consequent imbalance of DNA synthesis itself. Hence, agents interfering with any of these pathways could enhance the frequency of amplification. Viral integration must also be included in the list of insults that could provoke amplification either directly or indirectly, since several tumor cell lines that display HSRs and DMs are transformed with viruses (see Barker, 1982). It is not known whether the presumptive amplification is made possible by a cellular metabolic change induced by the virus (i.e., transformation), or whether the integrated viral DNA is directly responsible for, and becomes a part of, the amplified unit.


67

B. GROWTH-PROMOTING SUBSTANCES The synergistic role of TPA in the studies of Brown et al. (1983) is unclear, since its action is thought to be at the cell surface as a mitogenic agent. Varshavsky (1981a) has shown that TPA and related nontoxic phorbol ester tumor promoters are all able to increase the frequency of MTX-resistant 3T6 colonies selected in single-step drug treatments by approximately 100-fold when the phorbol ester is present at optimal concentration at the time of MTX selection. However, TPA analogs that are inactive as tumor promoters (e.g., phorbol or phorbol- 12,13,20-triacetate) do not enhance the frequency of MTX-resistant colonies surviving at any selective concentration of MTX. In most colonies, resistance was shown to be due to DHFR gene amplification. Most surprisingly, the hormones insulin, epidermal growth factor, and arginine vasopressin, all of which are mitogenic for 3T6 cells, act in a manner similar to TPA (Barsoum and Varshavsky, 1983). In addition, the effects of TPA and insulin on increasing the frequency of DHFR gene amplification were approximately additive when used together. Varshavsky points out that the mitogenic potential of TPA and hormones could enhance the colony-forming ability of clones that have amplified the DHFR gene, but which would not survive without the mitogen. Another interpretation is that these mitogenic agents put a larger percentage of the population in a metabolic state that is required for amplification (e.g., the DNA synthetic period), implying that replicon misfiring (Varshavsky, 198 1b) or other aberrant DNA replication is responsible for gene amplification. TPA has also been shown to induce the expression and replication (amplification?) of bovine papilloma virus in mouse cells that normally harbor the virus in a nonproductive, nonreplicating state (Amtmann and Sauer, 1982). Whether its action in this system is related to the synergistic effects of TPA described above is not clear. This result could be interpreted to mean that TPA creates a cellular ambience that favors replication of the virus or unscheduled replication of chromosomal origins of DNA synthesis. Alternatively, it may directly affect controlling elements that interact with the virus to suppress or induce its expression andlor replication.

VI. Proposed Mechanisms of Sequence Amplification Is it possible, at this juncture, to fit the largely phenomenological observations that have been reviewed here into any coherent model for the mechanism(s) involved in gene amplification in mammalian cells? Since there seem to be major differences in the modes andlor manifestations of amplification between cell types, and sometimes between different loci in the same cell, there clearly exist variations of any central mechanism. But we will attempt to outline certain plausible models that may be discounted or supported by future experiments.

68

JOYCE L. HAMLIN ET A L

The data to be reconciled can be summarized as follows. Mammalian cells can amplify DNA sequences many hundreds of times, possibly at random positions within the genome. The amplified units (amplicons) that have been examined are almost always very large; in cases of drug resistance, the amplicon is much larger than the gene that confers resistance on the cell. The amplicons are apparently arranged in tandem, linear arrays. Depending upon the cell type and the particular locus in question, the multiple copies can be integrated stably into preexisting chromosomes as HSRs (often at the location of the parental gene), or they can exist as extrachromosomal acentromeric double minutes. In some cell lines, the multiple amplicons appear to be able to shuttle between these alternate configurations. Since the net result of amplification is that a particular locus is now present in a supernumerary amount relative to other loci in the same cell, the mechanism must involve either multiple replications of that locus during the S period, or some other mechanism related to recombination that increases the copy number of a given locus relative to the rest of the genome. Possibly gene amplification is a combination of both processes. A. UNEQUAL SISTERCHROMATID EXCHANGE One suggested model is unequal exchange between sister chromatids, with the result that one chromatid obtains two copies of a sequence for which the other chromatid is now deleted. In order for the initial duplication to occur (according to current concepts of recombination in mammalian cells), homologous DNA sequences have to flank the locus in question (A in Fig. 3) in order to provide a basis for pairing; recombination occurs outside the locus itself. If A includes sequences coding for a selectable gene, then only the cell receiving two copies of A will survive, and the other daughter cell will be killed by drug selection. Once duplication occurs, it is now possible for recombination to occur again, either at the original elements that provided homology (X in Fig. 3), or within the sequence A itself, again by a staggered mispairing mechanism. As the number of amplifications increases, recombination should become more likely, since there will be more possibilities for mispairing. It would also be possible to generate collections of amplicons of different size in a single cell if it is further assumed that staggered recombination can occur in regions both within and flanking the core amplicon, A, at highly repeated elements that are dispersed at different positions throughout the region. Amplification would provide a collection of repeated units that would not have precisely defined boundaries, due to recombination occurring at slightly different positions each time, but a core element (A) should be observed that would usually include the gene itself (which is selected). Many different kinds of junction fragments could be formed by a model of this kind, but it might be expected that at least some vestige of the


I

* 3

=&r ,

x x xA x x x x

Staggered Pairing

x x x x x q

Exchange

X

4

69

A

A

L * +

Replication and mitosis

Dauahter Chromosomes

FIG. 3 . Unequal sister chromatid exchange. The locus A contains a gene or genes (possibly selectable), and is flanked by repetitive elements (x) that could be ribosomal, satellite, or other tandemly repeated sequences. An unequal, homologous recombination event occurs between two x elements situated on either side of A in the two chromatids (as indicated in 1 above). After mitosis and another round of DNA synthesis, one cell is deleted for A on this chromosome, and the other cell now has two copies of A on this chromosome. If A itself contains dispersed repetitive elements (y), as indicated in 1 , then unequal exchanges can occur within A itself, leading to amplicons of variable length and constitution.

repeated dispersed elements that form the basis for recombination might be maintained in each junction fragment. Another prediction is that some highly repeated elements should flank the original locus in the parental cell which has only one copy of the amplicon per chromosome. A further prediction of this model is that the frequency of sister chromatid exchange in HSRs should be higher than in other chromosomal regions, owing to the greater opportunities for mispairing in a tandem array of repeated amplicons. In addition, agents that provoke recombination and sister chromatid exchange, such as those that damage DNA, might be expected to increase the frequency of amplification events. According to this scheme, double minutes would involve internal recombination between adjacent amplicons on the same Chromatid, releasing circular elements made up of monomers or higher multiples of the amplicon arranged in tandem. The differences between cell lines and between loci in a single cell line in their propensity to form either DMs or HSRs would have to be explained by subtle differences related to intrachromatid recombination, and could depend on the nature of the locus itself with respect to the distribution of the recom-

70


binogenic sequence elements. It might be expected that certain loci would have a greater likelihood to be amplified than others which are not flanked on both sides by the proper elements. In order to explain the stable fixation of genes as HSRs after prolonged culture of cell lines that originally contained DMs, tandemly arranged amplicons would have to integrate in such a way that the resulting sequences had little propensity for the intrastrand recombination events that are suggested to lead to extrachromosomal DMs. At the present time, there are experimental observations that support this model and those that conflict with it. The genome of mammalian cells is peppered with hundreds of thousands of simple sequence elements (e.g., Alu and Ah-like sequences, satellite DNAs, etc.) that are dispersed both within and outside of genes (Jelinek and Schmid, 1982). Clearly, elements of this kind could provide the requisite homologies for the recombination events suggested, provided that they are positioned appropriately with respect to the amplicon. Recombination between highly repeated satellite DNAs apparently occurs quite often in mammalian cells, and Bostock and Clark (1980) have shown that satellite DNAs form a major part of the DHFR amplicon in the MTX-resistant murine melanoma, PG19T3. In addition, the karyotype of these cells is very unstable, with the multiple copies shuttling frequently into the chromosomes as HSRs and out as DMs. Bostock and Clark have proposed that recombination between DMs at homologous sequences, with subsequent integration by recombination into chromosomes, could account for the formation of HSRs. A disproportionation by intrastrand recombination would then reverse the process and generate DMs that would presumably be variable in size, depending upon the number of amplicons between the two sites of recombination. It is difficult to see how this process could lead to the multiple, uniformly sized DMs that are observed in a single cell in most other systems, however, unless it is assumed that in these other cases, the original excision event occurred early in the amplification process, and multiple DMs of similar size are generated by the DNA synthetic process in a given S period. Another important observation is that ribosomal DNA, a highly repeated element in mammalian cells, is included in the CAD amplicon in Syrian hamster cells (Wahl et al., 1983), and Wahl and co-workers have suggested that rDNA cistrons flanking the CAD locus could be responsible for the original duplication and subsequent amplification of this gene by the process outlined above. Support for this concept derives from studies in bacteria, in which duplications of DNA sequences flanked by rDNA genes are very frequent occurrences (Anderson and Roth, 1977). In addition, in rat hepatoma (Tantravahi et al., 1981), sarcoma (Tantravahi et al., 1982), and erythroleukemia (Murao et al., 1982) cell lines, the rDNA genes have been shown to be amplified, and the HSRs containing these genes are often located on chromosomes that contain nucleolar organizer regions in parental rat cells. In these systems, it is not known what other sequences are


71

contained in the amplicon, but it is clear that the repetitive rDNA elements themselves could be providing the basis for unequal recombination. Furthermore, Henderson and Mcgraw-Ripley (1 982) have pointed out that in many human neoplasms, rDNA genes are frequently involved in translocations and other recombinational rearrangements, and are often amplified, suggesting that the repetitive nature of these sequences aids recombination and amplification. Chasin et al. (1982) have addressed the question whether HSRs display a higher frequency of sister chromatid exchanges, as would be predicted by the recombination model for gene amplification. By utilizing the BUdR/Hoescht 33258 method devised by Latt (1974) to illuminate sister chromatid exchanges in MTX-resistant CHO cells, they found that the frequency within the HSR was not greater than in other chromosomal regions, and, indeed, was somewhat depressed. These data therefore do not support one prediction of the recombination model. B. REREPLICATION The model that has received the most attention in recent years states that a given DNA sequence can undergo multiple rounds of DNA synthesis prior to mitosis (Schimke, 1982; Varshavsky, 1981b; Hamlin et al., 1983). This could occur randomly at a low frequency at any replicon, due to leakiness in the control mechanism that normally prevents reinitiation. Alternatively, a mutation at an origin could allow reinitiation, and would, of course, perpetuate the property in that particular replicon, facilitating further amplifications. Agents that interfere with DNA synthesis or that damage DNA could increase the frequency of this kind of mutation. Alternatively, a variety of agents could provoke reinitiation by somehow transiently affecting the structure of the origin. If mammalian chromosomal replicons are defined by fixed origins and termini, then this model predicts that the unit of amplification may be equivalent to the domain of a chromosomal replicon. The amplification mechanism could be relatively precise and could therefore generate uniformly sized amplicons (both within a given cell and between different cells derived from the same parent). However, it is also possible that the replication forks could terminate at different positions within or outside of the replicon during each amplification event, generating a collection of amplicons whose center is the same, but whose size is different. In the latter case, each amplicon could have more than one origin of DNA synthesis. It is also possible that uncontrolled DNA synthesis could initiate at sites not usually utilized as origins, especially if the cellular DNA is damaged by agents that generate single-stranded breaks, etc. A very heterogeneous collection of amplicons would be expected by this mechanism, since neither the origin nor the termini would be fixed at any sequence.

72


Regardless of the mode of rereplication of a given sequence, some mechanism must operate to join the extra DNA copies into tandem arrays (either as HSRs or as DMs), else the onion skin arrays characteristic of chorion gene amplification would result. Thus, some form of recombination must operate in any scheme that invokes multiple rounds of DNA synthesis as the inductive event. In one variation, repeated initiations at one origin would generate side-by-side duplexes with only the parental strands actually covalently integrated into the chromosomal DNA fiber as a whole (Fig. 4A). The nonintegrated supernumerary duplexes would be free to ligate end-to-end, either with themselves to form circles, or with other duplexes to form tandem (and possibly circular) arrays that could be released as DMs. HSRs could be formed by the integration of these single or tandem structures into either of the duplexes containing a parental strand. Alternatively, HSRs could be the result of recombination of the DMs at some later time into sites close to or distant from the original replicon. A mechanism for the joining of supernumerary copies of the amplicon to form tandem repeats (either chromosomal or extrachromosomal) is suggested by the studies of Bullock and Botchan (1982). They have examined the amplification and excision of integrated SV40 viral genomes in transformed rodent cells after treatment with mitomycin C or fusion with permissive monkey cells. In this system, an increase in the viral copy number in high-molecular-weight DNA precedes the appearance of free viral forms. In addition, excision and amplification require a functional T antigen. They therefore conclude that amplification is dependent on replication. The extrachromosomal products recovered are closed circles that are apparently formed by the pairing and recombination of two short homologous sequences, one present within the virus and one present in flanking cellular DNA. From one clonal cell line with a single integrated copy of the virus, many different excised copies can result in which different short homologies are utilized. However, from cell lines with tandemly duplicated viral inserts, a homogeneous collection of unit length autonomous forms results, apparently because in this circumstance, the exact and extensive (4 kb) homology between the two tandem viral copies overrides the short homologies used by single copy inserts. Bullock and Botchan propose that the single-stranded regions generated at replication forks during onion skin rereplication facilitate the pairing of the two homologous sequences that straddle the origin. Depending upon whether recombination occurs between sequences on the same or on different duplexes, the products are released from the chromosome as autonomous elements, or result in the in situ tandem duplication of the region contained between the two regions of homology (see Fig. 4A). They further propose that mitomycin C or fusion with permissive cells induces or activates enzymes that are used for viral replication, which is the triggering event in this model. Whether the mechanism of induced replication of SV40 is related to that responsible for amplification of chromosomal DNA sequences remains to be


73

A

-0-

T

T

0

0

FIG.4. Proposed mechanisms for the formation of tandem repeats or extrachromosomal elements after rereplication. (A) If a replicon is reduplicated prior to mitosis, an onion skin structure could result, in which two of the four daughter duplexes are not covalently attached to the original chromosomal DNA fibers. Recombination events could occur between homologous elements that happen to flank the origin of replication, as indicated by X in the diagram. Recombinations between two elements on either side of the origin would result in a closed circle if the elements were on the same duplex, or in a tandem duplication if the elements were on different duplexes. In the latter case. integration into the original chromosomal fiber occurs if one of the duplexes contains a parental strand. (B) If DNA is arranged in a series of loops affixed to a nuclear matrix, each one of which represents a replicon, then rereplication during a single S period could result in the structure pictured above. Three adjacent replicons are shown, only one of which has undergone two cycles of replication. This structure is analogous to the one pictured in A, except that the termini indicated by the arrows in A are juxtaposed in B. Again, recombination is proposed to occur between the homologous elements X, on the same or on different duplexes.

74


seen. It is difficult to imagine how the chromosomal amplicons, which are usually hundreds of kilobases long, could arise by this mechanism, since it would be expected that at least some of the time the two short homologous sequences required would flank the origin more closely. However, the experiments with SV40 do illuminate the kinds of operations that cells are capable of performing on DNA, and variations of this mechanism could be involved in the amplification of chromosomal sequences other than viruses. Current ideas about the physical arrangement of DNA in the mammalian nucleus provide another suggestion for the formation of tandem arrays of replicated sequences that could account for the large size of most amplicons. A variety of microscopic and biochemical evidence has demonstrated that chromosomal DNA is arranged in loops attached to a subnuclear, proteinaceous scaffold or matrix (Worcel and Benyajati, 1979; Cook et al., 1976), and it has been proposed that replication occurs by feeding the DNA loops through a replication complex attached to the matrix (Pardoll et al., 1980). Once a replicon has been synthesized, the two daughter loops could end up with their four termini juxtaposed as in Fig. 4 9 , awaiting forks from adjacent replicons to approach in order to complete a complex resolution event (possibly involving topoisomerase) that unwinds any super coils ahead of the replication forks and fuses adjacent replicons. An occasional aberrant recombination event at the termini could then generate a head-to-tail tandem integration of the two DNA loops into one continuous strand, or could release a monomeric circle of the replicon. By this model, amplicons are formally equivalent to replicons, and would be of relatively uniform size in a given cell. It is also possible that recombination could occur at multiple sites located in the general vicinity of the replicon termini, generating amplicons of similar size, but terminating at slightly different positions. Furthermore, it could be imagined that while rereplication would increase the likelihood of this event by providing more than four termini simultaneously, it would not necessarily be required by this model. DNA damaging agents could increase the frequency of incorrect resolution or rereplication, or both. The DMs and HSRs generated by the process of rereplication could then be synthesized by the usual mode, utilizing the legitimate origin in bidirectional replication. Alternatively, the proposed circular form of tandem repeats (DMs) could replicate by the rolling circle mechanism observed in the amplification of extrachromosomal rDNA copies in Xenopus (Hourcade et al., 1973; Rochaix et al., 1974; Buongiorno-Nardelli et al., 1976). The data compatible with rereplication models come from several indirect experiments, none of which absolutely distinguishes between rereplication and recombination. The amplicons that have been analyzed so far all seem to be large, and could all be within the range of mammalian chromosomal replicons. However, since no amplicon has been cloned in its entirety, they may turn out to be much larger than replicons, by arguments discussed previously.


75

Other circumstantial evidence that implicates DNA replication in the process of amplification derives from studies on the pattern of synthesis of amplicons. The HSRs in three MTX-resistant Chinese hamster cell lines, as well as in a stable derivative of murine S 180 cells, have been observed to initiate replication at multiple loci along their length in early S, and to complete replication by the mid-S period (Hamlin and Biedler, 198 1; Milbrandt et a / ., 198 1; Kellems et al., 1982). In the murine and CHO cell lines that bear transfected, amplified genes as HSRs, multiple initiations within the HSR in early S have also been observed (Kaufman et al., 1983; Hamlin, unpublished observations). In the latter cases, it might be assumed that the cloned gene was introduced into a chromosomal replicon, and that subsequent amplification occurred via the origin of that replicon. These results imply that there is at least one origin of DNA synthesis per repeated unit, as would be expected if legitimate origins were involved in the amplification process. It has also been shown that there is probably only one origin per DHFR amplicon in a MTX-resistant CHO cell line (Hamlin et al., 1983), which lends weight to the argument that an amplicon is equivalent to a parental replicon. As discussed earlier, Schimke and colleagues have demonstrated that rereplication can be induced in a population by a variety of drugs that interfere with DNA synthesis (Tlsty et al., 1982; Brown et af., 1983). However,the amplifications observed in this experimental situation need not be the primary mechanism that accounts for all amplifications, particularly those observed in the absence of obvious insults and/or selections (e.g., neuroblastomas, carcinomas, etc.). The other point worth making is that there may be no single mechanism by which all amplifications can be explained. Certain chromosomal regions that contain repetitive elements such as satellite or rDNA may undergo a few rounds of unequal sister chromatid exchange that establishes an unstable condition in the nucleus. The extra DNA may not be able to affix itself properly to the matrix, encouraging aberrant resolution events during subsequent DNA synthesis that lead to further amplifications. Alternatively, infrequent rereplication that produces tandem repetitions may stimulate subsequent unequal exchange. Even more complex models can be invoked in which both processes (rereplication/insertion and unequal exchange) occur continuously during the entire amplification process, in order to explain the complex sequence arrangements observed in some systems (e.g., Ardeshir et a l . , 1983).

V11. Concluding Remarks It is clear that much remains to be learned about DNA amplification in mammalian cells. One of the most promising areas of investigation is the analysis of sequences contained in amplicons. When it becomes possible to isolate entire

76

JOYCE L. HAMLIN E T AL.

amplicons in recombinant clones, as well as the parental domain from which the amplicon was derived, much more will be learned about the mechanisms responsible for this interesting genetic phenomenon. Analysis of junction fragments between repeated units will be particularly illuminating, since they may prove to contain the highly repetitive elements responsible for staggered recombination events. Alternatively, it may be possible to show that the ends of amplicons are equivalent to the termini of replicons. Another exciting area of investigation is the amplification of transfected genes, which locate themselves in new and apparently random chromosomal environments. It may be possible to find homologous sequence elements between these heterogeneous amplicon types that are required for the process itself. A most rewarding consequence of this rapidly expanding area of investigation is that whole new fields of vision have been opened. An understanding of the mechanism of gene amplification will necessarily tell us much about the physical and functional organization of DNA in chromosomes, and the complex processes of DNA synthesis and recombination in mammalian cells. The field of cytogenics will undoubtedly be aided as well by an understanding of the types of sequence that generate particular staining properties to chromatin. The study of gene regulation, which is a difficult endeavor in mammalian cells, may be aided by a kind of pseudogenetic approach in which a nonselectable gene can be cotransfected with an amplifiable gene such as DHFR or CAD, and the two genes can be coamplified by drug selection. The mode of regulation of the passenger gene (or lack of it) should shed light on the nature of the other elements involved in normal gene expression. In addition, the overproduction of virtually any protein for which the gene can be cloned will be allowed by this approach. This will be an important development for the purification of scarce biological peptides whose synthesis involves complex processing steps that could not be engineered in bacteria. Finally, the discovery that cellular counterparts of viral oncogenes can be amplified in some human tumors is a major advance in our understanding of the genesis 01cancer. The important work on the inductive effects of certain agents (including anticancer drugs) on amplification will modify our current drug treatment protocols toward more rational directions.

ACKNOWLEDGMENTS We would like to thank the many colleagues who sent us manuscripts prior to their publication. We would also like to thank Melinda Mills for her expert assistance in the preparation of the manuscript. Work in the authors’ laboratory was supported by grants from the NIH and The March of Dimes. J.D.M., N.H H . , and J.C.A. were supported by NIH postdoctoral fellowships, and J.L.H. was the recipient of an American Cancer Society Faculty Research Award.


77

REFERENCES Akerblum, L., Ehrenberg. A., Graslund, A., Lankinen, H., Reichard, P., andThelander, L. (1981). Proc. Nail. Acad. Sci. U . S . A . 78, 21 59-2163. Alitalo, K., Schwab, M., Lin, C. C., Varmus, H. E., and Bishop, J . M. (1983). Proc. Natl. Acad. Sci. U.S.A. 80, 1707-1711. Allen, G., and Fantes, K. H. (1980). Nature (London) 287, 408-41 1. Alt, F. W., Kellems, R. E., Bertino, J. R., and Schimke, R. T. (1976). J. B i d . Chem. 253, 13571370. Amtmann, E., and Sauer, G. (1982). Nature (London) 296, 675-677. Anderson, R . P., and Roth, J. R. (1977). Annu. Rev. Microbiol. 31, 473-504. Anderson, R. P., Miller, C. G . , and Roth, J. R. (1976). J. Mol. B i d . 105, 201-218. Andrulis, I. L.,Duff, C., Evans-Blackler, S . , Worton, R., and Siminovitch, L. ( 1 983). Mol. Cell. Biol. 3, 391-398. Ardeshir, F., Giulotto, E., Zieg, J., Brison, 0 . .Liav, W. S . L., and Stark, G. R. (1983). Mol. Cell. Eiol. 3, 2076-2088. Axel, R., and Roberts, J. M. (1982). Cell 29, 109-1 19. Balaban-Malenbaum, G., and Gilbert, F. (1980). Cancer Genet. Cytogenet. 2, 339-348. Baran, N., Neer, A,, and Manor, H. (1983). Proc. Natl. Acad. Sci. U.S.A. 80, 105-109. Barclay, B.J.,Kuntz, B.A., Little, J.G.,andHaynes, R. H. (1981).Can.J . Eiochem. 60, 172-194. Barker, P. E. (1982). Cancer Genet. Cytogenet. 5, 81-94. Barker, P. E., and Hsu, T. C. (1979). J. Natl. Cancer Inst. 62, 257-261. Barker, P. E., and Stubblefield, E. (1979). J. Cell B i d . 83, 663-666. Barker, P. E., Drwinga, H. L., Hittelman, W. N., and Maddox, A.-M. (1980). Exp. Cell Res. 130, 353-360. Barsoum, J . , and Varshavasky, A. (1983). Proc. Natl. Acad. Sci. U.S.A. 80, 5330-5334. Baskin, F., Carlin, S. C . , Kraus, P., Friedkin, M., and Rosenberg, R. N. (1975). Mol. Plzarmacol. 11, 105-117. Baskin, F., Rosenberg, R., and Dev, V. (1981). Proc. Natl. Arad. Sci. U.S.A. 78, 3654-3658. Beach, L. R., and Palmiter, R. D. (1981). Proc. Nail. Acad. Sci. U.S.A. 78, 2110-2114. Beach, L. R., Mayo, K. E., Durnam, D. M., and Palmiter, R. D. (1981). In “Developmental Biology Using Purified Genes” (D. D. Brown, ed.), pp. 239-248. Liss, New York. Biedler, J . L. (1982). In “Gene Amplification” (R. T. Schimke, ed.), pp. 39-45. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Biedler, J . L., and Riehm, H. (1970). Can. Res. 30, 1174-1 184. Biedler, J. L., and Spengler, B . A. (1976a). Science 191, 185-187. Biedler, J . L., and Spengler, B. A. (1976b). J. Narl. Cancer Inst. 57, 683-695. Biedler, J. L., Melera, P. W . , and Spengler, B. A. (1980). Cancer Genet. Cytogenet. 2, 47-60. Biedler, J . L.. Chang, T., Peterson, R. H. F., Melera, P. N., Meyers, M. B., and Spengler, B. A. (1983). In “Rational Basis for Chemotherapy,” pp. 71-92. Liss, New York. Bird, A. P. (1978). Cold Spring Harbor Symp. Quant. Biol. 42, 1179-1183. Birg, F., Dulbecco, R., Fried, M., and Kamen, R. (1979). J. Virol. 2, 633-648. Bishop, J . M. (1983). CeN 32, 1018-1020. Bostock, C . J., and Clark, E. M. (1980). Cell 19, 709-715. Bostock, C. J., and Summer, A. T. (1978). “The Eukaryotic Chromosome,” pp. 256-259. North Holland h b l . , Amsterdam. Bostock, C. J . , and Tyler-Smith, C. (1982). In “Gene Amplification” (R. T. Schimke, ed.), pp. 15-22. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Botchan, M., Topp, W . , and Sambrook, J. (1979). Cold Spring Harbor Symp. Quant. Biol. 43, 709-719.

78


Botchan, M., Stringer, J., Mitchison, T . , and Sambrook, J. (1980). Cell 20, 143-152. Brennand, J . , Chinault, A. C . , Konecki, D. S., Melton, D. W., and Caskey, C. T. (1982). Proc. Narl. Acad. Sci. U.S.A. 79, 1950- 1954. Brown, P. C . , Beverley, S. M., and Schimke, R. T. (1981). Mol. Cell. Biol. 1, 1077-1083. Brown, P. C., Tlsty, T. D., and Schimke, R. T. (1983). Mol. Cell. Biol. 3, 1097-1 107. Bullock, P., and Botchan, M. (1982). In “Gene Amplification” (R.T. Schimke, ed.), pp. 215224. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Buongiorno-Nardelli, M., Amaldi, F., and Lava-Sanchez, P. A. (1976). Exp. CellRes. 98,95-103. Chasin, L. A., Graf, L., Ellis, N., Landzberg, M., and Urlaub, G. (1982). In “Gene Amplification” (R. T. Schimke, ed.), pp. 161-165. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Chattopadhyay, S . K., Chang, E. H., Lander, M. R., Ellis, R. W . , Scolnick, E. M., and Lowy, D. R. (1982). Nature (London) 296, 361-363. Chin, D. J., Luskey, K. L., Anderson, R. G. W., Faust, J . R., Goldstein, J . L., and Brown, M. S . (1982). Proc. Narl. Acad. Sci. U.S.A. 79, 1185-1189. Clowes, R. C. (1972). Bacreriol. Rev. 36, 361-405. Coderre, J. A., Beverley, S. M., Schimke, R. T . , and Santi, D. V. (1983). Proc. Narl. Acad. Sci. U.S.A. 80, 2132-2136. Colantuoni, V., Dailey, L., and Basilico, C. (1980). Proc. Natl. Acarf. Sci. U.S.A. 77, 3850-3854. Colantuoni, V., Dailey, L., Della Valle, G . , and Basilico, C. (1982). J . Virol. 43, 617-628. Collins, S . , and Groudine, M. (1982). Nature (London) 298, 679-681. Collins, S. J . , and Groudine, M. T. (1983). Proc. Narl. Acad. Sci. U.S.A. 80, 4813-4817. Cook, P., Brazell, I. A,, and Jost, E. (1976). J . Cell Sci. 22, 303-324. Cowan, K . H., Goldsmith, M. E., Levine, R. M., Aitken, S. C., Douglass, E., Clendeninn, N., Nienhuis, A. W., and Lippman, M. E. (1982). J . Biol. Chem. 257, 15079-15086. Cowell, J. K. (1982). Anrru. Rev. Gene?. 16, 21-59. Criscuolo, B. A,, and Krag, S. S. (1982). J . Cell Biol. 94, 586-591. Crouse, G . F., McEwan, R. N., and Pearson, M. L. (1983). Mol. Cell. Biol. 3, 257-266. Dalla Favera, R., Wong-Staal, F., and Gallo, R. C. (1982). Nature (London) 299, 61-63. Daneholt, B., and Edstrom, J.-E. (1967). Cytogenetics 6 , 350-356. Debatisse, M., Berry, M., and Buttin, G. (1982). Mol. Cell. Biol. 2, 1346-1353. Della Valle, G . , Fenton, R. G., and Basilico, C. (1981). Cell 23, 347-355. de Saint Vincent, B. R., Delbruck, S., Eckhart, W . , Meinkoth, J., Vitto, L., and Wahl, G . (1981). Cell 27, 267-277. de Zamaroczy, M . , Marotta, R., Faugeron-Fonty, G . , Goursot, R., Mangin, M., Baldacci, G . , and Bernard, G . (1981). Nature (London) 292, 75-78. Dolnick, B. J., Berenson, R. J., Bertino, I. R., Kaufman, R. J . , Nunberg, J. H., and Schimke, R . T. (1979). J . Cell Biol. 83, 394-402. Edlund, T . , Grundstrom, T., Bjork, G . R., and Normark, S. (1980). Mol. Gen. Genet. 180, 249257. Epstein, H. F., Waterston, R. H., and Brenner, S. (1974). J . Mol. Biol. 90, 291-300. Ferris, S. D., and Whitt, G. S. (1980). Am. Nut. 115, 650-666. Fischer, G . A. (1961). Biochem. Pharmacol. 7, 75-80. Flintoff, W. F., Weber, M. K., Nagainis, C. R., Essani, A. K., Robertson, D., and Salser, W. (1982). Mol. Cell. Biol. 2 , 275-285. Fogel, S., and Welch, J. W. (1982). Proc. Natl. Acad. Sci. U . S . A . 79, 5342-5346. Fougere-Deschatrette, C . , Schimke, R. T.. Weil, D., and Weiss, M. C . (1982). I n “Gene Amplification” (R. T. Schimke, ed.), pp. 29-32. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Fuscoe, J. C., Fenwick, R. G . ,Jr., Ledbetter, D. H., and Caskey, C. T. (1983). M o l . Cell. Biol. 3, 1086-1096.


79

Fyrburg, E. A,, Kindle, K . L., Davidson, N., and Sodja, A. (1980). Cell 19, 365-378. Gaillard, C.. S t r a w , F., and Bernardi, G. (1980). Nature (London) 283, 218-220. Galas, D. J., and Chandler, M. (1981). Proc. Natl. Acad. Sci. U.S.A. 78, 4858-4862. Gasser, C. S., Simonsen, C. C., Schilling, J. W., and Schimke, R. T. (1982). Proc. Narl. Acad. Sci. U.S.A. 79, 6522-6526. Gatti, M., Rizzoni, M., Palitti, F., and Olivieri, G. (1973). Mutar. Res. 20, 87-99. Gattoni, S., Colantuoni, V., and Basilico, C. (1980). J . Virol. 34, 615-626. Gelb, L. D., Kohne, D. E., and Martin, M. A. (1971). J . Mol. Biol. 57, 129-145. George, D., and Powers, V. (1981). Cell 24, 117-123. George, D. L., and Franke, U . (1980). Cyfogenet. Cell Gener. 28, 217-226. George, D. L., and Powers, V. E. (1982). Proc. Natl. Acad. Sci. U.S.A. 79, 1597-1601. Gick, G. G., and McCarty, K. S., Sr. (1982). J . B i d . Chem. 257, 9049-9053. Gilbert, F., and Balaban, G . (1982). In “Gene Amplification” (R. T. Schimke, ed.), pp. 185-191. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Goldstein, A., Aranow, L., and Kolman, S. M. (1974). In “Principles of Drug Action,’’ chap. 8. Wiley, New York. Gottlieb, P. D. (1980). Mol. Immunol. 17, 1423-1435. Cottlieb, P. D., and Weeden, N. F. (1979). Evolution 33, 1024-1039. Graham. F., and van der Eb, A. (1973). Virology 52, 456-467. Grant, M. C., and Proctor, V . W. (1980). J . Phycol. 16, 109-1 15. Griffin-Shea, R.. Thireos, G . , Kafatos, F. C., Petri, W. H., and Villa-Komaroff, L. (1980). Cell 19, 91 5-922. Grossi, M. P., Corallini, A , , Meneguzzi, G., Chenciner, N., Barbanti-Brodano, G . , and Milanesi, G. (1982). Virology 120, 500-503. Hakala, M. T., Zakrzewski, S. F., and Nichol, C. A. (1961). J . B i d . Chem. 236, 952-958. Hamlin, J . L., and Biedler, J . L. (1981). J . Cell. Physiol. 107, 101-114. Hamlin, J . L., Montoya-Zavala, M.. Heintz, N. H., Milbrandt, J . D., and Azizkhan, J. C. (1982). In “Gene Amplification” (R. T. Schimke, ed.), pp. 155- 160. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Hamlin. J . L., Heintz. N . H., and Milbrandt. J . D. (1983). In “Mechanisms of DNA Replication and Recombination” (N. R . Cozzarelli, ed.) Liss, New York. Hanawalt, P. (1982). In “Gene Amplification” (R. T. Schimke, ed.), pp. 257-262. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Hanawalt, P. C., Cooper, P. K., Ganesan, A. K., and Smith, C . A. (1979). Annu. Rev. Biochem. 48, 783-836. Harshey, R . M., and Bukhari, A. I. (1981). Proc. Natl. Acad. Sci. U.S.A. 78, 1090-1094. Heintz, N . H . . and Hamlin, J . L. (1982). Proc. Nut/. Acad. Sci. U.S.A. 79, 4083-4087. Heintz, N. H., Milbrandt. J . D., Greisen, K . S., and Hamlin, J. L. (1983). Nature (London) 302, 439-44 1 . Henderson, A. S., and Megraw-Ripley, S. (1982). Cancer Genet. G-ytogenet. 6 , 1-16. Herreros, B.. and Gianelli. F. (1967). Nature (London) 216, 286-288. Hiscott, J . , Murphy, D., and Defendi, V. (1980). Cell 22, 535-543. Hiscott, J . B . , Murphy, D., and Defendi, V. ( I 98 I ). P roc. Narl. Arad. Sci. U.S.A. 78, 1736- 1740. Horns, R. C., Dower, W. J . . and Schimke. R . T. (1984). J . Cliti. Oncol. 2, 2-7. Hourcade, D.. Dressler, D., and Wolfson, J . (1973). Proc. Narl. Acad. Sci. U . S . A . 70, 2926-2930. Huberman. E . , McKeown, C. K . , and Friedman, J . (1981). Proc. Natl. Acud. Sci. U.S.A. 78, 31513154. Huberman, J . A,, and Riggs, A . D. (1968). J . Mol. B i d . 32, 327-341. Hyman, B. C., Cramer, J . H., and Rownd, R. H. (1982). Proc. Narl. Acad. Sci. U.S.A. 79, 15781582. Jakobssen, A. H., Dahllof. B., Mattinsson, T., and Levan, G . (1984). Submitted.

80


Jelinek, W. R., and Schmid, C. W. (1982). Annu. Rev. Biochem. 51, 813-844. Johnston, R. N., Beverley, S. M., and Schimke, R. T. (1983). Proc. Narl. Acad. Sci. U.S.A. 80, 371 1-3715. Kagi, J . H. R., and Nordberg, M. (1979). “Metallothionein.” Birkhaser, Basel. Kanda, N., Schreck, R., Ah, F., Bruns, G., Baltimore, D., and Latt, S. (1983). Proc. Nad. Acad. Sci. U.S.A. 80, 4069-4073. Kano-Tanaka, K., Higashida, H., Fukami, H., and Tanaka, T. (1982). Cancer Genet. Cytogenef. 5, 51-62. Karin, M., Andersen, R. D., Slater, E.. Smith, K . , and Henchman, H. (1980). Nature (London) 286, 295-297. Kaufman, R. J . , and Schimke, R. T. (1981). Mol. Cell. Biol. 1, 1069-1076. Kaufman, R. J., and Sharp. P. A. (1983). 1.Mol. B i d . 159, 601-621. Kaufman, R. J., Brown, P. C . , and Schimke, R. T. (1979). Proc. Nut/. Acad. Sci. U.S.A. 76,56695673. Kaufman, R. J., Brown, P. C., and Schimke, R. T. (1981). Mol. Cell. Biol. 1, 1084-1093. Kaufman, R. J., Sharp, P. A., and Latt, S. A. (1983). Mol. Cell. Biol. 3, 699-71 I . Kedes, L. H. (1979). Annu. Rev. Biochem. 48, 837-870. Kellems, R. E., Alt, F. W., and Schimke, R. T. (1976). J . Biol. Chem. 251, 6987-6993. Kellems, R. E., Harper, M. E., and Smith, L. M. (1982). J . Cell Biol. 92, 531-539. Kogoma, T., and Lark, K. (1975). J . Mol. Biol. 94, 243-256. Kohl, N. E., Kanda, N., Scheeck, R. R., Bruns, G., Latt, S . A., Gilbert, F., and Alt, F. W. (1983). Cell 359-367. Kopnin, B. P. (1981). Cytogenet. Cell Genet. 30, 11-14. Kozinski, A. W., Ling, S . K . , Hutchinson, N . , Halpem, M. E., and Mattson, T. (1980). Proc. Natl. Acad. Sci. U.S.A. 77, 5064-5068. Kuo, T., Pathak, S., Ramagli, L., Rodriguez, L., and Hsu, T. C. (1982). In “Gene Amplification” (R. T. Schimke, ed.), pp. 53-56. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Kurtz, D. T. (1981). J . Mol. Appl. Gen. 1, 29-38. Kutcherlapati, R. S . , Hwang, P., McDougall, J. K., and Botchan, M. R. (1978). Proc. Natl. Acad. Sci. U.S.A. 75, 4460-4464. Latt, S. A. (1974). J . Histochem. Cytochem. 22, 478-491. Lavi, S. (1981). Proc. Nut/. Acad. Sci. U.S.A. 78, 6144-6148. Levan, A,, and Levan, G. (1978). Heredifas 88, 81-92. Levan, A,, Levan, G., and Mitelman, F. (1977). Heredifas 86, 15-30. Levan, G . , and Levan, A. (1982). In “Gene Amplification” (R. T. Schimke, ed.), pp. 91-97. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Levan, G . , Mandahl, N., Bregula, V . , Klein, G., and Levan, A. (1976). Hereditas 83, 83-90. Littlefield, J. W. (1969). Proc. Natl. Acad. Sci. U.S.A. 62, 88-95. Long, E. O., and Dawid, I. B . (1980). Annu. Rev. Biochem. 49, 727-764. Maniatis, T., Fritsch, E. F., Lauer, J . , and Lawn, R. M. (1980). Annu. Rev. Genet. 14, 145-178. Mariani, B. D., Slate, D. L., and Schimke, R. T. (1981). Proc. Natl. Acad. Sci. U.S.A. 78, 49854989. Martinsson, T., Tenning, P., Lundh, L., and Levan, G. (1982). Hereditas 97, 123-137. Mattes, R., Burkardt, H. J., and Schmitt, R. (1979). Mol. Gen. Genet. 168, 173-184. McLachlan, A. D. (1979). Eur. J . Biochem. 100, 181-187. McMillin, D. E., and Scandalios, J. G. (1980). Proc. Nufl. Acad. Sci. U.S.A. 77, 4866-4870. Melton, D. W., Konecki, D. S . , Ledbetter, D. H., Hejtmancik, J. F., and Caskey, C. T. (1981). Proc. Natl. Acad. Sci. U.S.A. 78, 6977-7980. Meneguzzi, G., Chenciner, N., Coralinni, A., Grossi, M., Barbantini-Brodano, G . , and Milanesi, G. (1981). Virology 111, 139-153.


81

Meyer, J . , and lida, S. (1979). Mol. Gen. Genet. 176, 209-219. Milbrandt. I . D., Heintz. N. H., White, W. C . , Rothman, S. M., and Hamlin, J . L. (1981). Proc. Nad. Acad. Sci. U.S.A. 78, 6043-6047. Milbrandt, I . D.. Azizkhan, J. C.. Greisen, K . S . , and Hamlin, J. L. (l983a). Mol. Cell. B i d . 3, 1266- 1273. Milbrandt, J . D . , Azizkhan, J . C . , and Hamlin, J . L. (1983b). Mol. Cell. Biol. 3, 1274-1282. Moormann, R. I . M., Den Dunnen, J . T., Bloemendal, H., and Schoenmakers, J. G. G. (1982). Proc. Narl. Acad. Sci. U.S.A. 79, 6876-6880. Murao, A., Horita. Y.. Maeda, S . , Takahashi, R., Kano, Y., and Sugiyama, T. (1982). Cancer Genet. Cyrogenet. 6, 303-312. Murray, M. J . , Kaufman, R. J . , Latt, S. A., and Weinberg, R. A. (1983). Mol. CellBiol. 3, 32-43. Nathanson. S. G . , Vehara, H . , Ewenstein, B. M., Kindt, T. J., and Coliyan, J. E. (1981). Annu. Rev. Biochem. 50, 1025-1052. Nunberg, J. H., Kaufman, R. J.. Schimke, R. T., Urlaub, G., and Chasin, L. A. (1978). Proc. Nail. Acad. Sci. U.S.A. 75, 5553-5556. Oakeshott, J. G., Chambers, G . K., East, P. D., Gibson, J . B., and Barker, J. S . F. (1982). Ausr. J . B i d . Sci. 35, 73-84. Oscheim, Y . N . , and Miller, 0. L. (1983). Cell 33, 543-553. Pardoll, D. M., Vogelstein, B.. and Coffey, D. A. (1980). Cell 19, 527-536. Pater. M. M., Pater, A., di Mayorca, G., Beth, E., and Giraldo, G . (1982). Mol. Cell. Biol. 2, 837844. Perez-Cano, R., Girgis, S . I . , and Macintyre. I. (1982). Acra Enducrinol. 100, 256-261. Piccini, N., Knopf, J . L., and Gross, K. W. (1982). Cell 30, 205-213. Pronk, I. C., Frants. R. R., Jansen, W., Eriksson, A. W., and Tonino, G . J . M. (1982). Hum. Gener. 60, 32-35. Quinn, L. A., Moore, G. E., Morgan, R. T . , and Woods, L. K. (1979). Cancer Res. 39, 49144924. Rao, 1. N., and Rao, M. V. P. (1980). Genet. Res. 35, 309-312. Robbins, J . , Freyer, G. A., Chisolm, D., and Cilliam, T. C. (1982). J . Biol. Chem. 257, 549-556. Roberts, J . M., and Axel, R. (1982). Cell 29, 109-119. Roberts, J . M., Buck, L. B., and Axel, R. (1983). Cell 33, 53-63. Roberts, M., Huttner, K. M., Schimke, R. T . , and Ruddle, F. H. ( 1980). J . Cell B i d . 87, SG2211, Rochaix, J . D., Bird, A,, and Bakken, A. (1974). J . Mol. B i d . 87, 473-487. Rossana, C., Rao, L. G . , and Johnson, L. F. (1982). Mol. Cell. B i d . 2 , I 1 18- 1125. Rownd, R . H. (1982). In “Gene Amplification” ( R . T. Schimke, ed.), pp. 273-282. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Royal, A., Garapin, A., Cami, B., Perrin, F., Mandel, J. L., LeMeur, M., Bregegegre, F., Cannon, F., LePennec, J . , Chambon, P., and Kourilsky, P. (1979). Narure (London) 279, 125-132. Sainerova, H., and Svoboda, I . (1981). Cancer Genet. Cvtogener. Cancer 3, 93-100. Sambrook, J . , Greene, R., Stringer, J., Mitchison, T., Hu, S . L . , and Botchan, M. R. (1979). Cold Spring Harbor Svinp. Quant. Biol. 44, 569-584. Sattler, P. W.,and Mecham, J. S. (1979). J . Hered. 70, 352-353. Schaffner. W. (1980). Proc. Natl. Acad. Sci. U.S.A. 77, 2163-2169. Schilling, J., Beverley, S . , Simonsen. C . , Crouse, G., Setzer, D., Feagin, G . , McGrogan, M., Kohlmiller, N., and Schimke, R. T. (1982). In “Gene Amplification” (R. T. Schimke, ed.), pp. 149-154. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Schimke, R. T. (1982). Harvey Lect. 76, 1-25. Schmitt, R.. Bemhard, E., and Mattes, R. (1979). Mol. Gen. Gener. 172, 53-65. Schwab, M., Alitalo. K., Varmus, H. E., Bishop, J. M., and George, D. (1983). Nature (London) 303, 497-501. Solomon, E. (1980). Narure (London) 286, 656-657.

82


Spradling, A. C. (1981). Cell 27, 193-201. Spradling, A. C.. and Mahowald, A. P. (1980). Proc. Natl. Acad. Sci. U.S.A. 77, 1096-1 100. Spradling, A., Digan, M. E., Mahowald, A. P . , Scott, M. W., and Craig, E. A . (1980). Cell 19, 905-9 14. Spriggs, A. I . , Boddington, M. M., and Clark, C. M. (1962). Er. Med. J . 2, 1431-1435. Srirnatkandada, S . , Medina, W. D., Cashmore, A. R., Whyte, W., Engel, D., Moroson, B. A , , Franco, C. T., Dube, S . K., and Bertino, J. R. (1983). Biochemistry 22, 5774-5781. Sutou, S . (1981). Cancer Genet. C.ytoytogenet. 3, 317-325. Takanari, H., and Izutsu K. (1981). Cytogenet. Cell Genet. 29, 77-83. Tantravahi, U., Guntaka, R. V . , Erlanger, B. F., and Miller, 0. J . (1981). Proc. Narl. Acad. Sci. U.S.A. 78, 489-493. Tantravahi, U., Erlanger, B. F., and Miller, 0. J. (1982). Cancer Gener. Cyfogenef.5, 63-73. Tatsumi, K., and Strauss, B. S. (1979). J . Mol. Biol. 135, 435-449. Tlsty, T . , Brown, P. C . , Johnston, R., and Schimke, R. T. (1982). In “Gene Amplification” (R. T. Schimke, ed.), pp. 231-238. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Trent, J. M. (1982). In “Gene Amplification” (R. T. Schimke, ed.), pp. 99-105. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Trent, J. M., Burk, R. H., Olson, S., Horns, R. C., and Schimke, R. T. (1984). J . Clin. Oncol. 2, 8-15. Varshavsky, A. (1981a). Cell 25, 561-572. Varshavsky, A. (1981b). Proc. Narl. Acud. Sci. U . S . A . 78, 3673-3677. Villafranca, J. E., and Robertus, J . D. (1981). J . Biol. Chem. 256, 554-556. Wahl, G. M., Padgett, R. A,, and Stark, G . R. (1979). J . Biol. Chem. 254, 8679-8689. Wahl, G . M., Allen, V., Delbruck, S., Eckhart, W., Meinkoth, J., Padgett, R., de Saint Vincent, B. R., Rubnitz, J., Stark, G., and Vitto, L. (1982). In “Gene Amplification” (R. T. Schirnke, ed.), pp. 167-175. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Wahl, G . , Vitto, L., and Rubnitz, J. (1983). Mol. Cell. Biol. 3, 2066-2075. Wahli, W., Dawid, I. B., Wyler, T., Jaggi, R . , Weber, R., and Ruffel, G. V. (1979). Cell 16,535549. Wakabayashi, S . , Hase, T . , Wada, K . , Matsubara, H.,and Suzuki, K. (1980). J . Biochem. 87,227236. Weber, K. E., and Hoegerman, S . F. (1980). Exp. Cell Res. 128, 31-39. Wellauer, P. K., Reeder, R. H., Dawid, I. B., and Brown, D. D. (1976). J . Mol. Biol. 105,487505. Westin, E. H., Wong-Staal, F., Gelman, E. P., Favera, R. D., Papas, T. S., Lautenberger, J. A., Eva, A., Reddy, E. P., Tronick, S. R., Aaronson, S . A., and Gallo, R. C. (1982). Proc. Nutl. Acad. Sci. U.S.A. 79, 2490-2494. Wolman, S. R . , Craven, M. L., Grill, S . P . , Domin, B. A., and Cheng, Y .-C. ( 1983). P roc. Nutl. Acad. Sci. U.S.A. 80, 807-809. Worcel, A., and Benyajati, C. (1979). Cell 12, 83-100. Yeung, C. Y., Ingolia, D. E., Bobonis, C., Dunbar, B. S., Riser, M. E., Siciliano, M. J . , and Kellems, R. E. (1983a). J . Biol. Chem. 258, 833-8345, Yeung, C. Y.. Riser, M. E., Kellerns, R. E., and Siciliano, M. J. (1983b). J . Biol. Chem. 258, 8330-8337. Zieg, J . , Clayton, C. E., Ardeshir, F . , Giulotto, E . , Swyrd, E. A , , and Stark, G. R. (1983). Mol. Cell. Biol. 3, 2089-2098.

INTERNATIONAL REVIEW OF CYTOLOGY, VOL 90

Computer Applications in Cell and Neurobiology : A Review R. RANNEY MIZE Department of Anatomy and Division of Neuroscience, University of Tennessee Center for the Health Sciences, Memphis, Tennessee

I.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

11. The Microcomputer in the Research Laboratory . . . . . . . . . . . . . . . . . 111. Computer Systems for Microscope Control and Plotting . . . . . . . . . .

IV . V. VI. VII. VIII. IX. X.

Serial Section Reconstruction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Computer-Aided Morphometric Measurement . . . . . . . . . . . . . . . . . . . Video Image Processing and Analysis . . . . . . . . . . . . . . . . . . . . . . . . . Computer Uses in Photometry and Fluorescence Microscopy . . . . . . Computer-Automated Autoradiography and Immunocytochemistry . . Other Cell Biology Computer Applications . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

83 84 90 93 98 I03 107 111

117 117 119

I. Introduction Computer-aided quantitative analysis has come of age in cell and neurobiology . The development of large scale integrated circuits (LSI) and metal oxide semiconductors (MOS) has literally revolutionized our scientific lives. Inexpensive microprocessors and microcomputers based on these technologies are assisting cell and neurobiologists in almost every scientific activity, from acquiring and analyzing data and controlling instrumentation to writing manuscripts, searching the research literature, and ordering laboratory supplies. The reduction in price of computer hardware, particularly small personal computers like Apple and TRS-80, makes a laboratory computer accessible to almost everyone. In preparing for a recent presentation on microcomputer applications in cell biology for the American Association of Anatomists (Mize, 1983e), it became apparent that there were very few review articles describing the uses of microcomputers in cell and neurobiology research. A number of small noncommercial computer systems have been developed for particular applications in these fields, but I found them difficult to locate because they were published in many different journals. Commercial systems were sometimes also difficult to locate. This review attempts to bring this widely dispersed literature together in an integrated 83 Copyrighl 0 1984 by Academic Pres\, Inc. All nphts of rrproduclion in any h r m reserved ISBN 0-12-364490-Y

84

R . RANNEY MIZE

format. The review is divided into application sections describing computer uses in microscope control and plotting, serial section reconstruction, computer-aided morphometric measurement, video image processing and analysis, photometry and fluorescence microscopy, autoradiography and immunocytochemistry, and other cell biology computer applications. These sections are preceded by a section describing microcomputer hardware and software which should prove useful to laboratories contemplating the purchase of a microcomputer. In this review, I have largely restricted my discussion to technical papers which provide detailed descriptions of hardware and software and their application to particular research problems. Although these papers may also include research results, their primary emphasis is on the application rather than the findings of the research. This necessarily excludes a large number of research articles which utilize computer analysis but are principally research reports not concerned with the computer methodology. Although I have included main frame and minicomputer systems in the review, the emphasis is placed upon recently developed systems that use smaller, inexpensive microcomputers or single board microprocessors. Specific mention of computer hardware is included to aid those with a given computer product to locate software developed for those systems.

11. The Microcomputer in the Research Laboratory Two approaches to laboratory computer automation have evolved over the last decade (Doerr, 1978; Enke, 1982; Shipton, 1979). The first approach uses dedicated microprocessors to automate single laboratory functions such as control of analytical instruments or calculation of mathematical functions. The second approach uses more versatile minicomputers or microcomputers which can handle a wide variety of laboratory tasks. Dedicated vs flexible is the key distinction to be made between the microprocessor and microcomputer approach (Enke, 1982). Microprocessor systems are single board central processing units (CPUs) with minimal memory whose logic or programmed steps are usually hardwired into the machine (Doerr, 1978). Microcomputers include large memories, high-level languages, input-output controls, and a wide range of peripherals which make them highly flexible, general purpose instruments. There are advantages and disadvantages to each approach. Microprocessors are ideal for dedicated control of single functions because they are fast and cheap. Many manufacturers incorporate microprocessors into laboratory instruments for control of specific device functions such as sampling a voltage level, controlling a power supply, or readout of data. Microprocessors are also often used to perform repetitive mathematical functions which reduce data to intermediate results. Their principal advantage in data acquisition and

COMPUTER APPLICATIONS IN NEUROBIOLOGY

85

instrument control is speed. Analytical instruments which sample at high rates are thus well served by microprocessors. Instruments which perform lengthy calculations like Fourier transforms, integrations, peak detections, and linear to log conversions also effectively utilize the high speed of the microprocessor. The dedicated microprocessor nevertheless has several disadvantages in a laboratory environment. It must be programmed in machine language unless special program development devices are available. Programming in machine code is difficult and time consuming for the scientist. Machine codes are microprocessor specific so that the codes cannot easily be transported to other machines. Microprocessors often require the expertise of biomedical instrumentation and electronics shops. These are not always available at smaller institutions and their services can be quite expensive. In short, microprocessors are complicated devices which many scientists find difficult to understand. They often lack flexibility in programming and interfacing to peripherals. They are what the computer industry calls “unfriendly” devices. Microcomputers, by contrast, are friendly and have tremendous flexibility. They can be used as general purpose instruments in a variety of tasks. A microcomputer is a microprocessor-based CPU (central processing unit) to which is added read only memory (ROM) and random access memory (RAM), an input multiplexer, address and data buses, buffers for temporary data storage, a clock for timing functions, and control logic (Fig. 1) (Shapiro et al., 1976). Standard peripherals include a CRT display, keyboard, and some form of mass storage, usually a flexible (floppy) or hard disk. An operating system manages memory allocation and mass storage. One or more high-level languages are resident in ROM or software. Various interface cards are available for communication with external devices. A number of software packages for statistical analysis, graphics, terminal emulation, word processing, data base management, and numerical analysis are usually also available at a modest additional cost. These features offer numerous advantages to the scientist. Input-output control of peripheral devices is handled “automatically” by the microcomputer. A few elementary high level language commands conveniently drive printers, plotters, and graphics devices. Keyboard input, CRT display, memory allocation, and mass storage control are managed automatically by the microcomputer. Housekeeping operations such as file maintenance and program editing are handled using the computer’s operating system. High level languages are another major advantage of microcomputers. Most computer systems offer several interpretive or compiled languages, including BASIC, FORTRAN, PASCAL, and C. An ASSEMBLY language is often available as well. There is thus no need for the scientist to master machine or object code. Program development is much faster using high level languages. Five times fewer lines of code are required compared to ASSEMBLY (Brooks, 1975). Sophisticated editors and program debugging routines further simplify program-

86

R . RANNEY MIZE

Pi(, I . Rloch tli;ipr:iiii 0 1 the coiiiponeiit\ of an lritcl XOXO-based iiiicroconiptiter, including tlit' niicroptocehsor, real-tinic clock. input iiiiiltiplcxcr. arid address and data buws with as%ici;ited htilt'er~.RAM iiiitl KOM iiiciiiory. control logic. and an adtlress decoder lor peripheral intcrfacri art' also showii (Modii'icd fro111 Shapiro P I d / ., 1976. by periiiisbiori of Aiiirtrtr/ R e i ~ t w . \ , Palo Alto. c;lll~~~l-r~l~l.~

niing. High level languugc progrnmming is oltcn less cxpcnsivc since students willing to work for $4- 10 a n hour ol'ten know several o f these languages. There nre other vdvuntages to high level language programniing. Programs are easier to unclcrxtand and modify because high level languages require iiiore structure than do machine o r asscmbly cotles. High level l a n g i q e s are rnore easily transported from machine to machine, so programs can bc tnunsterred to other laboratory sltec. The most serious clisadvantage of the niicrocomputer is its slow data ncquisition spccd. Compiled languages such as FORTRAN may limit acquisition rates to 5 kl-lz. Interpretive BASICS may slow acquisition to SO Hz (Saiin. 19821). The slow exccution tiines of these languages can be partially overcome by writing instrument control and data acquisition subroutines in ASSEMBLY. Many microcomputers allow insertion of ASSEMBLY subroutines which high


87

level language programs can access. Programmable interrupts can speed data acquisition by allowing the computer to be interrupted by the peripheral when data are ready. Data collection speed can also be increased by using direct memory access (DMA) IiO. Using this mode, binary data can be collected from instruments and stored directly in memory at rates up to several hundred kHz, bypassing CPU control. The data can be converted, reduced, and operated on by the CPU when the acquisition process is complete. High speed data acquisition is thus quite feasible using microcomputers if special programming techniques are employed. There are a vast number of microcomputer systems available on the market today and the choice of a computer is often a difficult and bewildering process for the researcher. Some of the most popular microcomputers available today are listed in Table I . The table lists the computer’s CPU, memory options, the inputoutput ports available, the graphics resolution, operating system, and available languages. Although the list is not complete, it should nevertheless serve as a useful guide for potential microcomputer owners. A typical microcomputer system used in our laboratory is shown in Fig. 2. This Hewlett-Packard 9845T desk-top microcomputer has dual 16-bit NMOS-I1 processors, 187 kbytes of core memory, a medium resolution graphics CRT, built-in tape cartridge drives for mass storage, and a thermal line printer. The computer is interfaced to a digital plotter, an 8-in. flexible disk storage device, and a high-resolution digitizer (Fig. 2 ) . This computer was one of the earliest “micros” produced for use by scientists. We chose the computer for its bit-map graphics capabilities, ease of programming in BASIC, portability, and availability of peripherals, particularly the digitizing tablet. Although comparatively expensive by today’s standards, it has many of the desirable features of less expensive microcomputers available on the market now (Table I). One of the major advantages of the system for our purposes was the enhanced BASIC language. We initially had no expert programmers in our group and the BASIC language was easy to learn, simple to debug, and combined some powerful features of other high level languages. The H-P enhanced BASIC includes multicharacter variable names, I/O and graphics command sets, subprogram capabilities, labeled common, prioritized intempts, character string manipulation, and 6 dimensional numeric and string arrays. PASCAL-like statements provide a structured programming environment. Software can be written as independent modules using callable subprograms which are linked to the main program, similar to FORTRAN. The only major disadvantage of enhanced BASIC is that it is an interpretive language and thus executes quite slowly because the interpreter must translate each statement into machine code during program execution. Compiled languages execute far more rapidly because they are translated to machine code

SELtClEU -

~~

System

TABLE I 8 A N D 16 BIT MICROCOMPUTERS I-OK LABORATORY USE

CPU

Min RAM

Max RAM

1/0port\ dvalhhk

Graphics rewlution

Parallel RS-232

280 X 192 12 in. 640 x 240 12 in.

Applc 11E

6502

64K

I28K

TRS-SO Mudel 12

ZXOA

XOK

128K

Parallel

RS-232

H-P 85B

HP X hit or Z80A

32K

544K

ADVANTAGE

%XOA/HOXX

fJ4K

2S6K

128K

7.4M

8/16

H-P 16

DEC Rainbow

M6XOGil

1CK)

IBM PC XT

UEC PRO 325-350

TRS-80 Mvdel I6

ZXOA/808X

h4K

256K

8088

64K

64OK

F-I 1 (I I/23)

M68000i

64K

5 12K

64K

5 12K

128K

1M

Z80A

IBM INST

cs-9ooo

M6X000

192 X 256 I’arallcl ICBE-~X~ 5 in. RS-232 BCD Parallel 640 X 240 RS-232 12 in.

Parallel RS-232C IEEE-488 BCD RS-423 RS-232 Parallel KS-232

300 x 400 9 in.

Parallel

960

RS-232 IEEE-488 RS-423 Parallel RS-232 Parallel RS-232 1liEE-4XX

800 X 240 12 in. 300 X 400 12 in. X 140 12 in.

Operating systcin

Manufacturcr

BASIC PASCAL. FORTRAN C BASIC BASIC BASIC 1-ORTRAN PASCAI.

Apple Coinpuler, Cupertino, CA Tandy Corp. Radio Shack, Ft. Worth, TX

MS-DOS CPIM-80 GRAPHICS-DOS HP-BASIC CPiM-68K

BASIC- I6 FORTRAN-16 PASCAL- 16

North Star Computer, San Leandro, CA

BASK

Hewletl-Packard, Loveland, CO

CPIM-86/80 MS-DOS IBM-DOS CPlM 86 UCSD-P POS

M BASIC

CPIM-80 DOS TRS-DOS CP/M-BO H-P BASIC CPIM-SO UCSD-p

(RSX-11) UCSD-p

640 x 240 12 in. 768 x 4x0 12 in.

Languages availablc

TRS-DOS

MlJLTITASKlNG RT 0.5

P A X AI HPI.

~

c

BASIC

FORTRAN PASCAL BASIC PASCAL C BASIC FORTRAN BASIC PASCAL FORTRAN

Hcwlett-Packard, Loveland. CO

Digital Equipment, Maynard, MA IBM, Armonk, NY

Digital Quipmenr, Maynard. MA

Tandy Corp, Radio Shack, Ft. Worth, TX IBM, Arinonk, NY


89

FIG. 2. Microcomputer system. The niicrocomputer (Hewlett-Packard 9845T) includes a 560 X 455 bit-map graphics CRT (A). a thermal line printer (B), and two magnetic tape cartridge mass storage drives (C). The computer is interfaced to a digitizer (D, H-P Model 9874A). a 4-color digital plotter (E, H-P Model 9872A). and a dual-density, double sided floppy disk drive (F, H-P Model 9895A). (From Street and Mize, 1983, by permission of Elsevier Biomedical Press, Amsterdam.)

before the program is executed. The most popular compiled languages for scientific use are FORTRAN, PASCAL, and C. These languages are now usually available on most microcomputers (Table I). Software availability was also an important concern when we chose our microcomputer system. We found Hewlett-Packard software available for many laboratory functions. For example, we use Hewlett-Packard statistical packages for statistical analysis, editing, display, data formatting, and printing of our data. The statistical analysis software includes data editing, summary statistics, analysis of variance, linear and nonlinear regression, distribution analyses, and other parametric and nonparametric tests. We use a statistical graphics package for plotting data on the graphics CRT or digital plotter. Time interval plots, histograms, log-log plots, scattergrams, and 3-D plots are available using this package. Various ‘‘office’’-oriented software makes our microcomputer of great value in laboratory management and manuscript writing. We use a word processing package to prepare manuscripts; a terniinal emulator package for asynchronous communication with the university mainframe computer; data base management

90

R. RANNEY MIZE

software to enter, sort, search, and retrieve reprints; and an electronic spreadsheet program to manage laboratory purchases and expenditures. The 9845T is used to run graphics presentation software for preparation of figures, slides, and posters for scientific meetings and journal articles. Despite the availability of these commercial packages, we found it necessary to write our own specialized data acquisition and analysis software for microscope plotting, digitizing, and serial section reconstruction. The following sections describe these computer programs which we use in neuroanatomical research. Many other uses of computers in cell and neurobiology research which have been reported in the literature are also reviewed.

111. Computer Systems €or Microscope Control and Plotting

Computers can be used to control a variety of light microscope functions (Boyle and Whitlock, 1975). The microscope stage can be controlled by a computer if stepping motors are attached to the stage drives. The computer can both advance the stage in small increments and simultaneously keep track of stage position. Focusing can also be controlled by a computer if a drive device is attached to the focus knobs. Algorithms are available to automatically focus light microscopes. Several computer-assisted approaches to automated focusing have been described in the literature (Mason and Green, 1975; Ploem et al., 1979; Shoemaker et al., 1982). By adding a TV camera, photometer, or other imaging device, computers can also be used in analysis of microscope images (see Sections V1, VII, and VIII). Computers are also being used to control functions on electron microscopes (Engel et al., 1981; Herrmann et al., 1978; Hillman et al., 1980; Joy, 1982; Kirkland, 1982; McCarthy et a / . , 1982; Rez and Williams, 1982; Rust and Krahl, 1982; Smith, 1982; Statham, 1982). For example, a number of computer systems have been designed for control of STEMS (scanning transmission electron microscopes). Joy (1982), for instance, has designed software which monitors the operating conditions of a STEM. The system uses an Apple 11 Plus microcomputer with A / D and D/A converter cards. APPLESOFT BASIC programs provide a CRT readout of lens current, gun and specimen chamber vacuums, and stage position. Other applications for image storage and enhancement are discussed as well (Joy, 1982). More elaborate control systems are also available, some of which are sold commercially. These systems are particularly useful for controlling the scanning beam of a STEM in automated X-ray analysis (Herrmann et al., 1982; McCarthy et al., 1982; Rust and Krahl, 1982). Computers have been used for some time for microscope plotting. Computerbased plotters are used to produce maps of tissue being analyzed with the light microscope. The maps are usually produced by coding the stage position of the


91

microscope. The earliest of these computer microscope plotters was developed by Glaser and van der Loos ( 1965) almost 20 years ago. Computer-aided microscope plotters fall into two categories: dendrite-tracking systems used to study the dendritic branching patterns of cells, and microscope plotters used to map the spatial distribution of organelles within tissue. The dendrite-tracking systems plot the tree structures of neurons and measure such features as branch number, dendrite vector. and dendrite diameter. From these measurements, mathematical models of the functional characteristics of the cells can be generated. There are essentially two types of dendrite-tracking computer system (Capowski and Cruce, 1979). The first type uses an encoding device on the microscope stage to detect the position of dendritic branch points. The second type uses a projected image of the dendrite which is digitized by an external device. Capowski (1977), Capowski and Sedivec (1982). DeVoogd et ul. (1981), Overdijk et a / . (1978), and Wann et a / . (1973) have developed dendrite-tracking systems which record the positions of dendritic branches by encoding stage position. Encoding is accomplished on these systems by attaching stepping motors to the stage drives of the light microscope. Counting the pulses generated to drive the stepping motors and converting these counts to distance provides a measure of stage position. Z axis data to encode the depth of a profile within the tissue is taken from the fine focus knob of the microscope, which can also be fitted with a stepping motor. Branch positions are usually recorded by depressing a function key on the computer when the branch point lies under a cross-hair in the microscope binoculars. Other computer microscope systems use potentiometers or shaft encoders rather than stepping motors to encode position (Glaser and van der Loos, 1965; Mize, 1983a; Reed et al.. 1980). With linear potentiometers, position is read as an analog signal in which voltage is proportional to stage position. A I D converters translate the voltage signal to a digital value for computer input. With shaft encoders, the output pulses of the encoders are counted and translated to a unit of measure to represent stage position. In the second type of microscope tracking system, the microscope image is projected to another surface and measurements are taken from that surface. Such systems include images superimposed by a camera lucida onto a digitizing tablet (Green et a / . . 1979; Haug, 1979) and images transmitted to a television monitor via a video camera (Hillman et a l . , 1977; Lindsay, 1977; Paldino, 1979; Paldino and Harth, 1977; Uylings et a / . , 1981; Yelnik et u l . , 1981). Branch points of dendritic processes are digitized by positioning a CRT screen cursor or digitizer cursor over the branch point and digitizing the point. The completed “stick” reconstructions of the neuronal tree structure are displayed on the CRT. Various measurements can be made from these stick reconstructions, including dendrite diameter, distance between branches, branch length, the number of branches, and branch vectors. These measurements are computed automatically using spe-

92

R. RANNEY MIZE

cially developed algorithms. The dendrite-tracking systems have been useful in distinguishing cell classes (Glaser et a l . , 1979; Uylings et a l . , 1981) and in revealing parametric alterations in cell populations after various pathologies or experimental manipulations (Woolsey and Dierker, 1979, 1982). The video-based microscope plotters are faster than those which utilize stage encoders, but image quality and resolution are often sacrificed when the image is projected. Both video and encoder-based microscope plotters are susceptible to errors in the estimation of tissue depth, as Glaser (1982) has pointed out. When using dry objectives, objects will appear foreshortened in the Z axis unless a correction factor is used. This is a common but easily rectified problem in computer microscopy. Automatic video image analysis systems have also been developed which have algorithms for automatic “recognition” of Golgi impregnated or HRP filled neurons (Capowski, 1983; Coleman et al.. 1977; Garvey et a l . , 1973). This allows the computer to locate and trace the neuron so that branch points do not have to be plotted manually. However, the complexity of nerve cells as well as irregularities in staining density mandate close operator monitoring of the procedure to reduce computer misinterpretation of artifact. Other computer microscope systems are specifically designed to map the spatial distributions of organelles within a tissue specimen. These are called computer microscope plotters or pantographs (Curcio and Sloan, 1981; Foote et a l . , 1980; Forbes and Petry, 1979; Mize, 1983a; Reddy et a l . , 1973; Reed et al., 1980; Williams and Elde, 1982). These systems can be used to accurately plot the positions of labeled cells, other organelles, and various reaction products within a tissue sample. Our laboratory, for instance, has developed a microcomputer-assisted plotter for the electron microscope which maps the positions of various profiles within a section of tissue and calculates their distance (in microns) from the surfaces of the section (Mize, 1983a). The densities of the profiles within the tissue can also be calculated with the system. Stage position is measured using optical incremental shaft encoders which are attached to the stage drives of the electron microscope by gears. A display/control unit converts the encoder pulses to binary digits as well as providing an LED display of stage position. The 9845T microcomputer is used to control data input and to store, graph, and analyze the plots (Fig. 3). The software for the system includes four programs: (1) truce, used to draw around the outer edges of the tissue specimen; (2) plot, which maps the positions of profiles (synapses, cells) within the boundaries of the tissue; ( 3 ) analyze which includes algorithms for comparing trace and plot data and calculating distance from the tissue surfaces; (4) density, for sorting and counting profile types, measuring surface areas, and calculating profile densities. Commercial statistical software is used to analyze and graph the data and study profile distributions. The maps can be generated in four colors on a digital plotter (Fig. 4). The maps have demonstrated statistically significant


93

FIG.3 . Block diagram of an electron microscope plotter, which includes shaft encoders (Sh. Enc.) attached to the stage drives of the microscope, a display control box which converts encoder pulses to 16-bit parallel digital code, two 16-bit parallel interface cards to input the signals to the computer, and the H-P 9845T microcomputer, which includes a printer, CRT, plotter, and floppy disk. (Modified from Mize, 1983a. by permission of Elsevier Biomedical Press, Amsterdam.)

differences in the distributions of synapse populations which overlap qualitatively (Mize, 1983~).A microcomputer plotting system which also uses a Hewlett-Packard 9845 computer has been described by Williams and Elde ( 1982) for mapping the distribution of histochemical label in brain slices. Other mapping systems have been designed to plot the three-dimensional distributions of profiles through a volume of tissue. Foote et al. (1980), for instance, have developed a PDP 11/34 computer system with an Evans and Sutherland Picture System 11 graphics processor for this purpose. The system produces three-dimensional maps of neurons within brain nuclei. Comparisons of the distributions of cell groups in different brains can be made quantitatively using the system. Other computer-based plotters for producing three-dimensional maps have been reported (Curcio and Sloan, 1981; Johnson and Capowski, 1973).

IV. Serial Section Reconstruction The three-dimensional reconstruction of biological tissues is of great value both for analyzing the volume of structures and for modeling their three-dimensional molecular configuration. Three-dimensional information can often be ex-

Fic. 4. Computer-plotted map of synaptic terminals within the cat superior colliculus. The plotter outlines the tissue contours of the spccimen. The positions of retinal ( * ) and cortical (0) synapses are indicated by symbols. (From Mize. 1 9 8 3 ~ .by permission of Springer-Verlag. New York.)


95

tracted from single sections using stereo pair imaging or optical slice techniques (see Turner, 1981, for an extensive review of this topic, and Ghosh, 1975). Reconstruction from multiple serial thin electron microscope sections is often useful when high resolution is needed, or when the object to be reconstructed is large. Several tasks are required to reconstruct tissue from serial sections. Sections must be stacked, aligned, and displayed. In addition, rotation of the reconstructions is useful for studying their shape and spatial relationships. The history of serial section reconstruction has been reviewed by Gaunt and Gaunt (1978), Mannen (1978), and Ware and LoPresti (1975). Computer-based reconstruction has been reviewed by Macagno et u / . (1979) and Sobel et ul. (1980). A number of computerized electron microscope reconstruction systems were developed during the 1970s, most using specially fabricated instrumentation and minicomputer hardware (Glasser et ul., 1977; Hillman et al., 1977; Llinas and Hillman, 1975; Lubbers, 1977; Macagno et al., 1979; Perkins et al., 1979; Rakic et ul., 1974; Shantz and McCann, 1978; Veen and Peachey, 1977). Paralleling this development of reconstruction hardware were advances in software algorithms for reconstruction and image rotation (Dierker, 1976; Gentile and Harth, 1978; Gordon and Herman, 1977; Kam, 1980; Newman and Sproull, 1979; Veen and Peachey, 1977). Perhaps the best known reconstruction system introduced during this period is CARTOS (Computer Aided Reconstruction and Tracing of Sections), developed by Cyrus Levinthal’s group at Columbia University. The CARTOS system uses a PDP 11/34 computer and an Evans and Sutherland graphics display device. Profiles from serial sections can be entered in the computer system using various digitizing devices (Levinthal and Ware, 1972; Macagno et al., 1979; Sobel et a / . , 1980). A portable loaner system for CARTOS is available for use through a shared instrument resource grant from NIH. The portable system allows investigators to digitize serial sections in their own laboratories. Real-time rotation and analysis of the reconstructions is then performed at the home laboratory site in New York. Stevens et al. (1980) have developed a similar system for neuron reconstruction. Their system employs 35 mm film strips which have been produced from serial electron microscope negatives. The film strips are aligned and analyzed under computer control. The films are viewed with a video camera and displayed on a monitor. Advance of the film strips, fine alignment of the frames, and digitization of profiles are controlled by a specially designed Z80 microprocessor-based unit. The film is surveyed at different magnifications using a video zoom camera. The outlines of reconstructed profiles are digitized directly on a video screen by moving a screen cursor around the profile. Consecutive sections are microaligned by comparing one section with a previous section stored in a video memory. To align, the current and stored images are superimposed on a video monitor. The images are then rapidly alternated on the video

96

R. RANNEY MIZE

screen and one image moved with a joystick until alignment is achieved. Ideal alignment is obtained when apparent movement (“flicker”) is minimized. The logically aligned film strips can be replayed under computer control. Image rotation is accomplished using a DEC graphics package on a PDP 11/34 which runs specially developed rotation software. This approach is quite powerful. The film strips offer a convenient, easily stored record of the reconstructions. The alignment and digitization of the films are relatively rapid. The reconstructions can be reviewed and edited easily. On the other hand, the video hardware is expensive and producing the 35 mm movies is both time consuming and costly. Simpler, less-expensive systems have recently been developed which use microcomputer-based manual digitizers and graphics display devices. The digitizer is used for data entry, the graphics unit is used for aligning and displaying sections, and a floppy or hard disk is used for mass storage (Glasser et al., 1977; Johnson and Capowski, 1983; Macagno et ul., 1979; Perkins et al., 1979; Prothero and Prothero, 1982; Street and Mize, 1983). Our system, for instance, uses the Hewlett-Packard digitizer to trace outlines of cells or synaptic profiles within each serial section. The outlines are traced from 8” x 10” photographs, although projected slides or movies can also be used. Once digitized, two consecutive digitized outlines are displayed on the graphics CRT of the 9845T computer. The sections are aligned by translating and rotating one of the outlines with special arrow keys until it is superimposed over the other outline. The process works well although it is somewhat slow because the graphics screen must redraw each translation. Where there are slight tissue distortions we approximate a “best fit” by eye rather than using a complicated algorithm to match the sections. This works well since the human eye seems well-adapted to making these judgments rapidly. The procedure is continued until all sections are aligned. The X-Y coordinate values of the aligned profiles are then stored on floppy disks. A rotation algorithm allows us to display the reconstructed cells or synapses at different rotations in space with hidden lines removed (Fig. 5) (Street and Mize, 1983, 1985). Our system is programmed in enhanced BASIC. Because the H-P enhanced BASIC is an interpretive language, the rotation algorithms are slow. A solution to this problem is provided by graphics display systems (Glasser et al., 1977; Johnson and Capowski, 1983; Macagno et al., 1979). These devices have hardwired logic for rotation and translation and therefore provide real-time interactive graphics capability. Johnson and Capowski (1983), for instance, use a Neuroscience Display Processor for rapid rotation and translation of reconstructed images. This refreshed vector graphics system is driven by a PDP 11/45. Most of their reconstruction software is written in FORTRAN with callable assembly language subroutines for controlling peripheral devices. Another 3-D reconstruction FORTRAN program for entry and display of images (but with no facility for


97

FIG. 5. computer reconstruction of a retinal synapse in the cat lateral geniculate nucleus. The synapse has been rotated in three planes. Hidden lines have been removed to give a three-dimensional perspective to the reconstructions. Symbols represent regions of synaptic contact with other cells.

rotation) has been developed for microcomputers running under the CP/ M operating system (Prothero and Prothero, 1982). Computerized reconstruction has been used to study the dendrite structure and spatial relationships between neurons (Glasser et al., 1977; Llinas and Hillman, 1975; Macagno et al., 1979; Stevens et al., 1980; Street and Mize, 1983), the density of synaptic contacts on neurons (Mize et al., 1982; Stevens et al., 1980; Street and Mize, 1983), to develop structural models for viruses and other microbiological organisms (Perkins et a/., I979), and to examine the structure of mitochondria (Tenny et a l . , 1980; Veen and Peachey, 19771, chromosomes (Moens and Moens, 1981), and even single molecules (Perkins et a l . . 1979). Computerized reconstruction offers a major technical advance over manual techniques. The computer can be used to manipulate, measure, and store sectional data automatically. Computations for rotation can be executed rapidly with the computer. The reconstructions can be graphically displayed on computerbased graphics devices. None of these procedures is possible using manual techniques.

98

R. RANNEY MIZE

V. Computer-Aided Morphometric Measurement Quantitative analysis of cytological materials is becoming increasingly common in cell and neurobiology . Measurement of various geometric characteristics of cells and organelles is called morphometry. Planimetry and stereology are systematic techniques used in morphometry for measuring parameters such as object density, size, shape, and volume quantitatively. The introduction of microcomputer-based measuring systems greatly facilitates the application of these techniques to microscope materials. Microcomputer systems increase the speed and accuracy of data acquisition, allow for much more efficient mathematical analysis, and provide a convenient source of data storage and display. Computerized planimetry is a direct method for measuring such geometric features as cross-sectional area, perimeter (length), diameter, and shape. Two computerized approaches are used in planimetry: computerized analysis in which image data are entered manually with a digitizing tablet and semiautomatic and automatic video image analysis in which the image is reconstructed electronically. Digitizing tablets are electromechanical devices which have a tablet surface or platen usually embedded with an electrically active wire grid (Fig. 2). The wire grid is electrically referenced to a cursor or stylus (pen). The user collects data by tracing around the outer contours of profiles using the cursor (or an electronic stylus pen). The cursor “senses” the coordinate values representing the profile’s outline and transmits these values to the computer. The computer converts the values to measures of area, length, diameter, and shape. The digitized image can be a micrograph placed on the platen or a back projected slide or movie frame. The coordinate sensing techniques of early digitizers were mechanical (Veen and Peachey, 1977) or sonic (Cowan and Wann, 1973; Dunn e t a / . , 1975, 1977) but most modern digitizer tablets employ electrical wave sensors (Hewlett-Packard, Summagraphics, Talos). Digitizing tablets can be small and inexpensive (circa $300) or large and expensive (circa $10,000). Small data or graphics tablets used to lack resolution and accuracy and were most often used to manipulate cursors or data on a CRT. However, recently developed small tablets have excellent resolution and are quite accurate. Precision measurement digitizers have resolutions of up to 25 p m and often include special features such as function keys and LED displays. Many digitizers have built-in microprocessors which do much of the maintenance work of the digitizer (signal conversion from voltage to digital x, y coordinate values, origin setting, scaling, axis alignment, and skew correction). Digitizers can be interfaced to a microcomputer via any of the standard interface ports (the GP-IB IEEE-488 instrument interface, 8 or 16-bit parallel, and RS232 serial interfaces are most common). The x, y coordinate positions of each data point are usually transmitted to the microcomputer as binary numbers, although ASCIl values are also sometimes used. The microcomputer is used to convert

COMPUTER APPLICATIONS IN NEUROBIOLOCY

99

these data points to decimal values, to perform geometric calculations, and to store the data. The control of data input, storage, and calculation can be managed by assembly language routines or by high level language programs. A number of high level language programs for digitizing morphometric features have been described in the literature, Several FORTRAN programs have been written to drive digitizers interfaced to PDP (DEC) computers (Albright and Sawler, 1981; Cowan and Wann, 1973; Dunn e f al., 1975, 1977). BASIC language programs have also been developed for a variety of microcomputer digitizing systems (Dennino et a / ., 1978; Mize, 1983b; Pullen, 1982). Digitizing systems written in assembly language have also been reported (Green et al., 1979; Peachey, 1982). All of these programs have algorithms for calculating area and perimeter. Many also calculate other parameters such as diameters, Feret dimensions, angles of orientation, center of gravity, and various form or shape factors. Formulas for calculating some of these values are reported by Bradbury (1977). A large variety of inexpensive commercial morphometric digitizing systems are now available on the market (Table 11). Peachey (1982), for instance, describes a simple single board microprocessor-based system for about $2000. The system includes a 280 processor with clock, parallel I/O ports, 1K byte of RAM, and several EPROM (erasable programmable read only memory) boards. This single board computer is interfaced via an 8-bit parallel port to a Summagraphics BITPAD. The programs are written in assembly language and are “stored” on the nonvolatile EPROMS. The programs convert data points, calculate positive and negative areas and length, and print the results on an inexpensive printer. A more elaborate version of this system is available from Laboratory Computer Systems (Cambridge, Mass., Table 11). Other more expensive digitizing tablet systems are available commercially, some of which include flexible general purpose microcomputers (Table 11). Much of the cost of these systems is for the software, as the hardware is generally relatively inexpensive. We have developed a noncommercial, highly flexible digitizing system based around our H-P 9845T microcomputer and H-P 9874A digitizer. The programs are written entirely in H-P enhanced BASIC (Mize, 1983b). The system has been used to measure cell and synapse areas, to measure synaptic vesicle sizes and shapes, to measure the length of immunocytochemically stained collagen fibers, and to measure the size and density of intramembrane particles on freeze-fractured replicas. When measuring cells, the programs allow entry of data values for the number of elements contacting a cell, the number of elements within a cell (such as autoradiographic grains), cell type, and cell depth. These values are entered from the digitizer’s numeric keypad (Fig. 2). The operator then traces around the profile with the digitizer’s cursor. From the trace, the programs calculate cross-sectional area, perimeter, average diameter, and form factor (an index of circularity). The programs also compute contact and element densities

100

R. RANNEY MIZE TABLE I1 SELECTED COMMERCIAL IMAGEANALYSIS INSTRUMENTS

System

Computer

Resolution

Measurementsa

MICROPLAN

Microprocessor

0.025 MM

A,L,P,F,D,S,C

OPTOMAX

Apple IIE

0.10 MM

LADD 40000

Rockwell AIM 65

0.10 MM

A,L,P,F,D,S,C ANGLE, others A,L,P,F,D,S,C others

NUMONICS 1224EM

Microprocessor (8080)

0.25 MM

A,L,P,F,D,S,C ANGLE, others

MOP-30

Microprocessor

MM

A,L,P.F,D,S,C ANGLE, others

Manufacturer/ distributor

Digitizing tablet systems

International Review of Cytology: A Survey of Cell Biology, Volume 90

International Review Of Cytology: A Survey of Cell Biology

International Review of Cytology: A Survey of Cell Biology, Volume 205

International Review of Cytology: A Survey of Cell Biology, Volume 254

International Review of Cytology: A Survey of Cell Biology, Volume 198

International Review of Cytology: A Survey of Cell Biology, Volume 139

International Review of Cytology, Volume 185: A Survey of Cell Biology

International Review of Cytology: A Survey of Cell Biology, Volume 222

International Review of Cytology: A Survey of Cell Biology, Volume 11

International Review of Cytology: A Survey of Cell Biology, Volume 209

International Review of Cytology: A Survey of Cell Biology, Volume 105

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

International Review of Cytology: A Survey of Cell Biology, Volume 239

International Review of Cytology: A Survey of Cell Biology, Volume 57

International Review of Cytology: A Survey of Cell Biology, Volume 95

International Review of Cytology: A Survey of Cell Biology, Volume 35

International Review of Cytology: A Survey of Cell Biology, Volume 114

International Review of Cytology: A Survey of Cell Biology, Volume 233

International Review of Cytology: A Survey of Cell Biology, Volume 113

International Review of Cytology: A Survey of Cell Biology, Volume 195

International Review of Cytology: A Survey of Cell Biology, Volume 144

International Review of Cytology: A Survey of Cell Biology, Volume 118

International Review of Cytology: A Survey of Cell Biology, Volume 106

International Review of Cytology: A Survey of Cell Biology, Volume 48

International Review of Cytology: A Survey of Cell Biology, Volume 264

International Review of Cytology, Volume 195: A Survey of Cell Biology

International Review of Cytology: A Survey of Cell Biology, Volume 8

International Review of Cytology: A Survey of Cell Biology, Volume 128

International Review of Cytology: A Survey of Cell Biology, Volume 263

International Review of Cytology: A Survey of Cell Biology, Volume 153

International Review of Cytology: A Survey of Cell Biology, Volume 90