Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Role of the cytoskeleton and substratum in cell topographic guidance Oakley, Carol 1995

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_1997-196348.pdf [ 46.52MB ]
Metadata
JSON: 831-1.0088060.json
JSON-LD: 831-1.0088060-ld.json
RDF/XML (Pretty): 831-1.0088060-rdf.xml
RDF/JSON: 831-1.0088060-rdf.json
Turtle: 831-1.0088060-turtle.txt
N-Triples: 831-1.0088060-rdf-ntriples.txt
Original Record: 831-1.0088060-source.json
Full Text
831-1.0088060-fulltext.txt
Citation
831-1.0088060.ris

Full Text

Role of the Cytoskeleton and Substratum in Cell Topographic Guidance By Carol Oakley D.D.S., University of Alberta, 1978 A THESIS SUBMITTED IN PARTIAL F U L F I L M E N T OF THE REQUIREMENTS FOR THE D E G R E E OF DOCTOR OF PHILOSOPHY in THE F A C U L T Y OF G R A D U A T E STUDIES (Department of Oral Biology) We accept this thesis as conforming to the required standard  T H E UNIVERSITY OF BRITISH C O L U M B I A July 1995 ©Carol Oakley, 1995  In presenting this degree  thesis  in partial  fulfilment  of the requirements  for an advanced  at the University of British Columbia, I agree that the Library shall make it  freely available for reference and study. I further copying  agree that permission for extensive  of this thesis for scholarly purposes may be granted  department  or  by his or  her representatives.  It  by the head of my  is understood  that  copying or  publication of this thesis for financial gain shall not be allowed without my written permission.  Department  of  Oral  Biology  The University of British Columbia Vancouver, Canada  Date  DE-6 (2/88)  September 26,  1995  11 ABSTRACT Topographic guidance refers to reactions of cells with the topography of their substratum and includes alterations in cell shape, orientation and polarity of movement. Current hypotheses focus on the cytoskeleton although most observations have been limited to cells that were already oriented with their substratum topography. In this thesis I addressed this shortcoming, tested the current theories of contact guidance and hypothesized that there may be a prime or principal cytoskeletal determinant of fibroblast orientation on micromachined grooves. I predicted that the principal cytoskeletal determinant of contact guidance would be the first cytoskeletal component to become aligned with the topography, would be most sensitive to size and arrangement of the topography and would be able to exert its influence independently. Furthermore, if the principal determinant was also the sole determinant, then contact guidance would not occur in its absence. Finally, the principal component would operate in other cell types.  The hypotheses and predictions were tested directly by examining the development of cytoskeletal alignment in relation to overall cell shape and alignment as fibroblasts spread on micromachined grooved substrata. Time-lapse cinematography, conventional and confocal epifluorescence microscopy and computer-assisted morphometry indicated that microtubules were the first cytoskeletal element to become aligned and microtubule alignment preceded alignment of the cells themselves.  Cell spreading and alignment in the presence of specific inhibitors of microtubules (colcemid) and actin microfilament bundles (cytochalasin) on micromachined grooves that differed in size demonstrated that the integrity of either the microtubule or actin system could be sufficient to produce alignment with an appropriate substratum topography but cells deficient in microtubules required significantly more time to accomplish these behaviours. These experiments demonstrated that microtubules were the principal but not the sole cytoskeletal determinant of contact guidance on micromachined grooves. Microtubules responded faster and to smaller topographic features than  microfilaments but unless the topography elicited alignment of both the microtubule and actin systems, the influence of microtubules upon cell alignment was tempered. I concluded that the substratum topography could selectively or differentially effect alignment of the microtubule and actin systems so that cell alignment and migration could be enhanced or retarded.  iv T A B L E OF CONTENTS Abstract  ii  Table of Contents  iv  List of Tables  xiii  List of Figures  xiv  Nomenclature  xviii  Preface  xxi  Acknowledgements  xxii  C H A P T E R 1 - INTRODUCTION I. Overview U. In Vitro Cell Adhesion to Artificial Substrata A . Physical Properties of the Substratum 1. Mechanical Rigidity 2. Cleanliness of the Surface 3. Size of Available Adhesion Sites B. Directed Orientation and Migration of Cells by the Substratum 1. Chemical or Adhesive Guidance a. Chemotaxis b. Galvanotaxis c. Hap to taxis d. Bidirectional Adhesive Cues e. Population Pressure 2. Physical or Topographic Guidance 3. Interactions between Directional Cues  1 3 3 3 4 5 6 6 6 7 7 8 9 10 11  III. In Vitro Topographic Guidance A . The Role of Artificial Substratum Composition, Dimensions and Geometries 1. Non-grooved Substrata 2. Grooved Substrata  12 14 15  B . The Response of Different Cell Types to Topographic Guidance  17  IV. Proposed Mechanisms for Topographic Guidance A . Extra-cellular Mechanisms - Microexudate Hypothesis  12  20 20  B. Cytoskeleton-based Hypotheses 1. Microfilament Hypothesis a. Convex Curvatures b. Concave Curvatures 2. Focal Contact Hypothesis 3. Microtubule Hypothesis  21 21 21 23 24 25  C. Stochastic Hypothesis  26  V . The Cytoskeleton A. Microtubules 1. Structure 2. Organization and Assembly a. Microtubule-organizing Center i. fibroblasts ii. epithelium b. M T Assembly 3. Post-translational Modifications a. Phosphorylation b. Glutamylation c. Acetylation d. Tyrosination/Detyrosination e. Detyrosination/Deglutamylation 4. Proteins Associated with MTs a. Structural or Nucleotide-insensitive Proteins b. Motor or Nucleotide-sensitive Proteins i. dynein ii. kinesin 5. Interactions of MTs with Other Cytoskeletal Elements and Cellular Components a. Intracellular Traffic b. Intermediate Filaments c. Actin MFs and Focal Contacts i. structural interactions ii. interactions with structural MAPs (a) depolymerization of MTs (b) depolymerization of actin M F bundles iii. possible mechanisms for M A P interactions between MTs and actin MFs 6. Physical and Chemical Interactions with MTs that Affect M T Dynamics a. Effects of the physical environment i. compression and tension effects upon M T assembly ii. tensegrity model of structural organization iii. effect of shear-free planes b. Chemical Agents i. colchicine and its derivatives ii. nocodazole 7. Effects of M T Depolymerization on Cell Behaviour B . Intermediate Filaments 1. Structure and Distribution 2. Interaction of IFs with other cytoskeletal elements and cellular components a. MTs b. Actin MFs c. Plasma Membrane d. Cytoplasmic Components 3. Mechanical Functions of IFs 4. Agents that Affect IFs 5. Effects of IF Collapse on Cell Behaviour  29 31 32 33 33 33 34 34 35 36 36 36 37 38 38 39 41 41 41 43 43 45 45 45 46 47 47 47 48 48 49 49 50 50 51 52 52 54 54 54 54 55 56 56 57 57 57  vi C. Actin 1. Structure 2. Actin Assembly 3. Actin-binding Proteins a. Classification of Actin-Binding Proteins b. Actin Monomer-binding Proteins c. Actin Filament-binding Proteins i. actin filament-severing proteins ii. actin filament-capping proteins iii. actin filament-connecting proteins (a) between actin filaments (b) between actin filaments and myosin (c) myosins (d) between actin filaments, the actin cortex and plasma membrane (i) Direct Interactions (ii) Indirect Interactions (a) spectrin- and ankyrn-like proteins (b) membrane attachment proteins 4. Actin Organization a. Fibroblasts b. Epithelium 5. Interaction of Actin MFs with other Intracellular components 6. Interaction of Actin MFs with the Physical Environment 7. Agents that Disrupt Actin Structure and Dynamics a. Phallotoxins b. Cytochalasins i. mechanisms of action ii. correlation between cell and cytoskeletal behaviours in the presence of cytochalasin D. Cell Attachments to the Substratum 1. Descriptions of Cell Attachment Sites Based on Appearance in Light and Electron Microscopy a. Extracellular Matrix Contacts b. Close Contacts c. Focal Contacts 2. Descriptions of Cell Attachment Sites Based on Macromolecular Structure a. Integrins b. Cytoplasmic Proteins Associated with Focal Adhesions 3. Formation and Maturation of Adhesion Sites 4. Functions of Focal Adhesions VI. Polarization and Locomotion of Tissue Culture Cells A. Terminology B. Role of the Cytoskeleton in Polarization, Formation of Protrusions and Directed Motility 1. Microtubules  59 59 60 61 61 62 63 63 63 64 64 64 65 66 66 66 67 67 68 68 69 69 69 71 71 72 72 73 75 75 75 76 76 77 77 78 79 83 86 87 88 88  vii 2. Actin  90  C. Factors that can Affect Protrusions and Locomotion  94  D. Summary  96  C H A P T E R 2 - STATEMENT OF T H E PROBLEM  97  C H A P T E R 3 - METHODS AND MATERIALS  100  I. Fibroblasts A . Cell Culture 1. Starting Cultures from Gingival Explants 2. Established Fibroblast Cultures  101 101 101 103  B. Culture Conditions 1. Control Conditions 2. Drug Conditions (Cytoskeletal-perturbing Agents) a. Colcemid Conditions b. Cytochalasin Conditions c. Cytochalasin and Colcemid Conditions  103 103 104 104 104 105  C. Observations of Living Cells 1. Seeding the Substrata a. Cell Suspensions b. Droplets of Cell Suspension 2. Effects of Gravity 3. Assembly of Pentz Chambers a. Fabrication of Pedestals b. Mounting of Pedestals and Substratum Wafers c. Final Assembly for Viewing with Upright Microscope d. Final Assembly for Viewing with Inverted Microscope 4. Injection of Cell Suspension or Medium into Pentz Chambers 5. Documentation of Cell Behaviours 6. DICMicroscopy  106 106 106 106 107 107 107 108 108 109 109 110 111  D. Micromachined Substrata 1. Micromachining a. Cleaning b. Oxidation c. Photolithography d. Oxide Patterning e. Final Etching f. Titanium Coating 2. Groove Patterns a. R grooves b. N grooves c. V N grooves 3. Final Preparation of Substrata a. Washing b. Radio-frequency Glow-discharge Treatment  112 112 112 113 113 113 113 113 113 114 114 114 116 116 116  viii E. Preparation of Samples for Examination by Epifluorescence Microscopy 1. Staining to Visualize Cell Morphology 2. Staining of the Cytoskeleton and Associated Elements a. Staining of Microtubules, Kinesin, Vimentin Intermediate Filaments and Vmculk-conteining Focal Contacts i. standard protocol (a) wash (b) fix (c) permeabilize ii. Hollenbeck protocol (a) wash/ permeabilize (b) fix (c) wash/permeabilize iii. common protocol (d) quenching (e) blocking (f) indirect immunofluorescence i. double labelling ii. primary antibodies iii. secondary antibodies b. Staining of Filamentous Actin 3. Mounting of Stained Samples  119 119 121 121 122  122  122 123 124 126 127 128 130  F. Epifluorescence Microscopy 1. Cell Morphology and Quantitation a. Cell Height b. Computer-assisted Morphometry: Projected Area and Form Factor, Form Ell c. Dmax d. Orientation and Polarity 2. Microscopy of Cytoskeletal Elements a. Cytoskeletal Patterns-Descriptions and Criteria for Alignment i. actin ii. vinculin iii. microtubules iv. intermediate filaments  134 134 134 135 135  G. Statistics  135  II. Epithelium A. Cell Culture 1. Starting Cultures from Periodontal Ligament 2. Established Epithelial Cultures B. Epifluorescence Microscopy 1. Cell Morphology and Quantitation a. Single cells and Cells of a Pair or Cluster b. Computer-assisted Morphometry - Projected Area and Form Factors c. Orientation and Polarity C. Statistics  130 131 131 131 132 132 133  137 137 137 138 138 138 139 139 139 140  ix C H A P T E R 4 - RESULTS  141  Introduction and Overview  141  I. The Sequence of Alignment of Microtubules, Focal Contacts and Actin Microfilament Bundles in Fibroblasts Spreading on Smooth and R-grooved Substrata A . Smooth Surfaces 1. Cell Morphology and Quantitation 2. Cytoskeleton and Associated Elements a. Actin b. Focal Contacts (Vinculin Staining) c. Microtubules d. Vimentin Intermediate Filaments e. Kinesin 3. Time Lapse Observations on Smooth Surfaces  144 145 145 145 145 146 146 146 146 147  B . R Grooves 1. Cell Morphology and Quantitation 2. Cytoskeleton and Associated Elements a. Overview of Alignment of Cytoskeletal Elements on R Grooves b. Development of Cytoskeletal Orientation on R Grooves i. Actin ii. Focal Contacts (Vinculin) iii. Microtubules iv. Microtubule Organizing Center v. Vimentin Intermediate Filaments and Kinesin 3. Time Lapse Observations on R Grooves  148 149 149 150 151 152 153 153  C. Summary  154  n. Behaviours of Fibroblasts and their Cytoskeletons in Response to Conflicting Topographic Cues A . Changes in Cell Morphology as Migrating Cells Approached the Groove Intersections  147 147 148  180 180  B. Cytoskeletal Behaviours of Cells Spreading at Intersecting Grooves  182  C. Summary  184  HI. The Effect of Colcemid on Topographic Guidance by R Grooves A . Smooth Surfaces 1. Cell Morphology and Quantitation 2. Cytoskeleton and Associated Elements a. Actin and Vinculin b. Microtubules c. Vimentin d. Kinesin 3. Time Lapse Observations on Smooth Surfaces  198 198 198 199 199 199 199 199 199  X  B. R Grooves 1. Cell Morphometry and Quantitation 2. Cytoskeleton and Associated Elements a. Actin and Vinculin b. Microtubules c. Vimentin and Kinesin 3. Time Lapse Observations on R Grooves  200 200 200 200 201 201 201  C. Summary  202  IV. The Effects of Cytochalasin B and Cytochalasin B/Colcemid on Topographic Guidance by R Grooves A . Time Lapse Observations and Morphometry 1. Smooth Surfaces a. C B b. C B / C 2. R Grooves a. C B b. C B / C  218 219 220 220 220 221 221 221  B. Cytoskeleton and Associated Elements 1. CB-treated a. Microtubules, Vimentin and Kinesin i. barrier behaviour ii. ridge-edge behaviour iii. wall-hugging behaviour b. Actin 2. CB/C-treated a. Microtubules, Vimentin and Kinesin b. Actin  222 223 223 225 225 225 226 226 226 226  C. Summary  227  V . Sensitivity of Fibroblasts and their Cytoskeletons to Topographic Guidance and Topographic Compensation by R, N and V N Grooves A . Detailed Descriptions of Cell Behaviours and Cytoskeletal Elements by Treatment Condition 1. MT-containing Cells - Control and CB Conditions a. Control Conditions b. C B Conditions 2. MT-deficient Cells - Colcemid and CB/C Conditions a. Colcemid Conditions b. CB/C Conditions  250 251 251 251 252 253 253 255  B. Summary of Cell Behaviours by Substratum 1. Smooth Surfaces 2. Grooved Surfaces a. R Grooves b. N Grooves c. V N Grooves  256 256 256 257 257 258  C. Summary  258  xi  VI. The Response of Single, Pairs and Clusters of Epithelial Cells to Substratum Topography A . Cytoskeleton 1. Cytoskeletal Patterns in Cells Spread on Smooth Surfaces a. Actin b. Microtubules 2. Cytoskeletal Patterns in Cells Spread on R Grooves a. Actin b. Microtubules  283 283 283 284 284 284 285 286  B. Morphometry 1. Area 2. Shape - Form Ell 3. Orientation  286 287 287 288  C. Time Lapse Observations  288  D. Summary  288  C H A P T E R 5 - DISCUSSION  314  I. The Sequence of Alignment of Microtubules, Focal Contacts and Actin Microfilament Bundles in Fibroblasts Spreading on Smooth and R-grooved Substrata A. Cell Morphology  316 316  B. Cytoskeleton 1. Focal Contacts 2. Actin Microfilament Bundles 3. Intermediate Filaments 4. Microtubules  317 317 318 318 319  C. Possible Mechanisms for Microtubule Alignment  319  n. Behaviours of Fibroblasts and their Cytoskeletons in Response to Conflicting Topographic Cues  321  III. The Effect of Colcemid on Topographic Guidance by R Grooves  324  IV. The Effects of Cytochalasin B and Cytochalasin B/Colcemid on Topographic Guidance by R Grooves  329  V . Sensitivity of Fibroblasts and their Cytoskeletons to Topographic Guidance and Topographic Compensation by R, N and V N Grooves  332  VI. The Response of Single, Pairs and Clusters of Epithelial Cells to Substratum Topography  335  C H A P T E R 6 - CONCLUSIONS and FUTURE DIRECTIONS  338  I. Conclusions  338  U. Future Directions  342  BIBLIOGRAPHY APPENDICES Appendix 1. Tables of Cytoskeletal Patterns Appendix 2. Nocodazole-resistant microtubules  LIST of TABLES  Table 1  Page 282  Table A. 1  Page 370  Table A.2  Page 371  Table A.3  Page 372  Table A.4  Page 373  xiv LIST of FIGURES Figure 1  Page 115  Figure 2  Page 118  Figure 3  Page 156  Figure 4  Page 157  Figure 5  Page 158  Figure 6  Page 159  Figure 7  Page 160  Figure 8  Page 161  Figure 9  Page 162  Figure 10  Page 164  Figure 11  Page 166  Figure 12  Page 167  Figure 13  Page 168  Figure 14  Page 170  Figure 15  Page 172  Figure 16  Page 173  Figure 17  Page 174  Figure 18  Page 175  Figure 19  Page 176  Figure 20  Page 177  Figure 21  Page 178  Figure 22  Page 179  Figure 23  Page 185  Figure 24  Page 186  Figure 25  Page 187  Figure 26  Page 188  XV  Figure 27  Page 189  Figure 28  Page 190  Figure 29  Page 191  Figure 30  Page 192  Figure 31  Page 193  Figure 32  Page 194  Figure 33  Page 195  Figure 34  Page 196  Figure 35  Page 203  Figure 36  Page 204  Figure 37  Page 205  Figure 38  Page 206  Figure 39  Page 207  Figure 40  Page 208  Figure 41  Page 209  Figure 42  Page 210  Figure 43  Page 211  Figure 44  Page 212  Figure 45  Page 213  Figure 46  Page 214  Figure 47  Page 216  Figure 48  Page 229  Figure 49  Page 231  Figure 50  Page 233  Figure 51  Page 234  Figure 52  Page 235  Figure 53  Page 237  xvi Figure 54  Page 239  Figure 55  Page 240  Figure 56  Page 242  Figure 57  Page 244  Figure 58  Page 245  Figure 59  Page 247  Figure 56  Page 248  Figure 61  Page 260  Figure 62  Page 262  Figure 63  Page 264  Figure 64  Page 266  Figure 65  Page 268  Figure 66  Page 269  Figure 67  Page 270  Figure 68  Page 271  Figure 69  Page 273  Figure 70  Page 275  Figure 71  Page 277  Figure 72  Page 279  Figure 73  Page 281  Figure 74  Page 290  Figure 75  Page 291  Figure 76  Page 292  Figure 77  Page 294  Figure 78  Page 296  Figure 79  Page 297  Figure 80  Page 299  XVII  Figure 81  Page 300  Figure 82  Page 301-  Figure 83  Page 302  Figure 84  Page 304  Figure 85  Page 306  Figure 86  Page 308  Figure 87  Page 310  Figure 88  Page 312  Figure A. 1  Page 376  Figure A. 2  Page 378  Figure A. 3  Page 380  Figure A. 4  Page 381  NOMENCLATURE Abbreviations used in the Thesis ABP  actin binding proteins  AAP  actin associated proteins  ADF  actin depolymerization factor  ADP  adenosine diphosphate  AEB  actin edge bundles  ATP  adenosine triphosphate  Ca2+  calcium ion  CB  cytochalasin B  CB/C  cytochalasin B/colcemid  CD  cytochalasin D  CD/C  cytochalasin D/colcemid  CLSM  confocal laser scanning microscope  CS buffer  cytoskeletal stabilizing buffer  DIC  differential interference contrast  Dm ax  maximum diameter  DMF  dimethyl formamide  DMSO  dimethyl sulfoxide  E cells  epithelial cells  ECM  extracellular matrix  EGF  epidermal growth factor  ER  endoplasmic reticulum  F-actin  filamentous actin  FN  fibronectin  FNR  fibronectin receptor  G-actin  globular actin  GDP  guanidine diphosphate  Glu tubulin  detyrosinated microtubules rich in tubulin with glutamine residues  GTP  guanidine triphosphate  IF  intermediate filaments  IRM  interference reflection microscopy  MAPs  microtubule associated proteins  MEM  minimal essential medium  MF  microfilaments  MT  microtubules  MT+/MF+  control cells with an intact cytoskeleton  MT+/MF-  cytochalasin-treated cells deficient in actin M F bundles  MT-/MF+  colcemid-treated cells deficient in M T  MT-/MF-  cytochalsin and colcemid treated cells deficient in M T and M F bundles  MTOC  microtubule organizing center  NC  nocodazole  N grooves  Narrow grooved pattern; 6-9 Jim pitch, 3-|im deep  OI  Orientation Index  PDL  periodontal ligament  PBS  phosphate buffered saline  PMN  polymorphonuclear leukocyte  RFGD  radio-frequency glow-discharge  RGD  arginine-glycine-aspartic acids  R grooves  Regular-grooved pattern; 30-|im pitch, 3-^im deep  SI  subfragment 1 of myosin  S.E.M.  Standard error of the mean  Tyr tubulin  microtubules rich in tubulin with tyrosine residues  VN  vitronectin  V N grooves  Very Narrow-grooved pattern; l-(im pitch, 0.5 \im deep  Vnc  vinculin  VNR  vitronectin receptor  ' X ' grooves  grooves leading into the intersection  ' Y ' grooves  grooves intersecting at right angles with ' X ' grooves  PREFACE Some of the material included in this thesis has been previously published or is in press, as noted below: Papers: 1. Oakley, C. and Brunette, D. M . (1993). The sequence of alignment of microtubules, focal contacts and actin filaments in fibroblasts spreading on smooth and grooved titanium substrata. J. Cell Sci. 196:343-354. 2. Oakley, C. and Brunette, D. M . (1995). Topographic Compensation: guidance and directed locomotion of fibroblasts on grooved micromachined substrata in the absence of microtubules. Cell Motil. Cytoskeleton 31:45-58. 3. Oakley, C . and Brunette, D. M . (1995). The response of single, pairs and clusters of epithelial cells to substratum topography. Biochemistry and Cell Biology. Accepted, in press. Abstracts: 1. Oakley, C. and Brunette, D. M . (1993). Confocal microscopy of the cytoskeleton of fibroblasts on grooved titanium. J. Dent. Res. 72:388. 2. Oakley, C. and Brunette, D. M . (1994). Effects of colcemid and cytochalasin B on fibroblasts spreading on titanium. J. Dent. Res. 73:401. These publications as well as this thesis are the principal work of the candidate, Carol Oakley . However, the thesis supervisor and senior author of the above-noted publications, Dr. D. M . Brunette, offered editorial comments on the manuscripts and contributed advice and suggestions throughout the course of the experiments that comprise these publications and this thesis. The candidate and senior author, as well as thesis supervisor, agree that the contributions of the respective parties are as stated above.  Carol Oakley, carfHida'te  A  Dr. D. M . Brunette, supervisor and co-author  xxii ACKNOWLEDGEMENTS I would like to thank my supervisor, Dr. D. M . Brunette, not only for his advice and guidance throughout the course of this thesis but also for his criticisms which I admit were not always appreciated at the time. In that regard, I thank Dr. Brunette for his tolerance and understanding and for maintaining his sense of humour. I am also grateful to my PhD Committee, Dr. E. Evans, Dr. V . J. Uitto, and Dr. W. Vogl. In particular, I would like to thank Dr. Vogl for his advice and comments regarding the cytoskeleton. I would also like to acknowledge Dr. L . Keshet who has collaborated with our laboratory.  I am indebted to Mrs. L . Weston and Mr. A. Wong for their technical support and to Ms. J. McDiarmid for her diligence in tracing and digitizing the cells. Thanks are also due to the numerous work-study students who assisted with portions of these thesis. I would also like to acknowledge the expertise and generous access to the laboratories of the Center for Microelectronics, Department of Electrical Engineering, University of British Columbia, directed by Dr. N Jaeger, and thank Mr. H . Kato for fabricating the micromachined surfaces.  I am also indebted to Dr. Doug Waterfield, Dr. Babak Chehroudi and my fellow students, Dr. Lee Chou and Dr. Ed Putnins for their friendship and moral support. I would also like to express my gratitude to my husband, John, for making many sacrifices and standing by me during the darkest hours; this is a man who survived not only his own PhD, but his wife's as well.  Finally, I would like to acknowledge the financial support of the Medical Research Council of Canada (MRC Dental Fellowship Grant No. 5-59108).  1 CHAPTER 1 INTRODUCTION I.  Overview  It is important to understand cell behaviours on artificial substrata because an increasing number of individuals are receiving artificial devices to replace damaged or missing tissues or organs. Such devices include total or partial hip and other joint implants or prostheses, cardiovascular shunts and catheters and dental implants. The success of implanted devices depends upon the interaction of cells with the artificial surface and specifically upon those interactions involving cell adhesion and migration along the substrata. These interactions are influenced by the nature of the substrata, such as its physical and chemical composition. Moreover, most artificial devices contain some degree of heterogeneous or anisotropic topography produced either deliberately or inadvertently during tooling or polishing procedures.  Initially, the effects of substratum topography on cell behaviour in vivo were obtained primarily from the overall performance or success of the implanted device, an outcome that was the result of many factors including surgical technique, cleanliness and biocompatibility of the implant material. However, there is now substantial evidence (eg. Chehroudi et al., 1989, 1990, 1991) indicating that specific dimensions and geometries of the substratum topography can affect the orientation of tissue ingrowth into and around an implanted device which in turn, affects the success or longevity of the implanted artificial device (see also review by Brunette, 1988b).  The in vivo responses of cells to substrata topographies are beyond the scope of this thesis, and it is difficult in any case, to study mechanisms of cell responses in such a complex milieu. This thesis focuses on delineating cytoskeletal behaviours that may be involved in cell responses to substratum topography in vitro by manipulating the substratum topography and the cellular cytoskeleton. This chapter will review some of the characteristic cell behaviours (Sections n, III) on artificial substrata along with current hypotheses (Section IV) that may explain these  phenomenon. This chapter will also review the chemical and physical features of artificial substrata (Section JJ) as well as the cytoskeletal components (Sections V) and cell behaviours (Section VI) that may mediate cell responses to the substratum.  3 II.  In vitro Cell Adhesion to Artificial Substrata  Features of the substratum that affect cell adhesion are discussed below. Other aspects of cell adhesion that involve the cytoskeleton are presented in Section V . D . (Cytoskeleton, Cell Attachment) of this chapter.  A.  Physical Properties of the Substratum  Maroudas (1973a, b) suggested that cell attachment to a surface depends upon the rigidity of the surface, cleanliness of the surface and the size of the available adhesion sites; these factors will be discussed below (see also reviews by Brunette, 1986c and 1988b).  1.  Mechanical rigidity  Cells may require some degree of mechanical rigidity of the substratum to withstand the tensile forces they exert (Maroudas 1973a). However, cells can also exert tensile forces on deformable substrata as they attach and migrate (eg. Harris, 1982). The amount of traction that cells exert is evident from the degree of wrinkling produced by cells spread on thin films of silicone rubber. Some cells such as transformed cells, neutrophils and growth cones of extending neurites exert little traction. In contrast, activated platelets, epithelial (E) cells and fibroblasts can cause significant deformation of silicone films (reviewed by Curtis and Clark, 1990) although the cumulative effects of many E cells are required to produce tension comparable to that produced by a single fibroblast (Harris, 1982; reviewed by Brunette, 1988b).  Because the traction forces exerted by different cell types differs, the nature and dependency of their attachments upon rigidity of the substratum may also differ. Curtis and Clark (1990) surmised that if an adhesive surface was isotropically (ie. lacking predetermined axes) deformable, then cells whose motility required significant traction forces would be unable to locomote or possibly even unable to spread.  4 2.  Cleanliness of the Surface  The hydrophilicity and hydrophobicity of the surface can be described by the wettability of the surface which is measured by the critical contact angle formed by liquid droplets spreading on the surface (eg. Baier and Meyer, 1988). Wettability is closely related to the free surface energy of a material. In general, materials such as glass and metals have a high surface energy and are highly hydrophilic and wettable. In contrast, materials such as silicone or teflon have low surface energy and are hydrophobic.  The wettability or surface energy of a material can be temporarily increased by a process termed radio frequency glow discharge (see also Chapter 3, Section I.D.3.b) which ashes away organic contaminants and renders surfaces sterile, highly hydrophilic and very receptive for adhesion (Doundoulakis, 1987; Gombotz and Hoffman, 1987).  However, cells do not attach to artificial surfaces direcdy. Instead, the hydrophilicity of the surface affects the adsorption of macromolecules onto the surface. Within a minute of exposure of a surface to medium, a 2-5 nm layer of macromolecules can adsorb onto a surface in a process termed interface conversion (Baier, 1986). On surfaces with a high surface energy, macromolecules may bend out of shape and develop more points of attachment for cells (reviewed by Baier, 1970; Kasemo and Lausmaa, 1988). In fact, the conformation of adsorbed cell adhesion molecules rather than their absolute quantity determines the extent of cell-to-substrate adhesion (Lewandowska et al., 1989). Thus, the surface chemistry of the substratum determines not only the quality and quantity of adsorbed molecules but their conformation as well. In turn, cells interact with a caipet of adsorbed macromolecules.  In vitro , the process of adhesion may resemble the in vivo process if the culture medium contains serum. Serum contains a number of extracellular adhesion molecules such as fibronectin and vitronectin. Fibronectin and vitronectin are glycoproteins that possess a particular sequence of  5 amino acids, arginine-glycine-aspartic acids (RGD) that are recognized by specific receptors (integrins) on the cell membrane (eg. Ruoslahti and Pierschbacher, 1987; see also Chapter 1, Section V.D.). Vitronectin is also known as serum spreading factor and accounts for the majority of adhesion-promoting activity in serum (Hayman et al., 1985) so that cells plated in the presence of serum are essentially plated onto a vitronectin-substrate (Burridge et al., 1988).  After the adsorption of medium and attachment molecules to the substratum surface, the second step in cell adhesion is the actual contact between the cell and the surface. One difficulty involves electrostatic forces because both the cell surface and most tissue culture surfaces are negatively charged and would be expected to repel (Weiss, 1975). However, the surfaces of cells are heterogeneous in their distribution of negative charges and areas with lower- than-average charge may approach the substratum closely enough so that attractive interactions predominate (Weiss, 1975). Microextensions of the cell surface may contain receptors that are specific for attachment factors that have adsorbed to the surface. Subsequently, changes in the plasma membrane and reorganization of the cytoskeleton consolidate the attachment as additional attachments are made and spreading continues. Cells attach to the substratum via several types of contacts that may be described by their microscopic appearance (see Chapter 1, Section V.D.I.) and may reflect different functions (see Chapter 1, Section V.D.3). For example, close contacts (see Chapter 1, Section V.D.I.) are associated with cell spreading (Izzard and Lochner, 1976) because they can provide the adhesion required to transmit to the substratum, the forces involved in the forwards movement of the marginal cytoplasm (Izzard and Lochner, 1980).  3.  Size of Available Adhesion Sites  Maroudas (1973b) concluded that different cell types have different requirements for the amount or size of substratum necessary to promote cell growth. That is, most cells require anchorage to a substratum as a stimulus for cell division and a surface could support cell growth simply by providing that minimum of mechanical support (see Maroudas, 1973a and Section n.A.l. above)  6 which allowed the cell to extend to its normal length.  Anchorage stimulus of attachment to a substratum can also be demonstrated on adhesive islands which have been imprinted on a non-adhesive substratum. The pattern of adhesive islands provides precise control of cell shape and precise control of the distribution of cell attachment sites or focal contacts, as identified by vinculin-staining plaques (see Chapter 1, Section V.D.); (O'Neill et al., 1990; see also Ireland et al., 1987). Significantly, anchorage stimulation is a function of total focal contact area and neither the arrangement of focal contacts nor their number is related to cell growth (O'Neill et al, 1990).  B.  Directed Orientation and Migration of Cells by the Substratum  Patterns of adhesiveness on a substratum can not only affect cell shape (Ireland et al., 1987; O'Neill et a l , 1990) but can also influence the orientation and direction of cell migration. In fact, several heterogeneous or anisotropic features of the environment, including the substratum topography can act as directional cues to influence cell orientation and the direction of migration. Directional cues can be chemical (adhesive) or physical (topographic or mechanical), (reviewed by Curtis and Clark, 1990).  1.  Chemical or Adhesive Guidance  a. Chemotaxis Chemotaxis is defined as the directed migration of cells along or against a concentration gradient of a soluble chemoattractant. Chemotaxis has been studied primarily in leukocytes such as neutrophils and monocytes (eg. Wilkinson, 1987) although fibroblasts can also respond to chemotactic signals (eg. Adelmann-Grill and Cully, 1990). The molecular mechanisms of chemotaxis are poorly understood, but it is likely that specific membrane receptors that recognize a gradient of specific molecules may be involved. In turn, the signal is transmitted into the cell and cytoskeletal reorganization, development of biased protrusions and directed motility ensue (eg. Fay  7  et al., 1989; reviewed by Zigmond, 1989; Stossel, 1994; see also Chapter 1, Section VI.).  b. Galvanotaxis Galvanotaxis refers to the response of cells to a voltage gradient and is dependent upon steady, long-lasting voltage gradients rather than gradients that are transient (Harris, personal communication). Different cell types respond differendy to the cathode (negative electrode) and anode (positive electrode). For example, macrophages and osteoclasts orient and locomote towards the anode whereas epithelial cells and neurite growth cones prefer the cathode (reviewed by Harris, personal communication). Some fibroblasts also orient and locomote towards the cathode (eg. Popov et al., 1991; Brown and Loew, 1994) but the predominant fibroblast response is to align perpendicular to the axis of the voltage gradient (Harris, personal communication). It is likely that in different cell types, different cellular mechanisms dominate or a different combination of mechanisms are operational.  When a voltage gradient is applied across a cell, the voltage inside a cell is initially lower than outside. Subsequently, the cell membrane facing the anode becomes hyperpolarized and the membrane facing the cathode becomes depolarized (reviewed by Harris, personal communication). In fibroblasts, the development of protrusions appears to be independent of actin polymerization (Popov et al., 1991) and Harris (personal communication) noted that voltage gradients seemed to favour the formation of new adhesions on the cathode-directed side of the cell and the breakage of adhesions on the anode side. In fact, the receptor or integrin (see Chapter 1, Section V.D.) for fibronectin is recruited from large clusters and redistributed, in a calcium-independent process, in more diffuse localizations towards the cathode (Brown and Loew, 1994).  c. Haptotaxis Carter (1965) observed that cells in culture do not readily adhere to a surface of cellulose acetate but, if this surface is metallized in vacuo, cells will adhere to it and spread. The amount of metal  deposited can be varied to produce a range of different degrees of cell adhesion and fibroblasts presented with a gradient of palladium will locomote against (up) the gradient (Carter, 1965). In fact, cell movement against the gradient is not only directional but the cell tracks are longer and indicate more efficient motility than on evenly-metallized surfaces (Carter, 1965). Carter (1965) suggested "that this phenomenon should be called 'haptotaxis' (Greek: haptein, to fasten; taxis, arrangement) to convey the idea that the movement of a cell is controlled by the relative strengths of its peripheral adhesions". Carter (1967) related cell adhesion to spreading tension and suggested that cell processes directed towards areas of adhesivity exerted tension on the remainder of the cell and thereby inhibited migration away from the adhesive areas.  Curtis and Clark (1990) defined haptotaxis as directed cell movements to adhesive guidance cues that are oriented or gradients of substratum adhesiveness.  d. Bidirectional Adhesive Cues The effects of chemotaxis, galvanotaxis and haptotaxis are typically described as unidirectional because cells respond to either a chemical, voltage or adhesive gradient by orienting with and locomoting against (up) the gradient. In contrast, environmental cues such as adhesive tracks, substratum topography (see Section II.B.2. below) and intercellular interactions (see Section II.B.l.e. below and also Chapter 1, Section VI.C) can evoke bidirectional responses.  The relative effects of physical (topographic and mechanical) and chemical (adhesive) influences have been difficult to establish because some techniques employed in fabricating either intended topographic or selectively-adhesive substrata have inadvertently introduced unwanted chemical or topographic heterogeneities (eg. review by Singhvi et al., 1994; see also Section EII.A. below).  Alternating parallel tracks of adhesive and nonadhesive substrata have been produced by a variety of techniques and in a range of periods or repeat spacing (eg. Ivanova et al., 1976; Britland et al.,  1992; Clark et al., 1992). Typically, cell attachment, spreading and locomotion are limited to the adhesive tracks, despite increasing population pressure over time, as a result of mitotic activity. However, the adhesivity of an initially non-adhesive substratum may increase over time (eg. Britland et al., 1992) because serum proteins are continuously adsorbed onto the substratum (see Section II.A.2. above). The absolute quantity of adsorbed proteins is less critical to cell adhesion than the conformation of the adsorbed proteins and since the surface chemistry of the substratum governs the conformation of adsorbed proteins (Lewandowska et al., 1989), some surfaces may not increase in adhesivity.  Different cell types respond differently to the periodicity or density of adhesive tracks. For example, the morphology and guidance of neurite extension varies with neuron type, the spacing of the tracks (Clark et al., 1993) as well as the angle of approach to the substrate boundaries (Tanaka and Kirschner, 1995). Colonies of E cells are largely unaffected by the spacing of adhesive tracks and only marginal cells demonstrate some alignment to the adhesive pattern (Clark et al., 1992). In general, alignment responses of single E cells and fibroblasts increase as the periodicity of the adhesive tracks increases. However, on surfaces with narrow nonadhesive regions, cell protrusions can bridge over these regions to span many periods of pattern and differences in the ability of cells to bridge over patterns may reflect the length of their protrusions (Clark et al., 1992). In order for adhesive tracks to exert a guidance response in the absence of other directional cues such as haptotaxis, chemotaxis or population pressure (see below), tracks must be narrow enough to prevent deviation from the desired orientation and isolated or narrow enough to prevent bridging (Clark et a l , 1992).  e. Population Pressure Population pressure or contact inhibition of locomotion (see also Chapter 1, Section VI.C) causes cells to move away from a densely-populated area towards a sparsely-populated area. This behaviour is evident in the centrifugal growth of cells from the edges of a tissue explant in vitro  10 (Middleton and Sharp, 1984) and in wound healing where cells along the wound margin migrate into the denuded area to repopulate the site.  Similar behaviour may also contribute to the formation of cell tracts in which fibroblasts become aligned locally at confluence (eg. reviewed by Trinkaus, 1984; Curtis and Clark, 1990; Harris, personal communication). Likewise, if adhesive tracks or substratum topography limit the availability of space for cell attachment and spreading, it is easy to imagine how population pressure could reinforce or amplify alignment and directed migration of fibroblasts along the long axis of the substratum cue. Thus, at high population densities the effect of topography upon cells is different than at low density. Topographies which have only a small degree of alignment at low cell densities, will have a significandy greater effect at high cell density (Curtis and Clark, 1990). In this thesis, the experiments with fibroblasts were limited to observations of single, isolated cells in order to avoid the influence of contact inhibition of locomotion and population pressure.  E cells respond differently than fibroblasts to population pressure. Contact between E cells results in contact-induced spreading (Middleton, 1976). Cells increase their spread area, form stable lateral contacts between adjacent cells and assemble into sheets (DiPasquale, 1975; Brown and Middleton, 1981; see also Chapter 1, Section VI.C). Although single E cells respond to adhesive (see above) and topographic (see III.B. below) cues, colonies of E cells respond poorly or not at all and it appears the formation of stable cell contacts can override the aligning effect of the substratum (reviewed by Curtis and Clark, 1990).  2.  Physical or Topographic Guidance  Harrison (1914) was first to describe the orientation response of fibroblasts and other tissue cells in tissue culture along thin glass fibers, scored surfaces or grooves and even spiders' webs. Weiss (1934, 1945) cultured fragments of embryonic chick brain or spinal ganglia in plasma and coined the term "contact guidance" to describe the directed outgrowths of nerve cells along fibrin fibrils or  glass fibers embedded in the plasma clots. Weiss (1934,1945, 1958,1959) also observed the aligned outgrowths of a variety of cells on a variety of surfaces including clotted plasma, reprecipitated collagen, fish scales, scored glass and glass fibers and noted the tendency of cells to be guided by the shape of the substratum to which they attached so that their long axes were parallel to the long axes of the grooves or fibers.  As initially described by Weiss (1934, 1945), contact guidance connotes cell outgrowth or movement (Curtis and Clark, 1990) and therefore Curtis and Clark (1990) introduced the term "topographic guidance" to include the shape, orientation, movement and polarity of movement of cells in reaction to their substratum. Contact guidance is the more familiar and more-commonly used term although contact guidance and topographic guidance are often used interchangeably (Singhvi et al., 1994) and this convention is observed in this thesis.  3.  Interactions between Directional Cues  Different types of directional cues may be combined to alter the probability of a guidance response (see also Section IV.C below) to a cue and the effect may be cooperative or conflicting. For example, the effects of population pressure can increase orientation to other directional cues (see above). As well, the probability of cells crossing a vertical step can be altered by rendering the upper surface of steps less adhesive. Compared with identical steps without adhesive gradients, cells cross more frequently from lower to higher adhesiveness and less frequently in the opposite direction (Clark et al, 1987).  In neutrophils, a topographic cue can enhance or reinforce a chemotactic cue. However, a chemotactic cue will override a conflicting topographical cue although the overall chemotactic response will be impaired (Wilkinson and Lackie, 1983; Wilkinson, 1987). In neuronal growth cones, galvanotactic growth towards the cathode can be also be impaired by conflicting topographic cues (reviewed by Curtis and Clark, 1990).  12 III.  In vitro Topographic Guidance  The following discussion will be limited to the in vitro observations of topographic guidance on different substrata and by different cell types, with emphasis placed on fibroblasts and epithelial cells. However, contact guidance also occurs in vivo and one example is evident during embryonic development when neural crest cells migrate along oriented extracellular matrix. Similarly, tumour cells may invade along oriented collagen in tissue planes and neuronal migration and axon outgrowth may be directed by glial cells in the brain (reviewed by Curtis and Clark, 1990).  In vitro, contact or topographic guidance has been observed in neurites, neutrophils, epithelium and fibroblasts (see Section III. B. below) and as noted above, occurs on a variety of substrata (see Section M . A . below). It is difficult to compare observations between different studies because they have utilized not only a variety of different cell types but a variety of cell lines of the same cell type. As well, cells have been plated on substrata of various natural and artificial compositions that have been prepared or fabricated by different techniques and that have different geometries and dimensions. For example, the ability of collagen gels to guide cells depends upon the method of gel preparation because some methods (eg. air-drying) may not maintain the threedimensional structure necessary to guide neutrophils or the elasticity necessary to guide fibroblasts (reviewed by Curtis and Clark, 1990; see also Section ffl.B. below).  A.  The Role of Artificial Substratum Composition, Dimensions and Geometries  The simplest method of producing an anisotropic surface geometry is to rub the surface with a paste containing different sizes of abrasive particles (eg. Ohara and Buck, 1979). Other methods include cutting or tooling the surfaces. Recently, photolithography, plasma etching (see Chapter 3, Section D.l.) and laser holography (eg. Clark et al., 1991) have been introduced. Although these latter techniques allow fabrication of precisely defined topographies, none of these techniques can eliminate differences in surface chemistry between abraded, tooled or etched and unetched regions  13 (reviewed by Singhvi et al., 1994). As surface chemistry influences the adsorption of proteins as well as the conformation of the adsorbed proteins (see Chapter 1, Section II.A.2 above), it is essential that surfaces are chemically homogeneous or effects of surface morphology upon cell behaviour cannot be considered separately from possible chemical differences (Singhvi et al., 1994). Nevertheless, when the same cell lines are used to compare the effects of spacing of adhesive cues (Clark et a l , 1992) with those of identically-spaced topographic cues (Clark et al., 1990), it is apparent that the two cues have opposite effects. That is, the increased density of cues decreases the alignment response of cells to adhesive cues but increases alignment response of cells to topographic cues. Thus, Clark et al., (1992) conclude that topographic guidance is independent of any adhesive pattern a topographical pattern may contain and in fact, the two cues could be antagonistic.  The protocol developed by Brunette (eg. 1986a, 1986b; see Chapter III, Section I.D.f) and employed for the fabrication of substrata used in the experiments included in this thesis, avoids the problem and possibility of chemical heterogeneity altogether (Singhvi et al., 1994). After fabrication, all substrata are evaporatively coated with a thin film of titanium which results in a completely homogeneous titanium surface (Chehroudi, Thesis 1991). Titanium was selected (Brunette et al., 1983) because of the reported high rates of clinical successes of dental implants made from titanium (Branemark et al., 1977) and because E cells grown on titanium exhibit the same attachment structures (Gould et al., 1981) as does epithelium attached to teeth.  One concern with textured surfaces, particularly those produced by polishing with abrasive particles, is the production of roughness. Rugophobia (Rich and Harris, 1981) describes the avoidance behaviours of fibroblasts to rough or pitted and bumpy surfaces and small, sub-micron dimensions appear more effective in deterring fibroblast attachment than larger hills and valleys or linear anisotropies (reviewed by Brunette, 1988b; Harris, personal communication). In fact, the failure of fibroblasts to enter grooves (eg. Ohara and Buck, 1979) may be related to roughness  14 within the grooves produced during the fabrication of the grooves.  1.  Non-grooved Substrata  Fibroblasts seeded onto cylindrical substrata such as fibers (eg. Dunn and Heath, 1976; Rovensky and Samoilov, 1994) appear resistant to bending around the fibers and instead, become elongated and oriented along the cylinder. Thus, cells appear to respond to the curvature of the substratum (Curtis and Varde, 1964) and alignment depends upon the radius of curvature of the substratum. Cell alignment occurs over an optimal range of fibers diameters, typically between 100 and 150 |im (reviewed by Curtis and Clark, 1990).  On wavy fibers or cylinders, fibroblasts favour the concave side over the convex side (Curtis and Clark, 1990). Spirals provide a graded set of curvatures in two dimensions. Cells near the center of the spiral are unable to spread completely, and this situation may be analogous to cylinders or spheres with very small diameters (Dow et al., 1987). Thus, spirals may act as cell traps although, away from the center of the spiral, cells spread preferentially in regions of lower curvature, away from the center (Dow et al., 1987).  The reactions of cells to hills and dimples, pits or valleys depends upon the size of the features relative to the size of the cells (Green et al., 1994). For example, small dimensions may be perceived as surface roughness and evoke a rugophobic response (see above). If the lips of circular pits are appreciably smaller than the size of the cell, they may immobilize the cells and act as cell traps (Curtis and Clark, 1990). Hill and valley dimensions that resemble the surface of an epithelial sheet of cells in curvature and scale, evoke strong alignment responses in fibroblasts which orient along the diagonals (Dow et a l , 1987). Large hills may be perceived as small cylinders and thus cell response varies with the radius of curvature (see above). Multiple tall hills or cylinders may be perceived as spikes and fibroblasts will successfully spread and flatten on the tops and sides of several spikes (Rovensky et a l , 1991). That is, cells can spread between  15 attachment sites on spikes that are separated from one another by wide spaces, demonstrating that cells do not require a continuous substratum in order to spread and acquire a flattened shape (Rovensky et al., 1991).  2.  G r o o v e d Substrata  Grooved substrata can be obtained by a variety of techniques and can be defined by the groove width, depth and geometry. On substrata with multiple parallel sets of identical grooves, the repeat spacing or pitch of the grooves is also important. For example, in cross-section, grooves may have parallel, vertical 'U'-shaped walls, 'V'-shaped walls (eg. Brunette, 1986a) or truncated ' V shapes (eg. Brunette et al., 1983; see also Fig. 1 in Chapter 3). The space or ledge between grooves is typically referred to as the 'ridge' and the spacing, period or repeat distance comprising one groove and one ridge is referred to as the pitch (eg. Brunette, 1983). The variety of substratum geometries, dimensions and fabrication techniques employed in different studies has recently been reviewed and tabulated by Singhvi et al., (1994) and therefore does not require repetition here. Instead, a summary of observations and generalized cell responses to grooved substrata is given below (see also reviews by Brunette, 1988a, b; Clark and Curtis, 1990; Singhvi et al., 1994).  1. Susceptibility to topography is cell-type specific and dependent on whether or not cell-cell interactions are allowed (see Section U.B.l.e. above and Section I H B . below).  2. Cells can react to single steps (Clark et al., 1987) or cliffs which consist of single convex and concave edges. There does not appear to be a systematic difference between cells encountering the step from the upper or lower side as similar proportions of cells cross or align at the step irrespective of the order in which the external and internal angles are encountered (Dow et al., 1987; Clark et al., 1987). Cells can also react to multiple steps; however, cells are strongly aligned by repeated grooves or features at a depth far less than that required to produce alignment at a  16 single step (Dow et al., 1987).  Because multiple features of the substratum such as edges or walls appear to have more effect on alignment than similarly-sized single steps (Clark et al., 1987, 1990), it is possible that cells are reacting to transition zones between ridges and grooves rather than to the bottom of the grooves (Curtis and Clark, 1990).  3. Substrata containing grooves do not present absolute barriers to cell orientation or locomotion across the grooves (eg. Brunette et al., 1983). Rather, the response of cells to features of the substratum can be predicted on a probabilistic basis (eg. Brunette, 1986a, b, 1988b; see Section IV.C below).  4. The effects of groove size and spacing upon cell alignment must also be considered in relation to the typical size of the cells being investigated (Green et al., 1994; Singhvi et al., 1994). That is, if the feature dimensions are larger than the size of a single cell, cells may fail to encounter the step or groove altogether or react to the grooves as a single step. There is also a hierarchy of the effectiveness of grooves of different dimensions. That is, if cells are exposed to the conflicting influences of grooves of different size, cells will preferentially align themselves with the large grooves and only secondarily with fine grooves (Brunette, 1986a; reviewed by Brunette, 1988a).  5. It is generally agreed that the degree of cell alignment depends upon the size of the feature and the number or density of features (see above; see also Green et al., 1994). As a general rule, cell alignment is enhanced by decreased groove spacing and by increased groove depth (reviewed by Curtis and Clark, 1990; Singhvi et al., 1994).  Little is known about which feature of the substratum is necessary to elicit contact guidance. For example, cell alignment can be primarily dependent on groove depth (Clark et al., 1990), repeat  17 groove spacing (Wood, 1988), ridge width (Dunn and Brown, 1986) or groove size (Brunette, 1986a). It should also be emphasized that the cell types and groove dimensions and shapes employed by these investigators differed and therefore direct comparisons between the results may be inappropriate (see review by Singhvi et al., 1994). A better approach may be to consider the results as relevant only to a particular cell type on substrata of specific composition, fabrication, geometry and dimensions. Nevertheless, the disparate results amongst different investigations may be explained by several possibilities: (1) Different cell types (Section EILB. below) may respond to different topographic cues (2) Different cells types may employ different cellular mechanisms (Section IV.B) which interact with the substratum (3) Different substratum dimensions and geometries (Section En.A. above) may elicit the response of different cellular mechanisms (see Section IV.B. below)  B.  The Response of Different Cell Types to Topographic Guidance  Different cell types differ in their response to substratum topography. For example, neuronal growth cones and different types of fibroblasts are impeded, to a similar degree by single steps, but neutrophil migration is not affected (Clark et al, 1987). The alignment response of neutrophils varies significantly with the three-dimensional array of the substratum in that neutrophils align poorly on grooved substrata (Clark et al., 1987; Curtis and Clark, 1990) and on the surface of dried-down two-dimensional aligned collagen gels. In contrast, neutrophils align to the axis of alignment of three-dimensional collagen gels (Wilkinson and Lackie, 1983; Wilkinson, 1987).  Single E cells, single fibroblasts (eg. Brunette 1986a, b; Clark et al., 1990, 1991) and neurites demonstrate alignment (Clark et al., 1990,1991) on some grooved surfaces. However, their sensitivities to groove size, pitch and depth vary. For example, neurites do not align on ultrafine topography (130 ran grooves and ridges, 100-400 nm depths; Clark et al, 1991) or 1-^im-deep and wider grooves (8 |im grooves, 20 }im ridges) although alignment is marked when groove depth is  18 increased to 2 |j,m (Clark et al., 1990). On larger grating substrata, the alignment of fibroblasts is affected by both pitch and depth with depth having the greater effect (Clark et al., 1991). The alignment of single E cells is not affected by groove pitch but is greatiy enhanced by increasing groove depth and to a greater degree than alignment of single fibroblasts (Clark et al., 1990). On ultrafine topographies, single fibroblasts and E cells are both well-aligned although the elongation of E cells increases significandy with groove depth (Clark et al., 1991). In contrast, colonies of E cells and their constituent cells, particularly at the colony centers, are not well aligned on either large (Clark et al., 1990) or ultrafine (Clark et al., 1991) grooves and it was concluded that stable intercellular contacts amongst E cells within a colony could override the aligning influence of the substratum (Clark et al., 1990, 1991; see also Section II.B.l.e. above).  Embryonic cells (Rovensky and Samoilov, 1994) and transformed cells (eg. Rovensky and Samoilov, 1994) are generally less susceptible to the substratum topography than their nonembryonic and nontransformed counterparts within the same cell line. With the exception of colonies of E cells, cells that are less susceptible to topographic cues share several cellular properties which have been summarized by Curtis and Clark, (1990):  1. The cytoskeletons and in particular actin microfilaments of less susceptible cells appear to be less organized and more labile than cytoskeletons (see Chapter 1, Section V.) in susceptible cells.  2. Less susceptible cells appear to be less adhesive for most substrata than nontransformed fibroblasts and epithelial cells.  3. Less susceptible cells appear to exert less traction on deformable substrata than susceptible cells.  4. The above three factors can be combined to explain the locomotory behaviour of less  19 susceptible cells. Less susceptible cells tend to display more frequent changes in the direction of migration; that is, they display less directional persistence of movement.  These factors are all related to organization of the cytoskeleton and their relevance will become clearer as cytoskeletal mechanisms proposed for topographic guidance are discussed in the following section and after the cytoskeleton is discussed in Section V .  20 IV.  Proposed Mechanisms for Topographic Guidance  A number of theories have been proposed to explain contact guidance and will be reviewed below (see also reviews by Dunn, 1982; Trinkaus, 1984; Brunette, 1988a, b; Harris, personal communication).  A.  Extra-cellular Mechanisms - Microexudate Hypothesis  Weiss (1934) observed the guidance behaviours of nerve cells and fibrocytes on fibrin fibrils in clotted plasma and proposed that the cells respond to the molecular orientation of their environment. Specifically, oriented fibrillar or ultrafibrillar aggregations of polar colloid particles or micellae are produced in the plasma medium as the result of oriented mechanical forces such as exhibited by stretching or pressing the plasma medium or stroking the medium with a hairbrush or needle (Weiss, 1934). In turn, the oriented fibrin fibrils act as mechanical structures in the medium and guide the outgrowth of the cells (Weiss, 1934). Nerve cells are guided by oriented fibrin fibrils to a much lesser extent than fibrocytes; however, when fibrocytes and sheath cells are mixed with nerve cells in culture, the nerve processes orient parallel to nearby sheath cells and fibrocytes that are aligned to the fibrin fibrils (Weiss, 1934). Similarly, when two explants of chick embryonic tissue are placed a short distance apart onto clotted plasma, the outgrowing cells from each explant are oriented towards each other (Weiss, 1959). Weiss suggested that this "twocenter" effect is the result of cells exerting tension on the fibrin fibers which in turn, produces oriented lines between the explants. As fibroblasts grow out from the explants, they become oriented parallel to the stress lines and migrate along them towards the other explant (Weiss, 1959). Weiss was correct in assuming that cell activity caused the alignment of fibrin but oriented fibrils may be an effect of directed migration rather than a cause (Trinkaus, 1984).  Weiss (1945) also observed oriented cell outgrowths from explants placed along scratches scored into mica and hypothesized that capillary suction forces within the channels oriented the fibrillar colloidal exudate produced by the explants. In turn, the migrating cells followed the microexudate  21 and became oriented. Weiss also suggested that capillary forces produced a passive oriented flow or wetting of colloidal exudate along a glass rod or fiber and that outgrowing cells became oriented along the axis of the fibers by following the trail of exudate (Weiss, 1945).  Significantly, Weiss (1959) also documented the alignment of freely-suspended cells along grooves, in the absence of a tissue explant, but failed to offer a satisfactory explanation (Curtis and Clark, 1990). Curtis and Varde (1964) used suspended cells, in the absence of serum, and could not detect a stainable microexudate yet the cells still aligned along glass fibers and in grooves. Dunn (1982) argued that it is difficult to demonstrate guidance by oriented ultrastructures because of the difficulty in preparing a substratum which is uncontaminated by larger oriented structures. Moreover, changes in cell behaviour on grooves or cylinders occur at angles or curvatures which are most unlikely to alter the capillary spread of a microexudate (Curtis and Varde, 1964; Dunn and Heath, 1976). Overall, the microexudate theory has won little support as an explanation for contact guidance and Harris (personal communication) summarized Weiss' explanations as "erroneous notions about cell spreading being a passive wetting of the substratum by a sort of directional capillarity rather than as an active process of locomotion."  B.  Cytoskeleton-based Hypotheses  Curtis and Varde (1964) suggested that the effect of topology on cell behaviour was direct and could be attributed to a reaction of the cell surface to the topology of the substrate. Current theories for the mechanism of contact guidance differ in specifics (see below) but all generally agree that cells align with the substratum topography in a manner which minimizes distortions in their cytoskeleton or attachment components (Dow et al., 1987).  1.  Microfilament  Hypothesis  a. Convex Curvatures Dunn and Heath (1976) proposed a cellular mechanism to explain how cells could detect the shape  22 of a substratum. Their model was based on the directed migrations of fibroblasts on cylinders and prisms and on mechanical properties of the cytoskeleton which imposed restrictions on the formation of cell protrusions. Specifically, they postulated that inflexible bundles of microfilaments or stress fibers (see Chapter 1, Section V . C A a . ) acted as a reference standard for detecting the curvature of the substratum. Because microfilaments are also involved with the cell's locomotory system, stress fibers would be well-suited to translate the detected curvature into a directed locomotory response (Dunn and Heath, 1976; Dunn, 1982). Stress fibers are typically observed to stretch between focal adhesions at the base of leading edges and the nuclear region of fibroblasts (see Chapter 1, Section V . C A a . and Section V.D.) and range in length from about 520 [im. The longest stress fibers in a typical fibroblast subtend an angle of about 4 degrees to a flat substratum and any change in inclination of the substratum of more than this angle could interfere with the bundle geometry (Dunn, 1982). On cylinders with small radii of curvature (<100|im) or at prism edges with intersecting angles greater than 16 degrees, linear stress fibers would have to bend and therefore they could not operate or even assemble. Thus, the shape of the substratum restricted the length of stress fibers in any given direction. If they were too short to exert sufficient traction to allow cell spreading or locomotion in that direction, the direction of a cell's locomotion was shifted to another direction.  Recendy, the microfilament hypothesis was revisited to explain the disparate alignment responses of different cell lines on cylindrical substrata in that the innate pattern of actin microfilament bundles within different cells lines may predetermine their orientation abilities on cylinders (Rovensky and Samoilov, 1994). For example, cell lines which are practically devoid of actin microfilament bundles and cell lines which posses circular arrays of microfilament bundles display weak reactions of elongation and orientation on a cylinder. In contrast, cell lines containing mainly straight bundles of actin microfilaments strongly align along cylinder axes (Rovensky and Samoilov, 1994).  23 b. Concave Curvatures Dunn and Heath (1976) restricted their original microfilament-based hypothesis to predictions of cell locomotion on convex curvatures and suspected that concave curvatures may not interfere with the locomotory machinery in the same manner.  However, the microfilament-based hypothesis was extended as an explanation for contact guidance of fibroblasts on grooved substrata by Dunn (1982) who suggested that substrata with 'V'-shaped grooves could be considered as a repeating sequence of sharp convex and concave changes in inclination and that cells reacted to the convex discontinuities of grooved substrata in the same manner as they reacted to isolated convex curvatures. Dunn and Brown (1986) combined the microfilament-based hypothesis with the focal-contact-based hypothesis of Ohara and Buck (1979; see Section IV.B.2. below) and suggested that focal contacts could form in grooves but the direction in which cells could exert traction on them was restricted by the walls of the grooves. Thus, a cell could not exert a force on a focal contact at a shallow angle if the focal contact was within the groove and the direction of the force was perpendicular to the axis of the groove, since the stress fibers associated with the focal contact would be bent. Consequently, expansion of cell margins and cell locomotion would be restricted in directions other than those parallel with the grooves (Dunn and Brown, 1986).  Significantly, Dunn (1982) also acknowledged that fibroblasts can locomote efficiently without either focal contacts or stress fibers (eg. Couchman and Rees, 1979) although their hypothesis was predicated on the presence of these structures. Dunn (1982) argued that regardless of whether close contacts and a less obvious actin meshwork were involved, traction was still transmitted to the substratum and it had not been demonstrated that cells lacking focal contacts and stress fibers were incapable of responding to substratum shape. Brunette (1988a) reported that cell alignment on grooved surfaces preceded the formation of discrete microfilament bundles and therefore aligned microfilament bundles were an effect rather than cause of cell alignment on grooved  24 surfaces.  Cells also align on very shallow (100-400 nm deep) ultrafine topographies (Clark et al., 1991) and Dunn (1991) recently agreed that their original actin-distorting hypothesis (Dunn and Heath, 1976) may not fully explain such behaviours; although, a modified application of their focal contact/stress fiber model (Dunn and Brown, 1986) may be still be the primary orienting influence of cells on ultrafine grooves (Dunn, 1991; see Section IV.B.2. below for further discussion).  2.  Focal Contact Hypothesis  Ohara and Buck (1979) observed that fibroblasts aligned on grooves with a periodicity less than the cell length, a periodicity specifically excluded from predictions of the microfilament hypothesis of Dunn and Heath (1976). As well, the cells observed by Ohara and Buck (1979) preferentially aligned along the ridges rather than within the grooves and cells often bridged over grooves, avoiding them altogether.  Ohara and Buck reasoned that because the cells were stiff, they could not attach within a groove and therefore cells spanned over grooves and utilized the hmited areas for attachment available on the ridges. Cells could lie along the top of several ridges and still align with the axis of the groove pattern, but microfilament bundles would not be distorted as predicted by Dunn and Heath (1976). Therefore, Ohara and Buck (1979) proposed that linear focal contacts (see Chapter 1, Section V.D.) governed alignment on grooves of narrow pitch. The dimensions of focal contacts (about 0.25-0.5 (im wide and 2.0-10 |im long) could only be accommodated on the ridges if the focal contacts were aligned along the long axes of the ridges. Significantly, as the width of the ridges between the grooves decreased, cell alignment increased because this orientation represented the maximum linear contact available to the cells (Ohara and Buck, 1979).  However, as argued above, cell locomotion is not necessarily associated with focal contacts. As  25 well, cells are able to form focal contacts to the bottoms and sides of grooves (Dunn and Brown, 1986; Brunette, 1988b). Moreover, cell alignment on narrow-pitch grooves increases on deeper grooves of similar pitch (eg. Clark et al., 1990, 1991) and Brunette (1988a) argued that it seemed unlikely that the formation of focal contacts would be affected differently by different groove depths. Dunn (1991) proposed that on closely-spaced grooves, each focal contact may span several grooves and become corrugated by sinking into them. Thus, any microfilament bundle associated with part of the focal contact that lies within a groove, would be restricted in the range of direction in which it can exert force (Dunn, 1991). In comparison to shallow and wide grooves, it is easy to imagine that deep and narrow grooves would more efficiently restrict non-parallel orientations of the microfilament bundles and hence cell orientation parallel to the long axis of the grooves would be amplified.  The general increase in cell alignment on grooves of narrow pitch may also be explained by the increased density of edges on such substrata. Sharp edges or discontinuities of the substratum may enhance the formation of focal contacts and actin condensations (Curtis and Clark, 1990; Dunn, 1991), possibly through a mechanism of contact-forming precursors (Izzard and Izzard, 1987; Izzard, 1988; reviewed by Harris, personal communication; see also Chapter 1, Section V.D.3.).  3.  Microtubule Hypothesis  The role of microtubules (MT) in contact guidance has received relatively little attention despite reports that the orientation of MTs precedes a functional polarization of the cell that results in directed cell movement (see Chapter 1, Section VI.B.l.).  Ohara and Buck (1979) described MTs and microfilament bundles aligned parallel to the cell axis, and they found that, unless the grooves influenced the orientation of the cell as a whole, there was no effect on the cell's components. Brunette (1986a, b) reported that for cells aligned with  26 grooves, alignment of all the cytoskeletal filaments reflected the orientation of the cell as a whole. Dunn and Brown (1986) observed that, although cells were well oriented with the groove pattern, MTs only very occasionally appeared to be influenced by the grooves.  Vasiliev et al. (1970) reported the alignment and directed migration of fibroblasts on fish scales. When MTs are disrupted by colchicine (see Chapter 1, Sections V.A.6.b.i. and V.A.7.) contact guidance is abolished and cells behave similarly to colchicine-treated cells on smooth surfaces. That is, cells become polygonal, apolar and stationary because MTs are essential to effect these behaviours on both smooth and anisotropic substrata (Vasiliev et al., 1970).  Recendy, Svitkina et al. (1995) reported that the pattern of microfilament bundles in cells on cylinders depends upon the integrity of MTs. Control fibroblasts on cylinders are elongated and contain numerous straight microfilament bundles oriented along the axis of the cylinder. When MTs are disrupted by colcemid (see Chapter 1, Section V.A.6.b.ii.), cell shape becomes polygonal and less elongated. The alignment of microfilament bundles is disturbed in that longitudinal fiber bundles are significandy reduced but transverse fiber bundles are increased. These authors concluded that MTs in fibroblasts spread on cylinders promote cell elongation and shift the longitudinal and transverse orientation of microfilament bundles to a monodirectional one and with the dominance of longitudinal bundle alignment (Svitkina et al., 1995).  C.  Stochastic Hypothesis  Harris (personal communication) observed that the current hypotheses of contact guidance view cell extension in directions perpendicular to the fibers of grooves as an inhibitory process and questioned whether the process of cell extension may instead be favoured along the axes of fibers or grooves.  In fact, both possibilities were suggested by Brunette (1988b) who noted that cells do not respond  27 to the substratum as an "all or nothing affair". Rather, cell response to a topography can be predicted on a probabilistic or statistical basis (Brunette 1986a, b, 1988b; Clark et al., 1987). The probability of a cell making a successful protrusion and adhesions in a given direction may be reduced by specific features of a topography, and similarly, protrusions and adhesions made in other directions may be favoured (Brunette, 1988b; Clark et al., 1987). As well, the magnitude of a specific feature may reduce or enhance the probability of a protrusion made by a cell being successful. That is, a protrusion in the direction contacting a feature would result in the axial polarization and/or movement of the cell in that direction (Curtis and Clark, 1990). Cell locomotion (see Chapter 1, Section VI.) comprises cycles of several processes including protrusion, adhesion and retraction and presents an opportunity for small probabilistic effects on any single process to be magnified and hence produce orientation of the entire cell (Brunette, 1988b).  The durability of adhesions to features of the substratum is related to the stability of cell processes at or along those features and therefore is important in effecting cell orientation (Brunette, 1986a). For example, time-lapse observations of cells on grooves indicate that cell processes near to or at the edge of the grooves are less stable than those formed on a flat surface and these processes give way when tension is applied to them during cell locomotion (Brunette, 1986a).  Time-lapse observations also suggest that the guidance of cell locomotion results from interactions between the grooves and the cell's leading edge rather than from interactions between the substratum and mechanically-stiff long linear cytoskeletal elements (Brunette, 1986a, b, 1988a). Lamellae are flexible structures that can bend around groove edges and apparently enable fibroblasts and epithelial cells to emerge from or descend into grooves (Brunette, 1986a, b). Fibroblasts migrating from one set of grooves to another intersecting groove change their alignment and direction of migration and the leading lamellae becomes oriented to the second set of grooves before the cell body is reoriented (Brunette, 1988a). Collectively, these observations  28 suggest that lamellae are flexible structures that may also contain the machinery responsible for contact guidance on grooves (Brunette, 1988a). Thus, it is not surprising that the lamellar regions of fibroblasts on shallow grooves conform intimately to the underlying grooves whereas the long axes of cells often cross the grooves and are not aligned (Clark et al., 1990).  Thus, it appears that a single hypothesis cannot sufficiendy explain topographic guidance. As noted above, it is possible that different cell types respond to different topographic cues or that different cell types employ different cellular mechanisms. It is also possible that different substratum dimensions and geometries evoke the response of different cellular mechanisms. Therefore, the data and conclusions presented in this thesis are limited specifically to behaviours of porcine epithelial cells and isolated human gingival fibroblasts on titanium-coated micromachined grooves.  Although hypotheses for contact guidance are often presented as discrete cytoskeletal mechanisms, it should be noted that the cytoskeletal elements and their associated proteins and components interact in a complex manner to perform coordinated cell spreading and locomotion. This fact will become evident in the following sections (Sections V., VI.).  29 V.  The Cytoskeleton  The cytoskeleton is a complex network of three principal types of protein filaments that extends throughout the cytoplasm of eucaryotic cells. The three fdament types are microtubules, intermediate filaments and actin filaments. Each filament is formed from a different protein monomer which is assembled into a variety of structures along with a variety of associated proteins. The filament systems and their associated proteins interact to allow cells to adopt different cell shapes and to perform coordinated and directed movements of organelles within the cells as well as of the cells themselves, such as the crawling of cells across a substratum. As a result, the cytoskeleton has also been termed the "cytomusculature" of the cell (Alberts et al., 1989).  Different cell types such as neurons, muscle cells or specialized epithelial cells can display unique or characteristic organizations of their cytoskeletons. As well, the cytoskeletons within the same cell types are organized differently throughout different phases of the cell cycle, such mitosis.  Because investigations that comprise this thesis were limited to interphase fibroblasts and to a lesser extent interphase non-specialized epithelial cells, discussion of the cytoskeleton in this thesis will focus largely on the cytoskeleton of these cell types. As well, detailed discussion will be limited to those components and features of the cytoskeleton and associated elements that are most relevant to the findings of this thesis. That is, investigations were limited specifically to microtubules, vimentin intermediate filaments and actin microfilament bundles in fibroblasts and microtubules and actin microfilament bundles in epithelial cells. In addition, kinesin and vinculincontaining plaques were studied in fibroblasts. Vimentin intermediate filaments were found to play a minor role in the topographic guidance of fibroblasts and therefore receive a minor review; cytokeratins of epithelial cells were not central to this thesis and are therefore not included in this review.  30 The fundamental structure and organization of cytoskeletal and associated elements along with their putative roles within cells will be presented. Often the role of a cytoskeletal element has been established by disruption of that filament system and therefore a discussion of cytoskeletalperturbing agents and their effects upon cell behaviour is included. Unfortunately, many interesting topics have been omitted or have received little attention for the sake of brevity, but references to appropriate review articles have been included throughout.  The question of just how cytoskeletal elements interact to become organized into asymmetric, dynamic arrays required for cell spreading, polarization and locomotion is still largely unanswered (Cooper and Mitchison, 1995) but models or proposed mechanisms for these behaviours are presented briefly throughout Section V and in greater detail in Section V I of this chapter.  31 A.  Microtubules  Microtubules (MT) form a variety of structures associated with a variety of cellular functions. In many differentiated epithelial cells, MTs are organized into stable structures called axonemes which form the core of cilia and flagella. In nonneuronal interphase cells, MTs often form radial, polarized cytoplasmic arrays which are involved in intracellular transport and in morphologic events such as polarization and directed cell migration. However, during cell division, the cytoplasmic arrays disappear and MTs organize into spindles which are involved in chromosome movement (reviewed by Alberts et al., 1989; Darnell et al., 1986).  The degree of M T assembly is coupled to the soluble tubulin pool in the cytoplasm and this pool, in turn, regulates the level of tubulin synthesis (autoregulation), (Suprenant, 1993). MTs are versatile and dynamic structures and these properties are conferred upon MTs by their heterogeneous primary structure and by several levels of regulation or modification of their structure which can, in turn, influence their dynamic behaviour and further interactions (Gelfand and Bershadsky, 1991). Similarly to a domino effect, local biochemical differences in the structure of tubulin affect M T function and govern the interactions of MTs with other proteins which, in turn, further modify the interactions of MTs with other intracellular organelles and cytoskeletal elements. Furthermore, the extra-cellular environment can influence M T behaviours. Some of the factors that can affect M T functions are listed below along with appropriate references, and some of these topics are included in the following discussion: a. transcriptional and assembly differences (eg. Joshi and Clevelend, 1990; Luduena et al., 1992) b. post-translational modifications of tubulins and MTs (eg. Murphy, 1991) c. association of MTs with various microtubule-associated proteins (MAPs) including MTassociated motors (eg. Chapin and Bulinski, 1992; Hirokawa, 1994) d. interaction of MTs with other cytoskeletal elements and cellular structures (eg. Gelfand and Bershadsky, 1991) e. interaction of MTs with the physical extra-cellular environment (eg. Hill and Kirschner, 1982a)  32 and pharmacologic agents (Dustin, 1984).  1.  Structure  MTs are polymers of tubulin comprised of two polypeptides designated a-tubulin and 6-tubulin. The a and 6 tubulin monomers share 36-42% of their amino acid sequences (Luduena et al., 1992) and the longest of these shared sequences is a glycine cluster that may be a binding site for GTP (Dustin 1984). Each ocB-dimer binds 2 molecules of GTP, one per monomer, forming a "aB2GTP" complex. The 6 subunit binds one GTP and this GTP is considered an exchangeable GTP as it can be exchanged with nucleotides from solution. The a subunit binds the other molecule of GTP but this GTP is non-exchangeable (Mandelkow and Mandelkow, 1995).  The disparate amino acid sequences between a and B tubulin isotypes may be related to differences between isotypes in their affinities for different microtubule-associated proteins (MAPs; see Chapter 1, Section V.A.4.a.) and colchicine (see Chapter 1, Section V.A.6.b.i.) and may also account for some of the different functions of MTs (reviewed by Joshi and Cleveland, 1990; Murphy, 1991). In fact, one tubulin isoform, gamma (y) tubulin, may serve a special function as a nucleation site for MTs within microtubule-organizing centers (see section V . A.2.a. below).  Some residues in B tubulin can form intra-chain disulfide bridges which introduce steric interferences that can prevent the assembly of molecules containing this particular cross-link (Luduena and Roach, 1991). However, only some of tubulin's sulfhydryl groups are 'assemblycritical' and these may be specific targets for tubulin ligands (such as colchicine; see Section V.A.6.b.i. below).  The a-B tubulin dimers are about 4-5 nm in diameter and 8 nm long. The globular dimers are assembled into protofilaments which are longitudinal rows of repeating tubulin dimers, linked aB to aB. This "head to tail" assembly imparts an intrinsic polarity to the protofilaments and to the M T  33 filament itself which consists, in vivo, of 13 parallel protofilaments encircling a hollow-appearing centre. More correctly, the dimers may actually be linked 6a to 6a (Mandelkow and Mandelkow, 1995) as the polarity of the protofilaments is such that a-tubulin points to the plus or fast-growing end and 6-tubulin points to the minus or slow-growing end. The assembled protofilaments form M T filaments that are 24 nm in diameter but vary in length.  2.  Organization and Assembly  a. Microtubule-organizing Center i. Fibroblasts In fibroblasts, interphase MTs are nucleated and organized from a microtubule-organizing center (MTOC) or centrosome. That is, one end, the minus end, of each M T is sequestered within the M T O C and the other end, the plus end, extends into the cytoplasm, towards the plasma membrane.  The centrosome contains amorphous pericentriolar material which surrounds a pair of centrioles set a right angles to each other. Centrioles are cylindrical in shape and in cross section, each centriole resembles a pinwheel consisting of a cylindrical bundle of 9 rods. Each rod consists of a triplet of MTs, fused along their length, and each triplet consists of one complete 13-protofilament M T and two partial MTs (reviewed by Alberts et al., 1989; Darnell et al., 1986; Glover et al., 1993).  Cytoplasmic MTs do not emanate directly from the centrioles within the centrosome. Instead, cytoplasmic MTs emanate from the periocentriolar material and therefore this amorphous material is likely the true microtubule-organizing center. Gamma-tubulin (y-tubulin) is found within the periocentriolar material but it is not incorporated into the walls of the MTs themselves (reviewed by Luduena et al., 1992; Joshi 1994). y tubulin appears essential for initiation of M T assembly as well as the anchorage of MTs to the MTOC. y tubulin interacts physically with the 6-subunit of the a6 tubulin dimer, thus orienting the a subunit at the plus ends of a M T (reviewed by Luduena et al, 1992; Mandelkow and Mandelkow, 1995). The net effect of the M T O C appears to be the  34 inhibition of tubulin polymerization from the minus end (Cassimeris 1993), the end that is sequestered at the M T O C . As a result, intracellular MTs in fibroblasts have only one free end from which to grow or disassemble. ii. Epithelium In some E cells the organization of MTs differs, depending upon the confluency of the cells in culture. In isolated M D C K cells, MTs originate from a broad region containing the centrioles but after intercellular contacts are established, nucleating material is relocalized and nucleation can occur throughout the cytoplasm (Bre et a l , 1987, 1990).  b. M T Assembly Within cells, the minus ends are located within the MTOC and the plus ends radiate outwards into the cytoplasm, leaving only the plus ends available for polymerization and depolymerization. Within a single cell, most MTs are growing slowly and a few MTs are rapidly shrinking. Infrequent but rapid transitions between these phases occur asynchronously and in an apparendy random fashion within a population of MTs (reviewed by Caplow, 1992). The transitions of MTs from growth to rapid shrinkage (termed 'catastrophe') or transition from rapid shrinkage to growth (termed 'rescue') are explained by a model termed dynamic ^stability (Mitchison and Kirschner, 1984) which appears to be the major pathway of M T turnover in living cells (reviewed by Cassimeris, 1993).  The transition frequencies between growth and shrinkage differ throughout the cell and with different phases of the cell cycle. For example, in fibroblasts, MTs at the cell periphery demonstrate a higher turnover rate and higher transition frequency than MTs near the cell center (Sammak et al, 1987). Several studies have reported M T growth rates of 3.5-7 ^tm/minute and shrinkage rates of 14-17 (im/minute but transitions from shrinkage to growth phases are not completely random (reviewed by Gelfand and Bershadsky, 1991). For example, growth rates for a M T are not constant even during one period of growth and rescues from shrinkage can occur  35 several times at the same site with MTs regrowing, several times, along the same paths (Schulze and Kirschner, 1988). As well, although similar rates of elongation and rapid shortening can be observed in interphase and mitosis, the respective transition rates between elongation and shrinkage differ in interphase and mitotic cells (Cassimeris 1993).  In cell cultures, the dynamic behaviour of interphase MTs is also cell-type specific and is attributed to differences in the frequency of rescue (Shelden and Wadsworth, 1993). For example, MTs in cultured E cells are less dynamic than MTs in fibroblasts (Pepperkok et al., 1990; Wadsworth and McGrail, 1990). That is, MTs in E cells appear more stable because they display an increased average half-life (or reduced turnover rate) that may be related to the increased length and number of MTs in E cells (Wadsworth and McGrail, 1990). In E cells, increased confluency also decreases M T dynamics (Pepperkok et al., 1990), increases M T stabilization and causes a relocation of MT-nucleating material throughout the cytoplasm (Bre et al., 1990). In contrast, in fibroblasts, M T dynamics are not affected by cell density (Pepperkok et al., 1990; Nagasaki et al., 1992).  Non-tubulin, inter and intracellular regulatory factors can influence M T elongation, catastrophe and rescue, as well as their transition frequencies. Specifically, post-translational modifications of MTs and the binding of MT-associated proteins (MAPs) to MTs can influence the reactivity and longevity of MTs.  3.  Post-translational Modifications  Microinjection experiments with tubulin monomers revealed that the majority of interphase MTs are very dynamic and are rapidly exchanging tubulin units. However, 10-20% of MTs exchange subunits more slowly, requiring up to an hour (Kreis 1987) and even 16 hours (Webster et al., 1987) before evidence of the labelled tubulin appears. In addition to being less dynamic, this population of MTs is also more resistant to depolymerization by M T antagonists (Kreis 1987;  . '  36  Piperno et al.; 1987; Khawaja et a l , 1988) (eg. colcemid, nocodazole, see Chapter 1, Section V.A.6.b.).  Four types of biochemical, post-translational modifications of tubulin have been identified. However, a clear cause-effect relationship between these modifications and alterations in M T dynamics or M T resistance to depolymerizing by agents such as nocodazole has not been identified. That is, the presence of modified tubulin is neither required nor sufficient to decrease M T dynamics or confer nocodazole-resistance (Schulze et a l , 1987; Khawaja et al., 1988; Webster et al., 1990).  The modifications observed in stable MTs appear to be explained best by the observation that these MTs exist long enough to permit the particular modification to be completed (Gelfand and Bershadsky, 1991). Therefore, post-translational modifications are the result of M T stability and not the cause. Overall, modification may be a biochemical signal of M T stability to other cellular components that could utilize or further stabilize the MT. For example, detyrosination (see V.A.3.d. below) by itself does not stabilize MTs but some MAPs associate specifically with detyrosinated M T and maintain their stability (reviewed by Murphy 1991).  a. Phosphorylation This modification occurs in a 6 tubulin isoform that is neuron-specific. The functional significance of this modification is unclear (Luduena et al., 1992).  b. Glutamylation This modification consists of the addition of up to 6 glutamate residues to a glutamate residue in a-tubulin and 6-tubulin but the functional significance of these changes is not known (Joshi and Cleveland, 1990; Luduena et al., 1992).  c. Acetylation Acetylation occurs on a specific lysine residue on a tubulin (Piperno et al, 1987). This type of modification has been associated with stable and ordered arrays of MTs in 3T3  37 fibroblasts (Prescott et al., 1989) and HeLa epithelial-like cells but was not observed in PtK2 kidney epithelial cells (Piperno et al, 1987). However, it is unclear whether acetylation confers stability directly to MTs or whether acetylation is a signal for other factors which subsequently confer stability (Piperno et al, 1987).  Acetylated a-tubulin is also present in axonemes, basal bodies, primary cilia, midbodies and centrioles. Assembly of cytoplasmic MTs starts in the region of the centrioles but acetylation of atubulin in interphase MTs lags behind polymerization of the majority of MTs. Conversely, the deacetylation of the a-tubulin subunits occurs faster than the depolymerization of tubulin (Piperno et al., 1987).  d. Tyrosination/Detyrosination Some isotypes of a tubulin are synthesized with a tyrosine as their carboxyl-terminal amino acid. Carboxypeptidase B can remove this tyrosine from either the tubulin monomer or polymer and expose the glutamic acid residue. This form of tubulin is referred to as Glu-tubulin.  Tyrosine can also be re-added to tubulin monomers by ATP-driven tubulin-tyrosine ligase but the extent of tubulin tyrosination does not appear to affect its polymerization into MTs. MTs containing the tyrosinated-tubulin are referred to as Tyr-tubulin (Thompson 1982; Gundersen et al., 1984).  Typically, individual MTs in vivo contain varying amounts of Tyr and Glu a- tubulin. In vivo, Tyr MTs appear to arise from a pool of soluble tubulin monomers containing only Tyr tubulin. Glu MTs arise through a time-dependent, post-polymerization modification of Tyr MTs in which tyrosine is removed by the carboxypeptidase. A l l long-lived MTs become significantly detyrosinated as Glu MTs and therefore Glu-MTs are associated with increased stability. In contrast, MTs that are turning over rapidly, do not exist long enough to become detyrosinated and  38 achieve significant levels of Glu tubulin; therefore they remain as Tyr MTs. When Glu MTs eventually do depolymerize, the resulting Glu monomers are retyrosinated and the cycle is completed (reviewed by Bulinski and Gundersen, 1991).  In interphase cells, MTs are heterogeneous in their dynamics and distribution. Tyr-MTs are more prevalent and they radiate from the MTOC and reach the cell margins. Glu-MTs are typically located near cell centres, near the MTOC, and appear more sinuous than Tyr-MTs (Gundersen et al., 1984). As well, in fibroblast wounding experiments, cells immediately adjacent to the wound acquire an array of Glu MTs oriented towards the direction of impending migration into the wound. This array appears rapidly and in conjunction with the reorientation of the centrosome towards the wound edge and precedes the actual translocation of the cells (Gundersen and Bulinski, 1988). Glu MTs are also more resistant to depolymerization by MT-antagonists (eg colchicine) (Khawaja et al., 1988) and are involved in contact inhibition of motility (Nagasaki et al., 1992). When fibroblasts become surrounded by virtue of cell-cell contact, they lose their characteristic array of Glu MTs as Glu MTs are lost from regions of the cytoplasm subjacent to the contact sites. However, the Glu MTs persist in free cell margins in noncontacting areas and their distribution is not related to the orientation of the M T O C or the cessation of forward locomotion.  e. Detyrosination/Deglutamylation Recently, it was discovered that a-tubulin could also be modified by the removal of 2 carboxylterminal residues, tyrosine and glutamic acid, to yield A-tubulin. A-tubulin cannot be retyrosinated by tubulin tyrosine ligase (see Section V.A.3.d. above) and it forms very stable MTs in neurons (Mandelkow and Mandelkow, 1995).  4.  Proteins Associated with MTs  As their name implies, microtubule-associated proteins (MAPs) are proteins that are associated with MTs. However, MAPs may also interact with actin and intermediate filaments and this  39 discussion is presented in Section V.A.5.b. and V.A.5.C. below.  By operational definition, MAPs are proteins that associate with tubulin through in vitro cycles of assembly and disassembly; however, some proteins (eg. kinesin and dynein) are associated with MTs in vivo but do not recycle with MTs in vitro (Luduena et al., 1992; Mandelkow and Mandelkow, 1995). Therefore, MAPs are broadly defined as proteins associated with MTs in vivo. A l l MAPs can be divided into two groups according to their sensitivity to nucleotides, GTP or A T P (reviewed by Chapin and Bulinski, 1992; Hirokawa 1994). Those M A P s whose binding to MTs is sensitive to nucleotides also possess MT-activated ATPase or GTPase activity and this group includes kinesin, dynein and dynamin. MAPs can also be distinguished functionally (Mandelkow and Mandelkow, 1995) and this grouping is utilized in this thesis.  a. Structural or Nucleotide-insensitive Proteins Often, the term ' M A P s ' is used to designate only that group of microtubule-associated proteins that are nucleotide-insensitive and structural. Thus, MAPs can be described as nucleotide-insensitive because their association with MTs is not influenced by ATP or GTP. M A P s are also structural because they bind to, stabilize and promote the assembly of MTs. MAPs includes M A P 1 A , IB, M A P 2A, 2B, 2C, tau, big tau and M A P 4 (reviewed by Hirokawa 1994). The M A P I species and M A P 4 have a widespread tissue distribution but the M A P I species is enriched in amount in neuronal cells. M A P 2 and tau are neuron specific and are the best characterized MAPs.  The general, shared characteristics and abilities (reviewed by Wiche et al., 1991; Chapin and Bulinski, 1992; Hirokawa, 1994) of structural MAPs are discussed below. A l l members of this group can stimulate the assembly of MTs as well as stabilize MTs in vitro but only some members can produce similar effects in vivo, in microinjection and transfection studies (Chapin and Bulinski 1992).  40 Structural M A P s are fibrous phosphoproteins and the degree of their phosphorylation can alter their shape and hence their function (Wiche et al., 1991; Chapin and Bulinski, 1992; Hirokawa 1994). For example, phosphorylation of tau and MAP2 reduces their abilities to stimulate M T polymerization and ATP may indirecdy control the level of M A P phosphorylation (reviewed by Gelfand and Bershadsky, 1991; see also Friden and Wallin, 1991).  Structural M A P s possess a site or 'projection domain' whose function is unknown but has been implicated in cross-bridging MTs to each other and to other cytoskeletal elements such as intermediate filaments and to organelles (reviewed by Chapin and Bulinski, 1992; Luduena et al., 1992).  Structural M A P s also possess a site or domain for binding MTs and this site also appears to function independently from the rest of the M A P molecule to ionically bind MTs, stimulate M T assembly and stabilize MTs from depolymerization (reviewed by Chapin and Bulinski, 1992; Luduena et al., 1992; see also Chapin and Bulinski, 1994). Each MT-binding site on a M A P can bind a tubulin molecule to form a nucleus or a critical mass for polymerization.  Structural M A P s may also assemble and stabilize MTs by crosslinking or cross-bridging tubulin subunits. Different M A P isoforms contain a variable number of MT-binding sites and if each site in a M A P binds to a tubulin molecule, then a single M A P could interact with several tubulins and further stimulate M T assembly (Luduena et al., 1992). Differences in the number and spatial array of MT-binding sites within different MAPs are also reflected in differences in the lateral bundling of MTs (eg. Chapin and Bulinski, 1992; Weisshaar et al., 1992; Edson et al., 1993) and specific spacings between MTs may be important for unimpeded directed transport along MTs. MAPs, either alone or in conjunction with other proteins, may also serve as spacers between MTs and other cytoskeletal elements and organelles or vesicles destined for MT-based transport (reviewed by Gelfand and Bershadsky, 1991; Wiche et al., 1991; see also Chapter 1, Section V.A.5.a.).  41 b. Motor or Nucleotide-sensitive Proteins M T motor proteins are functionally defined as MT-associated enzymes that use the energy released from nucleotide hydrolysis to generate force and motion, relative to the M T (Skoufias and Scholey, 1993).  MT-associated motors contain two superfamilies of homologous motors, the dyneins and kinesins, which can be distinguished, in in vitro motility assays by their direction of movement along MTs. Recently, the two families have been more accurately distinguished on the basis of their sequence similarities as bidirectional motors within each family have been discovered (Endow 1991). The following discussion is limited to dynein and kinesin in interphase cells although some authors also include dynamin as a superfamily. However, it is unclear whether this protein is a true MT-based motor or whether it functions as a regulatory GTPase in sorting membranes (Bloom, 1992; Skoufias and Scholey, 1993).  i. Dynein (see reviews by Schroer 1991, 1994a, b) Dynein is a motor generally associated with transport towards the minus ends of MTs although at least one isoform can move bidirectionally, depending upon its phosphorylation (Endow 1991). Dynein is credited with mediating the distribution of the Golgi (Wehland and Willingham, 1983; reviewed by Vale, 1987).  ii. Kinesin Kinesin can track along single M T protofilaments, moving towards the plus ends. Although single protofilaments are the smallest track for kinesin-directed transport, parallel protofilaments are necessary for continuous, smooth transport. Kinesin may be able to interact with more than a single M T at the same time (Hoyt, 1994), but there appears to be just one kinesin-binding site per aB tubulin dimer (Howard, 1993; see also Svoboda et al., 1993).  42 Kinesin is a cytoplasmic motor associated with movement towards the plus ends of MTs, directing membrane transport from the cell center outwards towards the plasma membrane. Kinesin has been associated with mitochondria (Goldman 1971; Ball and Singer, 1982; Bershadsky et al., 1991; Leopold et al., 1992), the movement of secretory vesicles between the Golgi region and the endoplasmic reticulum (Lippincott-Schwartz et al., 1995; reviewed by Cole and LippincottSchwartz, 1995), extension of the endoplasmic reticulum (Terasaki et al.,1986; Dabora and Sheetz, 1988) and transport of lysosomes (Hollenbeck and Swanson, 1990) and pigment granules (Rodionov et al., 1991) to the cell periphery (reviewed by Schroer, 1991; Vale, 1987).  In many cells, the intracellular distribution of intermediate filaments (IF) is dependent upon MTs (Terasaki et al., 1986; Hynes and Destree, 1978a; Blose et al., 1984; Hollenbeck et al., 1989) and upon kinesin, specifically, because microinjection of anti-kinesin results in the perinuclear collapse of IFs. Kinesin may form cross-bridges between MTs and IFs or, perhaps an unidentified membranous component connects IFs to MTs (Gyoeva and Gelfand, 1991). Continuous transport of this membranous component by kinesin towards the cell periphery, would also maintain the distribution of IFs. However, anti-kinesin would disrupt transport of the membrane component and therefore, also disrupt the distribution of IFs (BloOm, 1992).  Immunofluorescence staining for kinesin reveals a punctate codistribution along MTs (Hollenbeck, 1989; Hollenbeck et al., 1989; Okuhara et al., 1989) and IFs (Gyoeva and Gelfand, 1991). However, the colocalization of kinesin and MTs may more accurately reflect the association of kinesin-positive vesicles or organelles along the MTs (Pfister et al., 1989; Brady and Pfister, 1991; Leopold et al., 1992).  Recently, kinesin was identified as a specific mediator of polarization because antibodies specific to kinesin, when injected into fibroblasts, inhibit polarity very much in the same way as colchicine (Rodionov et al., 1993) (see also Section V.A.6.b. below). In instances in which cytoplasmic  43 MTs have been disrupted by agents such as colchicine, kinesin appears to codistribute with some stress fibers, in the absence of M T (Okuhara et al., 1989).  5.  Interactions of MTs with other cytoskeletal elements and cellular components  The interaction of MTs with other cytoskeletal elements and cellular components forms high-order assemblies of which MTs are an integral part, and this interaction may be another level of control of M T dynamics and behaviour within the cell (Gelfand and Bershadsky, 1991).  The association and interaction of MTs and MT-motors with cellular structures are summarized below. Many of the intracellular functions of MTs and the nature of their interactions with other structures within the cell have been determined by observing the consequences when MTs are disrupted by various agents (see Section V.A.6.b. below).  a. Intracellular Traffic The cytoskeleton and in particular MTs are important in mediating the transport of vesicles and organelles between different cell compartments, the plasma membrane and the cell exterior. MTs are involved in both secretory and endocytic pathways and it appears that a stable population of MTs (Cole and Lippincott-Schwartz, 1995) are associated with the Golgi (Wehland and Willingham, 1983; reviewed by Kreis, 1990) and elements of the endoplasmic reticulum (ER) (Terasaki et al., 1986; Dabora and Sheetz, 1988) although their distribution appears to be mediated by dynein and kinesin, respectively. When MTs are depolymerized, the ER gradually retracts to the cell center and Golgi stacks disperse to the cell periphery (reviewed by Vale, 1987). These structures remain functional but membrane traffic between organelles and to the cell surface becomes much less efficient (Cole and Lippincott-Schwartz, 1995).  Intracellular transport was believed to be mediated solely by the dynein and kinesin motor families along MTs (reviewed by Vale, 1987) but evidence is increasing for actin-based transport involving  44 myosin motors and actin microfilaments (see Chapter 1, Section V.C.). Thus, some functions of the actin and M T systems are interrelated and even partially redundant (reviewed by Fath and Burgess, 1994; Hoyt, 1994; Langford, 1995).  Vesicles or organelles may contain more than one motor; that is, both an actin-based motor and a MT-based motor. For example, vesicles derived from the Golgi possess both myosin I and cytoplasmic dynein (Fath et al., 1994). Alternatively, it is possible that a single motor may be capable of moving along both MTs and actin tracks. Organelles in extruded squid axoplasm can move unidirectionally, in an ATP-dependent manner, along both MTs and actin microfilaments (MF) and switch between them (Kuznetsov et al., 1992, 1994). Movement of organelles along the MTs is faster but in the nocodazole-induced (see Chapter 1, Section V.A.6.b.ii.) absence of MTs, movement persists along MFs possibly due to a myosin-like motor that is capable of moving on both actin and tubulin tracks (Kuznetsov et al., 1992, 1994; reviewed by Titus, 1993). As well, in fibroblasts, kinesin colocalizes with some stress fibers in the absence of MTs, although its motor activity along the stress fibers was not specifically identified (Okuhara et a l , 1989).  The majority of the data currently supports a model in which the motors that provide the force for directed particle transport are bound to an organelle or vesicle and interact transiently with a track, either a M T or actin M F (reviewed by Wiche et al., 1991; Leopold et al., 1992). Most organelles have the capacity to bind both plus- and minus-end directed motors and shared activator molecules (of both motors) may control the direction of organelle movement (Cole and Lippincott-Schwartz, 1995). A model which has emerged from the work of Kuznetsov et al., (1992) in squid axoplasm, further suggests that actin filaments provide transport within local regions of the cytoplasm. MTs provide the tracks for transport of organelles over long distances and provide the mechanism for translocation of organelles from one functional site in the cell to another (Langford 1995).  45 b. Intermediate Filaments In many cultured cell types, intermediate filaments (IF) are codistributed, in a time-dependent manner, along MTs (eg. Hynes and Destree, 1978a; Blose et al., 1984; Terasaki et al., 1986) and this coalignment depends upon kinesin (Gyoeva and Gelfand, 1991). When MTs are depolymerized in some cultured cell types, IFs collapse into bundles or coils around the nucleus (Hynes and Destree, 1978a; Terasaki et al., 1986; Hollenbeck et al., 1989).  c. Actin Microfilaments and Focal Contacts The M T and actin microfilament (MF) systems are closely interrelated functionally as well as structurally and both interactions may be mediated by MAPs and associated motor proteins (reviewed by Langford, 1995). i. structural interactions The extent to which the actin and M T intracellular-transport systems appear to be coordinated and possibly redundant suggests that the two systems may be physically linked, perhaps by a motor complex such as dynactin (Lees-Miller et al., 1992; reviewed by Langford, 1995).  As well, MTs are observed to interact with cytoplasmic actin MFs to form end-to-side contacts (Schliwa and van Blerkom, 1981) and MTs can bind and interact direcdy with actin MFs (Holifield and Heath, 1992). MTs frequently change their course at sites of contact with actin MFs and bends in MTs appear to be the result of their structural interaction with other cytoskeletal elements (Schliwa and van Blerkom, 1981). Some MTs terminate close to the plasma membrane but other MTs change direction and bend or curve near the plasma membrane to conform to the contours of the cell (eg. Weber et al., 1976). This bending behaviour of MTs may demonstrate that tensile forces exerted by actin MFs of the cell cortex can determine the placement of MTs (Holifield and Heath, 1992), (see also Chapter 1, Section V.C.7.b).  The state of organization of MTs appears to modulate the assembly and mobility of cortical  46 structures (reviewed by Wang, 1991) including the recruitment of cytoplasmic myosin into the cortex of some cells (Bornens et al., 1989). The state of organization of MTs can also modulate the location and arrangement of actin MFs and of vinculin (a marker of focal contacts; see Chapter 1, Section V.D.) within the cell (David-Pfeuty, 1983); however, neither MTs nor IFs are required for the formation or maintenance of actin MFs and vinculin patches (Goldman, 1971; DavidPfeuty, 1983; Gotlieb et al., 1983; Forry-Schaudies, 1986; Hollenbeck et al., 1989).  During the formation of cell processes (see Chapter 1, Section V L B . , Section VI.C and review by Condeelis, 1993) cytoplasmic protrusions are initially devoid of cytoskeletal substructure. The transition of these initial protrusions to established protrusions is marked by the development of actin-fiber meshworks and the appearance of MTs (Fisher et al., 1988; DeBiasio et al., 1988; Conrad et al., 1989). MTs appear to stabilize the protrusions as retraction of a protrusion occurs only in areas where MTs are absent (Conrad et al., 1989; Geiger et a l , 1984b). MTs also direct and potentiate the formation of cell-substratum contacts (Geiger et al., 1984b; Rinnerthaler et a l , 1988) and subsequently, the formation of tension between stabilized attachment sites directs formation of actin M F bundles (Greenspan and Folkman, 1977; Kolega, 1986). Cell spreading subsequently continues along the axis of tension and protrusions in directions other than those parallel to the direction of tension are restricted (Kolega, 1986). In turn, the tensile force generated by actin MFs in nonmuscle cells can also determine the placement of MTs in the cytoplasm (Holifield and Heath, 1992).  ii. interactions with structural MAPs The in vivo use of M T - or stress fiber-perturbing agents (see Chapter 1, Sections V.A.6.b. and V.C.7.b., respectively) has demonstrated a biochemical interrelationship between MTs and actin MFs and this interaction may be mediated by MAPs (eg. Pollard et al., 1984). The induced depolymerization of MTs could release MAPs which would be available to interact with the actin system and stimulate its contractility and organization into stress fibers (Danowski 1989; Danowski  47 et al., 1992; Kajstura and Bereiter-Hahn, 1993; Keller and Niggli, 1993). Likewise, the induced depolymerization of actin stress fibers could release MAPs or, free binding sites on M A P s which could in turn inter-react with MTs to stabilize and bundle MTs into lateral arrays (see ii.(b) below).  (a) Depolymerization of MTs Cells exposed to MT-inhibitors demonstrate an increase in their intracellular content of filamentous (F) actin of between 10% (Carlsson and Blikstad, 1981) and 20% (Blikstad and Carlsson, 1982), measured by DNAse assay, and as high as 140%, measured by microfluorimetry (Kajstura and Bereiter-Hahn, 1993). Moreover, the depolymerization of MTs also induces the formation of Factin bundles and new focal contacts in living density-inhibited fibroblasts (Danowski et al., 1992; Kajstura and Bereiter-Hahn, 1993). The functional significance of the increased F-actin content and its organization into stress fibers appears to be the increased contractile activity of these cells (Danowski, 1989).  fb) Depolymerization of Actin M F Bundles Cells exposed to actin-perturbing agents such as cytochalasin lose their actin M F bundles or stress fibers and develop arborized processes which contain collections of tightly and laterally-bundled MTs (Weber et al., 1976; Bliokh et al., 1980; Domnina et al., 1982). M T inhibitors can block and reverse disruption of the actin system (Danowski 1989; Danowksi et al., 1992; see also Lyass et al., 1988).  iii. Possible Mechanisms for M A P Interactions between MTs and actin MFs The interactions between MTs and actin MFs may be mediated by M A P s (eg Pollard et al., 1984; Griffith and Pollard, 1982) that exhibit tubulin-binding domains and at least one actin-binding domain (Keller and Niggli, 1993; Pedrotti et al., 1994; reviewed by Langford 1995). M A P s have demonstrated binding with globular (G) and filamentous (F) actin (see Chapter 1, Section V.C.I.). In addition, actin-crosslinking activity (Pedrotti et al., 1994) has been identified as has an  48 association of M A P s with actin stress fibers in cultured cells (Asai et al., 1985; Cross et al., 1993). Furthermore, at least one F-actin bundling protein which can bundle both actin MFs and MTs, forming co-bundles of both filaments, has been discovered (Itano and Hatano, 1991).  6.  Physical and Chemical Interactions with MTs that affect M T Dynamics  As discussed in Section V.A.5. above, the physical and biochemical interactions of MTs with other cytoskeletal elements can modify their behaviour and distribution. This section will expand discussion of (a) physical interactions and (b) chemical agents that perturb M T structure and assembly.  a. Effects of the physical environment In vitro, solutions of purified tubulin and MAPs form highly deformable polymers and these observations suggest that within cells, MTs form a cytoplasmic network that is not very rigid. Nevertheless, the large diameter, great length and relative stiffness of MTs, compared to actin MFs and IFs, allow MTs to serve as stabilizing elements within a cytoskeletal network containing the other filaments (Janmey et al., 1991).  Nevertheless, polymerizing MTs can generate mechanical forces strong enough to deform membranes. For example, tubulin polymerizing inside unsupported phospholipid bilayers of liposomes can distort free-floating liposomes to produce bipolar, cigar-shaped liposomes. MTs polymerizing inside liposomes attached to glass surface, produce liposomes which have multipolar projections of different diameters and with no preferred orientation (Hotani and Miyamoto, 1990). In fact, similar protrusions can be formed by MTs bundles in MAP-transfected cells, in the presence of cytochalasin (Edson et al., 1993).  However, within the cytoplasm, the course of MTs is altered upon direct interaction with actin filaments and the cell cortex and by tensile forces (Holifield and Heath, 1992; see Section  49 V.A.5.c.i. above). These observations may be explained by different models of M T behaviour and cell organization.  i. Compression and Tension Effects upon M T Assembly MTs are very sensitive to compression (Hill 1981; Hill and Kirschner, 1982a, b; Buxbaum and Heideman, 1988) and in fact, M T sensitivity to compression might be a mechanism for regulating the length of the polymers (Hill and Kirschner, 1982a).  MTs that are polymerizing and encounter a force or physical barrier experience compression and therefore, their growth is altered. For example, the free end of a M T polymer could theoretically continue to grow as long as free monomer was available or, until the polymer encountered a physical barrier. In that event, further addition of monomer would produce a rising compressive force that would stop growth. However, since MTs demonstrate great deformability, the polymer could bend away from the barrier and thus, alternative directions for growth would become available, avoiding compression effects. Conversely, tension appears to promote M T assembly (Hill and Kirschner, 1982b).  ii. Tensegrity Model of Structural Organization One model of the structural organization of cells, the tensegrity model (reviewed by Ingber 1993) is predicated on the balance of mechanical forces between competing M T struts and contractile M F networks such that MTs hold the cell outward against the contractile forces generated by the actin system. As well, the contractile forces borne by MTs can be transferred to and be borne by the substratum in a complementary and interchangeable fashion, through sites of cell attachment of the substratum.  The predictions of this model (Buxbaum and Heidemann, 1988; Ingber, 1993 ) have been tested in neurites (Dennerll et al., 1988); under conditions of increased compression, the concentration of  50 tubulin monomers increases at the expense of polymerized tubulin. The tensegrity model also predicts that increased loads are borne by the substratum when MTs are depolymerized and this has been demonstrated in fibroblasts (Danowski, 1989; Kolodney and Wysolmerski, 1992). As well, MTs in spreading hepatocytes experience different mechanical loads, depending upon the cell's ability to transfer forces from the MTs onto the substratum (Mooney et al., 1991). With decreased mechanical loads on MTs, polymerization of MTs is favoured and the concentration tubulin monomer decreases; conversely, increased mechanical loads on MTs results in their depolymerization and the concentration of monomer increases.  iii. Effect of Shear-Free Planes MTs in plant cells also respond to altered mechanical environments (Williamson, 1990; reviewed by Williamson, 1991) by demonstrating new axes of elongation along shear-free planes (Hush and Overall, 1991). Apparently, when shear-sensitive elements such as MTs, are exposed to anisotropic stress, either compression or tension, they respond by aligning in the shear-free plane which is defined as the plane perpendicular to the direction of the largest principal stress (Lintilhac, 1974).  b. Chemical Agents Because of the central role of MTs in mitosis (see Alberts et al., 1989; Darnell et al., 1986), several of the MT-perturbing agents were first recognized and identified through their anti-mitotic effects. Heat-shock, cold-treatment and high hydrostatic pressures can affect the mitotic spindle and produce polyploid cells. Similar effects are produced by different families of pharmacologic agents which interfere with mitosis by binding to different sites within the tubulin molecule and interfering with M T assembly. These pharmacologic agents include colchicine, nocodazole, vinca alkaloids, estramustine phosphate and taxol. However, discussion will be limited to colchicine and nocodazole (see Appendix 2) as only these tubulin ligands were used in studies comprising this thesis.  i. Colchicine and its derivatives Colchicine is an alkaloid extracted from the meadow saffron (Colchicum automnale). Its structure comprises 3 rings and derivatives of colchicine result from deletions or substitutions within these rings. However, some structural features are essential for anti-MT activity as evidenced by lumicolchicine which is missing one ring, does not bind to tubulin and therefore does not disrupt MTs (Dustin 1984). Nevertheless, lumicolchicine is a useful analogue to distinguish the specific tubulin-binding effects of colchicine from possible non-specific effects (Berlin and Fera, 1977; Aszalos et a l , 1985; 1986).  A commonly-used derivative of colchicine is colcemid or demecolchicine which is more reversible in action than colchicine because of differences in binding affinity to tubulin (Dustin, 1984).  Colchicine binds slowly to tubulin, much slower than the rate of M T assembly. Although each dimer of tubulin can bind a molecule of colchicine, M T assembly can be prevented when less than 4% of tubulin subunits are bound to colchicine (Dustin 1984). These observations are likely the result of colchicine binding to the high affinity site on G-tubulin to which colchicine binds irreversibly and in a temperature-sensitive fashion. However, colchicine can also bind to a second tubulin site at which binding is reversible and of lower affinity (Luduena and Roach, 1991).  Colchicine and its analogues are classed as sulfhydryl-directed reagents because they affect the accessibility of sulfhydryl groups in tubuhn. Sulfhydryl groups are important for the formation of intra-chain crosslinks in 8-tubulin and some of these sulfhydryl groups are critical to the assembly of tubulin (see section V . A . I , above). Colchicine binds to B-tubulin in a region which contains an assembly-critical sulfhydryl group and is also close to the nucleotide-binding site (Shearwin and Timasheff, 1994). Binding of colchicine appears to either (i) induce a conformational change in tubulin making the sulfhydryl less accessible or (ii) stabilize tubuhn and slow down an unfolding process which would normally expose the sulfhydryl groups (Luduena and Roach, 1991;  52 Shearwin and Timasheff, 1994). Thus, binding of colchicine to either the tubulin dimers or assembled MTs suppresses the formation of cross-linked aggregates of tubulin and the net effect is that polymerization of tubulin and further growth of MTs ceases. However, depolymerization of MTs is not affected by colchicine, and therefore existing MTs shrink and disappear, leaving only the centrioles and M T O C intact (Darnell et a l , 1986).  ii. Nocodazole Nocodazole (NC) is a member of the benzimidazole family and binds to tubuhn in a 1:1 ratio. N C can inhibit the binding of colchicine and therefore NC is thought to bind to the same region of tubulin as colchicine. However, the action of NC on MTs is more reversible than the action of colchicine (Dustin, 1984).  7.  Effects of M T Depolymerization on Cell Behaviour  The effects of M T depolymerization upon cell behaviour are variable, depending upon the cell type and may reflect differences in the organization of the cytoskeletons in the different cell types.  The disruption or lack of MTs does not prevent the attachment, spreading or motility of E cells (Dipasquale 1975; Vasiliev et al., 1975; Domnina et al., 1985; Middleton et a l , 1988), keratinocytes (Eutener and Schliwa, 1984), macrophages (Cheung et al., 1978; Glasgow and Daniele, 1994), polymorphonuclear leukocytes (PMNs) (Keller and Niggli, 1993) or fragments of PMNs (Malawista and de Boisfleury Chevance, 1982).  The effect of M T disruption on some fibroblasts such as chick heart fibroblasts, varies with their time in culture (Middleton et al., 1989). However, after the first 48 hours in culture, chick heart fibroblasts behave similarly to other fibroblastic cell lines whose M T system has been disrupted by MT-depolymerizing agents.  MTs are not required for fibroblasts to attach and spread although spreading is delayed and more variable (Kolodney 1972; Ivanova et al., 1976, Middleton et al., 1988) as compared to M T containing control fibroblasts. MT-deficient cells are also rounder and larger (Gail and Boone, 1971), likely due to the increased number of multinucleated cells (Vasiliev et al., 1970).  The most striking effect of MT-depolymerizing agents upon fibroblasts is that the cells lose their ability to polarize and to directionally locomote. Furthermore, these effects occur consistently and regardless of whether the substratum is tissue culture plastic (Kolodney 1972), glass coverslips (Ivanova et al., 1976), or collagen-coated glass coverslips (Middleton et al., 1988). As well, MTdeficient cells lose their ability to be aligned and guided along the topography of fish scales (Vasiliev et al., 1970).  Even after 72 hours, MT-deficient cells remain irregular and polygonal in shape (Kolodney 1972) as cells continue to ruffle around their entire cell peripheries with intermittent extension and retraction of protrusions. Stable leading edges fail to develop and cell translocation is either absent or random without persistence (Vasiliev et al., 1969; 1970; Gail and Boone, 1971; Goldman 1971; Kolodney 1972, Ivanova et al., 1976; 1980; Middleton et al., 1988) although occasional random displacements of some cells do result in collisions with adjacent cells (Vasiliev et al., 1970).  54 B.  Intermediate Filaments  Intermediate filaments (IF) will be discussed very briefly as they do not appear to play a central role in the observations significant to this thesis. As well, observations of IFs in this thesis were limited to fibroblasts. Typically, the cytoplasmic distribution of vimentin filaments in spreading cells is preceded by the distribution of MTs by several hours (see Section V.B.2.a. below) and therefore IFs do not appear to play a significant role during these early events. Furthermore, in culture conditions in which MTs are depolymerized, IFs are collapsed into perinuclear bundles yet their collapse does not affect spreading or directed motility (see Section V.B.5. below and also Chapter 4, Section III.).  1.  Structure and Distribution  TPs are cytoplasmic nonpolar protein fibers of 8-12 nm thickness which is 'intermediate' in size between MTs (24 nm) and actin filaments (7 nm), (Alberts et al., 1989; Darnell et al., 1986). Over 90% of cytoplasmic IF protein is assembled into filaments and during interphase, IFs are concentrated in the juxtanuclear region as IF 'caps' or perinuclear rings from which they radiate into the cytoplasm in atime-dependentmanner and in delayed association with MTs (eg. Hynes and Destree, 1978a; Terasaki et a l , 1986). Assembly of IFs appears to require a peri-nuclear organizing center or region from which filament assembly is initiated and candidates include the centrosomal region and the nuclear envelope and pore complexes (Skalli and Goldman, 1991). However, IF protein subunits can also be incorporated into a preexisting IF network, suggesting that IF subunits can exchange with polymerized IFs in a steady state or dynamic equilibrium.  2.  Interaction of IFs with other cytoskeletal elements and cellular components  a. MTs (See also Chapter 1, Section V.A.5.b.) In cultured cells, the interaction of IFs with MTs is unidirectional in that the distribution of IFs appears to depend upon MTs but the distribution of MTs does not depend upon IFs. That is, the disruption of IFs by microinjected anti-vimentin antibodies (Gawlitta et al., 1981; Klymkowsky  55 19981; Lin and Feramisco, 1981) has no apparent effect on M T organization, intracellular transport or even upon cell shape or motility (Gawlitta et al., 1981; Klymkowsky 1981; Lin and Feramisco, 1981).  Conversely, the disruption of MTs (eg. Hynes and Destree, 1978a; Wehland and Willingham, 1983; Blose et al., 1984; Forry-Schaudies et al., 1986) or kinesin function (Gyoeva and Gelfand, 1991) leads to the perinuclear collapse of IFs. In experiments using colchicine or its analogues to disrupt MTs, both MTs and IFs are affected and both will reform upon removal of the drug. The regrowth of IFs lags behind the regrowth of MTs, similar to the sequence of events observed in spreading cells (Hynes and Destree, 1978a). At least 4 hours are required before IFs codistribute with MTs which extend throughout the cytoplasm and to the plasma membrane. However, even in well-spread cells, IFs are not commonly observed in lamellae (Hynes and Destree, 1978a, Terasaki et al., 1986) and when present, IFs advance into lamellipodia after MTs (Terasaki et al., 1986).  b. Actin Microfilaments The interaction between IFs and MFs is likely of low affinity. Hence, the interaction between MTs and IFs appears to be dominant to the interaction between MFs and IFs (Cary et al., 1995).  Both MTs and IFs interact with actin MFs (eg. Pollard et al., 1984) and can form end-to side contacts with MFs within the cell (Schliwa and van Blerkom, 1981). In fibroblasts, IFs are also observed to run parallel to stress fibers towards adhesions sites at cell peripheries and nodules along M F bundles may act as foci from which IF diverge throughout the cell (Green et al., 1986).  IFs appear to be attached to the nuclear membrane, MTs and components of the actin cortex (see Chapter 1, Section V.C.3.c.iii.d.) and Bershadsky et al., (1987), (see also Tint et al., 1991) have proposed a model to explain the distribution of IFs when MTs or MFs are disrupted. These authors suggest that the continuous centripetal pull by the actin cortex (Bray and White, 1988) also  acts upon IFs but in an intact cytoskeleton, the centripetal pull is resisted by the anchorage of IFs to MTs. However, if MTs are depolymerized, anchorage of IPs to MTs is lost and the centripetal flow of the actin cortex pulls IFs towards the nucleus where IFs collapse into tight coils or bundles. Conversely, if the actin cortex and M F bundles are disrupted by cytochalasins (see Chapter 1, Section V.C.7.b.), the centripetal pull of the actin cortex is absent but the connection of IFs to MTs is maintained and therefore IFs codistribute with MTs which form numerous cell processes.  Similar to interaction between IFs and MTs, the interactions between IFs and MFs are also unidirectional. When IFs alone are disrupted (see Section B.5. below), the actin M F network is not affected (Gawlitta et al., 1981; Klymkowsky, 1981; Lin and Feramisco, 1981).  c. Plasma Membrane In fibroblasts, the ends of IFs often coincide with vinculin-containing adhesion plaques or focal contacts (see Chapter 1, Section V.D.) found between fibroblasts and their tissue culture substratum. Although not all ends of IFs have an association with focal contacts and vice versa, IFs appeal- to be closely associated with both dot (initial, immature contacts) and dash (mature) focal contacts (Bershadsky et al., 1987). It is possible that IFs (Bershadsky et al., 1987), along with MTs (Geiger et al., 1984b; Rinnerthaler et al., 1988), play a role in the maturation and stabilization of focal contacts. However, neither MTs nor IFs are required for the formation of focal contacts as they are observed in the presence of colchicine and hence the absence of M T and IF networks. Nevertheless, in these conditions, fibroblasts spread more slowly and inefficiently and cell morphology differs considerably from that of cells with an intact cytoskeleton (Bershadsky et al., 1987; also see Chapter 1, Section V.A.7.).  d. cytoplasmic components IFs may serve as a cytoskeletal scaffold around which other cellular and cytosolic components are  57 spatially organized. For example, IFs may play an ancillary role to MTs in controlling the distribution of mitochondria (Klymkowsky et al., 1989).  3.  Mechanical Functions of IFs  In addition to the non-mechanical functions of IFs described above, IFs appear to provide resilience against physical deformations by forming a mechanical scaffold. In vitro, polymerized vimentin proteins have unique viscoelastic properties suggesting that vimentin networks within cells maintain cell integrity. Compared to actin and MTs, vimentin networks are easily deformed at low stresses but they can withstand large stresses without losing elasticity. In fact, the more they are strained, the more resistant to deformation they become (Janmey et al., 1991). Under stresses that would rupture an actin network, the vimentin network would prevent excessive deformation. Thus, vimentin networks resist breakage and are sufficiently flexible without making the cytoplasm too rigid (Janmey et al., 1991).  If the cytoskeleton is organized into a tensegrity network (reviewed by Ingber, 1993; see also Chapter 1, Section V.A.6.a.ii.) then IPs may aid MTs to resist compression in cells. This is possible if MTs are stabilized against buckling by lateral interconnections with tensionally-stiffened IFs (Wang et al., 1993; Klymkowsky 1995).  4.  Agents that Affect IFs  As discussed above, the distribution of IFs is dependent upon MTs. Therefore, agents such as colchicine that disrupt MTs also disrupt IF networks which collapse into bundles or coils about the nucleus in many cell types in vitro.  5.  Effects of IF Collapse on Cell Behaviour  IFs are not required for the formation or maintenance of M T networks, intracellular transport, actin M F bundles or focal contacts (see Section V.B.2. above). As well, an intact IF system is not  58 required to maintain cell shape or cell motility (Gawlitta et al., 1981; Klymkowsky 1981; Lin and Feramisco, 1981) or mitotic activity (Gawlitta et al., 1981) in several cultured cell types.  One reason for the absence of effects following collapse of the IF network on cellular organization and behaviour is that cytoplasmic IFs appear to function at the level of the tissue or organism level rather than at the level of single cells (Klymkowsky et al., 1989). This is particulary evident in mutations of keratins which have been identified in a number of human diseases (reviewed by McLean and Lane, 1995).  59 C.  Actin  In animal cells, actin can be assembled into a variety of structures that perform a range of functions including participation in the determination of cell shape and providing the molecular machinery for cell motility and various contractile activities. As well, actin may participate in transmembrane signalling and intracellular transport.  In most eukaryotic cells, actin is the most abundant cytoplasmic protein and in nonmuscle cells, actin is distributed throughout the cytoplasm and is concentrated in the cell cortex. The cell cortex is a dense network of actin filaments and associated proteins just beneath the plasma membrane. It functions as a cortical cytoskeleton that maintains an isotropic tension that resists deformation and supports the pliable plasma membrane (Bray and White, 1988). However, the cell cortex is also a variable structure, its precise form differing not only amongst different cell types but differing in different regions of the same cell. It is the plasticity of the cell cortex that enables cells to generate extensions from their cell surfaces, phagocytose, change shape and to translocate.  This section will briefly review the structure, assembly and organization of actin and some of the associated proteins that control these processes and facilitate the generation of force for cell movements. It should also be noted that discussion will be limited to actin organization and function in fibroblasts.  1.  Structure  Actin microfilaments (MF) are 7nm-thick filaments that are also known as F-actin. MFs are polymers of globular or dumbbell-shaped actin subunits known as globular actin or G-actin. Each actin monomer can associate with either one molecule of ATP or A D P and each G-actin subunit has an intrinsic polarity (reviewed by Bremer and Aebi, 1992). Because the monomers are assembled head to tail, all the 'heads' of the subunits point in the same direction and consequently, the assembled filaments also have an intrinsic polarity (Alberts et al., 1989; Darnell et al., 1986).  60 A l l actin isoforms bind the SI (subfragment 1) or head fragments of Type II myosin (Darnell et al., 1986) (see Section V.C.3.c.iii.c. below). Each actin molecule in an actin filament can bind one myosin head in the same orientation to each actin subunit and therefore myosin binding reveals the structural polarity of the filament. In the absence of ATP, the myosin heads project laterally, at 45 degrees, from the filament and the appearance of decorated or myosin-bound actin in the electron microscope resembles a series of arrowheads. The point of the arrowhead corresponds to the 'pointed end' which is the minus end; the barb end of the arrowhead corresponds to the 'barbed end' which is the plus end (reviewed by Alberts et al., 1989; Darnell et al., 1986).  2.  Actin Assembly  Because actin filaments are polar, the kinetics of polymerization are different at the two ends. As a monomer subunit is incorporated into the polar filament, a conformational change occurs and it preferentially affects the plus end at which growth is 5-10 times faster than growth at the minus end. However, a steady state can be reached at which monomer addition at the plus end and monomer dissociation at the minus end occur at the same rate. Thus, the length of the polymer remains constant and subunits flux or treadmill through the polymer, constandy being transferred from one end to the other (reviewed by Alberts et al., 1989; Darnell et al., 1986; Pollard and Cooper, 1986; Pollard, 1990).  Motile cells must be able to maintain high rates of actin polymerization and depolymerization, simultaneously, at different sites in the cytoplasm and must carefully regulate the distribution and three-dimensional architecture of their actin structures. In vivo, the organization of actin is accomplished through complex interactions between actin and a vast array of actin-associated proteins (AAPs) or actin-binding proteins (ABPs).  There are at least 48 classes of ABPs comprising at least 100 members which can nucleate, sever, cap, depolymerize, stabilize and cross-link actin filaments (reviewed by Button et al., 1995). In  61 most animal cells, 50% of actin is unpolymerized (Alberts et al., 1989). However, very little exists as a free monomer as most G-actin is bound to actin monomer binding proteins. The other 50% of actin is polymerized into filamentous arrays that include single filaments, meshworks or isotropic gels, and bundles which arise from the action of another group of ABPs. Still other groups of ABPs link actin filaments together or to other cytoplasmic and membrane components, and others provide the mechanical work for motile behaviours. Moreover, the various groups of ABPs interact with each other, both competitively and cooperatively.  A discussion of the in vivo dynamics of actin will be presented largely in conjunction with the discussion of the various classes of ABPs (see Section V.C.3. below). Coordinated interactions between different ABPs and the effects of different cell signals will also become apparent during discussion of mechanisms of protrusion and locomotion (Chapter 1, Section VI).  3.  Actin-Binding Proteins  a. Classification of Actin-binding Proteins Classes of ABPs can be broadly distinguished on the basis of their effects on actin organization although some ABPs fall into more than one category (Vandekerckhove, 1990; Vandekerckhove and Vancompernolle, 1992). ABPs can also be categorized on the basis of their interaction with either G-actin or F-actin (Pollard and Cooper, 1986) and this criteria is used in the following discussion of ABPs. In turn, ABPs interacting with F-actin are subdivided on the basis of their effects upon actin organization (Hartwig and Kwiatkowski, 1991). 1. ABPs that interact with monomelic G-actin 2. ABPs that interact with actin filaments (F-actin) to (a) , sever filaments (b) . cap filaments (c) . connect actin filaments with other actin filaments, other proteins or other cellular components. This group of ABPs includes (i) caldesmon and tropomyosin and therefore this section will also  62 include a discussion of myosin which is responsible for the contractile processes in the cell (see Section V.C.c.iii.c. below) and (ii) proteins that link actin filaments to the actin cortex and to specialized regions of the plasma membrane involved in cell adhesion complexes (see Chapter 1, Section V.D.).  ABPs share certain characteristics that control their activity and affect their interaction with actin (reviewed by Rozycki et al., 1994). For example, particular actin sequences appear to be the preferred binding target of different ABPs (reviewed by Vandekerckhove and Vancompernolle, 1992) and therefore different groups of ABPs may compete for actin binding. Competition amongst different groups of ABPs could regulate actin response to different stimuli. For example, within the same cell, several different pools of actin monomer coexist. These pools can be individually mobilized because the ABPs controlling the pools may be less competitive with each other than with other groups of ABPs (reviewed by Vandekerckhove and Vancompernolle, 1992).  ABPs vary in the number of actin-binding sites that they possess and the activity of these sites can be controlled by cellular signals such as C a  2 +  concentrations and phosphoiriositides (reviewed by  Vandekerckhove 1990).  b. Actin monomer-binding Proteins Members of this group of ABPs form a complex with G-actin that does not polymerize as well as free or unbound G-actin. Thus, actin-monomer-binding ABPs sequester monomers in a pool or reservoir of nonpolymerizable complexes that can be released for polymerization upon receiving the appropriate signal. The monomer-binding ABPs also rebind monomers released by depolymerization and therefore monomer-binding ABPs 'buffer' free monomeric actin and can regulate actin dynamics (reviewed by Fechheimer and Zigmond, 1993). Monomer-binding ABPs include profilin, 8-thymosins, actin-depolymerization factor (ADF)/cofilin and actobindin (Fechheimer and Zigmond, 1993; Sun et al., 1995).  63 c. Actin Filament-binding Proteins (i) Actin filament-severing Proteins Some members of this family include gelsolin, severin, fragmin and villin. Gelsolin can be considered the prototype. Their primary mode of action is the Ca -regulated severing of actin 2+  filaments and subsequent capping of the actin fragments, thus preventing further assembly (eg. Janmey and Stossel, 1987; Stossel, 1990).  (ii) Actin filament-capping Proteins Capped filament ends are the result of a competition between G-actin and a capping protein for the same end of a filament. Capping can occur at either the plus (barbed) or minus (pointed) ends of a filament (Pollard and Cooper, 1986). Examples of barbed-end capping proteins include the fragmin/severin group and gelsolin/villin group. As well, cytochalasin (see Chapter 1, Section V . C.7.b.) may compete with these capping proteins for the plus end (Pollard and Cooper, 1986; Cooper, 1987). The characterization of pointed-end capping proteins is incomplete but examples may include acumentin, found in macrophages, fj-actinin, found in skeletal muscle, and the complex of spectrin and band 4.1 (Pollard and Cooper, 1986).  The simplest capping mechanism occurs when an actin monomer is modified so that it can still bind to F-actin, but its binding prevents the further addition of monomers. Capping also occurs when a non-actin protein binds to the filament end, preventing addition of monomers, or binds close to the end of a filament and introduces steric interferences or a conformational change in the terminal subunit, thereby preventing further growth (Vandekerckhove, 1989). Binding of a protein to the end of a filament can also interfere with dissociation of the terminal subunit or interfere with both addition and loss of subunits (Pollard and Cooper, 1986).  Thus, binding of capping proteins can stabilize filament length and stop treadmilling. In turn, filament length affects the viscosity of the cytosol which affects cytoplasmic flow, movement of  64 particles and possibly myosin-mediated contractions (Pollard and Cooper, 1986; Kolega et al., 1991).  Capping proteins may also be involved in the anchorage of actin filaments to the plasma membrane or other cellular components and this role of ABPs is discussed below.  (iii) Actin filament-connecting Proteins (a) between actin filaments Interconnecting ABPs cross-link or bundle actin filaments and thus require at least 2 actin-binding sites. In general, ABPs that crosslink filaments form actin gels and are elongated molecules such as filamin and spectrin. ABPs that bundle filaments are smaller, globular proteins such as aactinin and fimbrin. However, the effects of these interconnecting ABPs upon actin arrays also vary with the concentration of the A B P and actin (Otto, 1994).  (b) between actin filaments and myosin Caldesmon and tropomyosin bind to the sides of F-actin and cooperate to protect the filament from mechanical forces and other ABPs that could destabilize the filament. In non-muscle cells, caldesmon also regulates the interaction between actin filaments and myosin and stimulates the binding of tropomyosins to F-actin. Caldesmons are localized in membrane ruffles and are distributed periodically along stress fibers, complementary to the distribution of a-actinin (Vandekerckhove, 1990).  Tropomyosins form filaments that bind to actin and strengthen the actin filaments and as noted above, enhance the binding of myosin to actin. Tropomyosins compete with filamin and a-actinin whose binding in turn affects caldesmon and myosin binding. Overall, multiple interactions amongst ABPs are responsible for a complex variety of actin assemblies.  65 (c) Myosins Myosins are a family of mechanochemical proteins with actin-activated ATPase activities (Korn and Hammer i n , 1990). There may be up to 7 different classes of myosins, based the on similarity of their amino acid sequences, and it is likely that cells contain multiple types of myosins (Titus, 1993).  Myosins all contain a motor domain which is attached to a variety of different tail domains that are largely responsible for the varied functions of myosins (see below). That is, the different tails determine which particular cellular process the motor unit will drive (Spudich 1989; reviewed by Ruppel and Spudich, 1995). Myosin n, also known as conventional myosin, can be considered a dimer because it contains two globular heads (or SI fragments) and a tail region that spontaneously assembles into thick filaments characteristic of striated muscle. A l l other myosins are unconventional because they contain a single head and do not appear capable of assembling into filaments (reviewed by Alberts et al., 1989; Cheney and Mooseker, 1992; Titus, 1993).  In non-muscle cells, myosin II can form small bipolar assemblies that act on actin filaments from opposite directions so that two points in the cell are drawn closer together. Depending upon the anchorage sites of the actin filaments to which the myosin filaments are bound, a variety of movements can result and they may also involve the interaction of unconventional myosins (Spudich, 1989).  The tail regions of unconventional myosins can interact with membranes, either directly or through linker proteins as well as interact with actin. Myosin I is considered the prototype of unconventional myosins and the tail domain of this myosin can bind directly to membranes and to actin filaments, thereby linking membranes to the filaments and exerting force on them. Myosin I can also anchor its tails to F-actin, crosslinking the actin filaments and possibly facilitating filament sliding (Cheney and Mooseker, 1992). Thus, unconventional myosins have been implicated in  66 vesicle and organelle transport and endocytosis (see Chapter 1, Section V.A.5.a.; reviewed by Titus, 1993). As well, the redistribution of myosins during the development of cell polarity implicates them in the generation and maintenance of polarity as well as in cell migration (Spudich, 1989; Collins et al., 1991; see also Chapter 1, Section VI.). Myosin I is found in close association with the plasma membrane located at peripheral cell sites such as leading edges of locomoting cells. In particular, filopodia, lamellipodia and ruffles (see Chapter 1, Section VI.A.) contain the highest concentration of myosin I. In contrast, myosin II is concentrated at the rear of locomoting cells (Korn and Hammer HI, 1990; Titus, 1993; Conrad et al., 1993), (see Chapter 1, Section VLB.).  (d) between actin filaments, the actin cortex and plasma membrane (i) Direct Interactions The actin cytoskeleton can associate with the plasma membrane directly or indirectly. Direct interactions between the actin cytoskeleton and plasma membrane are mediated by integral membrane proteins and two examples are ponticulin, unique to Dictyostelium , and the transmembrane receptor for epidermal growth factor (EGF). Ponticulin can bind directly to the sides of actin filaments as well as nucleate the assembly of filaments.  Actin is also suspected as the cytoskeletal attachment protein to the EGF receptor. The binding of E G F to its receptor activates the receptor which in turn, initiates a cascade of events including ruffling and reorganization of the cytoskeleton prior to cell proliferation (reviewed by Hitt and Luna, 1994).  (ii) Indirect Interactions In addition to cytoplasmic actin filaments, actin also forms a network of filaments under the plasma membrane. This network is called the cell cortex and it is a dynamic structure that provides rigidity to the plasma membrane and also determines its shape. For example, cell surface structures such as microspikes, filopodia and membrane ruffles (see Chapter 1, Section VI.A.) all contain an actin  67 cytoskeleton. The actin cortex may also participate in phagocytosis, secretion, endocytosis, cell locomotion, cell-to-cell and cell-to-substratum interactions and transmembrane signalling (reviewed by Bray et al., 1986; Bretscher, 1991; Bretscher 1993). The different functions attributed to the cell cortex depend on a balance of cooperative and competitive interactions amongst ABPs, in particular those ABPs that form a link between cytoplasmic actin, the actin cortex and the plasma membrane. As well, different ABPs in different cell types confer different structure and organization to the cell cortex and therefore also to the associated mechanical properties of the cortex (reviewed by Bray et al., 1986). (a) spectrin- and ankyrin-like proteins Two prototypes of ABPs associated with the actin cortex are spectrin and ankyrin which are found in erythrocytes on the cytoplasmic side of the plasma membrane. Nevertheless, variations of these ABPs likely occur in the plasma membrane domains of all higher cells as their plasma membranes also require the stability and flexibility of actin meshwork (reviewed by Alberts et al., 1989).  In erythrocytes, spectrin molecules link together into two-dimensional tetramers by binding their ends to short actin filaments. This creates a deformable, nedike elastic mesh that ties adjacent to the plasma membrane and that can function as molecular struts or springs between the actin filaments. The actual link between the meshwork and the plasma membrane is provided by ankyrin and protein 4.1 that binds to both spectrin and the cytoplasmic domain of specific transmembrane proteins (reviewed by Hitt and Luna, 1994). (b) membrane attachment proteins In other cells such as fibroblasts, actin filaments often have their ends or sides very close to the plasma membrane. Historically, vinculin has been the prime candidate for linking F-actin directly to the plasma membrane. However, other evidence suggests that vinculin is but another intermediary A B P (eg. reviewed by Pollard and Cooper, 1986) between F-actin and one of the components of adhesion plaques which will be discussed in Section V . D . of this chapter.  68 4.  Actin Organization  a. Fibroblasts ABPs play a central role in cellular mechanisms which distinguish the two ends of an actin filament and control its orientation and location in the cytoplasm. Most actin filaments appear to be oriented with their minus or pointed ends oriented to the cell interior and it is possible that these ends are often left free as they are relatively inactive with respect to polymerization and depolymerization (eg. Alberts et al., 1989). In contrast, the plus or barbed ends are the predominant site of incorporation of actin subunits (Wang, 1985) and these ends are situated close to or at the plasma membrane (eg. Alberts et al., 1989); although, evidence does exist for distinct populations of actin filaments that differ in their polarity profiles as well as organization and stability (Lewis and Bridgman, 1992; see also Chapter 1, Section VI).  In cultured fibroblasts, cytoplasmic actin filaments are often formed into stress fibers which are collections of contractile bundles of actin filaments, myosin and tropomyosin. Stress fibers are located adjacent to the plasma membrane, next to and generally parallel to the cell's plane of attachment. At one end they insert into focal contacts (see Chapter 1, Section V.D.); at their other end they can insert into a perinuclear network of IF or into another focal contact (Buckley and Porter, 1967; Alberts et al., 1989). Stress fibers also give off small branches of filament bundles that course through the cytoplasm and merge with a general filamentous network. Stress fibers are aligned in directions of physical stress and are formed as a result of tension. They are also labile structures in that their dimensions and intracellular distribution can change (Buckley and Porter, 1967). That is, stress fibers are assembled near both the leading and trailing edges of motile cells in culture but their orientation can change according to the direction of migration. As well, the fibers can merge or fragment from either end of the fiber but elongation of stress fibers occurs only in a centripetal direction, towards the nucleus (Wang, 1984; Theriot and Mitchison, 1991).  69 b. Epithelium In some epithelial cells, actin is organized into polygonal filament networks that radiate from distinct foci and often interconnect with networks from adjacent foci (Lazarides, 1976; Gordon and Bushnell, 1979; Ferns et al., 1992). These foci are termed geodomes and are thought to act as nucleating sites and organization centers for newly forming actin bundles (Lazarides, 1976; Gordon and Bushnell, 1979). During cell spreading, the meshwork filaments are converted into actin bundles but the organization of actin around geodomes is not lost (Middleton and Sharp, 1982).  5.  Interactions of Actin MFs with Other Intracellular Components  The interaction of actin MFs with MTs (see Chapter 1, Section V.A.5.C.), M A P s (see Chapter 1, Section V.A.5.c.ii) and IFs (see Chapter 1, Section V.B.2.b.) has already been discussed along with the possible role of the actin system in intracellular transport (see Chapter 1, Section V.A.5.a.).  Actin filaments, along with MTs and IFs, play an important role in the sorting and regional localization of mRNA. In particular, the movement to and maintenance or concentration of actin mRNA behind the lamellipodia of motile cells (eg. Singer et al., 1989; Hoock et al., 1991) requires intact actin filaments (Sundell and Singer, 1991; reviewed by Singer, 1992). Coincidentally and conveniently, this localization of actin mRNA corresponds to same site behind the lamellipodia at which G-actin is concentrated (Cao et a l , 1993), actin subunits recycle (Wang, 1985; Theriot and Mitchison, 1991) and precursors of cortical actin filaments form (Giuliano and Taylor, 1990; Heath and Holifield, 1991), (see also Chapter 1, Section VI).  6.  Interactions of Actin MFs with the Physical Environment  In vitro, actin filaments form rigid networks that are very resistant to stress and are more resistant to deformation than MTs or IFs (see also Chapter 1, Sections V.A.6.a and V.B.3.). These  70 characteristics make actin a suitable component of the cell cortex because this is where environmental forces upon the cell are likely to be greatest. As well, forces required for cell locomotion are generated there, in conjunction with transmembrane substratum-attachment sites (see Chapter 1, Section V.D.). At high strains, actin networks rupture and become fluid. This behaviour is important in cell motility in which protrusions deform rapidly and cytoplasmic flow is required (Janmey et al., 1991).  Like MTs, the in vitro polymerization of actin in unsupported artificial membranes can produce sufficient force to distort the spherical membranes into deformed (Cortese et al., 1989) predominantly disc shapes (Miyata and Hotani, 1992). Furthermore, the actin filaments polymerizing within liposomes align along the periphery of the liposomes and assemble into thick bundles, possibly in response to shearing forces (Cortese and Frieden, 1988) or their restriction within small spaces (Miyata and Hotani, 1992). In fact, the turnover of actin networks under in vivo shearing forces has been considered as a factor in the dynamics of stress fibers (Cortese and Frieden, 1988), which form in response to tension (Kolega, 1986; Alberts et al., 1989) and may be visible manifestations of tension field lines (Greenspan and Folkman, 1977).  The stress-induced alignment of actin filaments can be observed in vivo. For example, the tails of polarized fibroblasts contain bundles of actin MFs that extend parallel to the long axis of the cell and are thus aligned with the stress field; if the tails are detached from the substratum, the aligned M F bundles lose their orientation and collapse into a random mesh-like array (Chen, 1981). In fish epidermal epithelium held under tension by micromanipulation or by the locomotive activity of other cells, actin MFs are aligned parallel to tension and they do not reorient against such force (Kolega, 1986). As well, MFs in plant cells align along the lines of principal stress (Lintilhac, 1974). In addition to biological evidence for stress-induced alignment of M F networks, a mathematical model for this phenomenon has been developed (Sherratt and Lewis, 1993) and confirms that MFs are very sensitive to changes in stress and will spontaneously align in a single  71 direction in response to the principal components of stress. Mathematical models have also been developed to describe the effects on M F assembly against a force (Hill, 1981; Hill and Kirschner, 1982a, b) and they are similar to those already described for MTs (see Chapter 1, Section V.A.6.a.L).  The tensegrity model (see Chapter 1, Section V.A.6.a.ii.) of cytoskeletal organization can also be used to interpret behaviours of actin MFs. However, actin M F bundles are physically associated with focal contacts which provide a link to the substratum or extracellular matrix (ECM), and this relationship is an integral part of the tensegrity model (reviewed by Ingber, 1993). Therefore, the role of actin in the tensegrity model will be discussed together will the role of focal contacts in the tensegrity model, in Section V.D.4. of this chapter.  7.  Agents that Disrupt Actin Structure and Dynamics  Many details of actin dynamics and assembly as well as the role of the actin system in phagocytosis and cell locomotion have been derived by the use of agents that perturb these activities. These agents include phallotoxins, which stabilize MFs, and latrunculins, tolytoxin, clostridial toxins and cytochalasins which cause disruption of the actin system. However, only phallotoxin and cytochalasin will be reviewed because they were the only actin ligands used in studies comprising this thesis.  a. Phallotoxins Phallotoxins are poisonous alkaloids produced by the mushroom Amanita phalloides. Phallotoxins do not bind G-actin but bind actin subunits within actin filaments and stabilize the filaments from depolymerization. They accomplish this by binding more tightly to F-actin than to G-actin and by reducing the dissociation rates at both ends of the filament to almost zero (Cooper, 1987). Thus, in phallotoxin-treated cells, the pool of F-actin increases dramatically whereas the pool of G-actin decreases dramatically. However, the rate of synthesis of G-actin and amount of actin mRNA  72 increase significantly, suggesting that actin is autoregulated, similar to tubuhn (Serpinskaya et al., 1990).  One disadvantage to the use of phallotoxins in living cells is that they do not permeate cell membranes of most cells and must be microinjected. As well, the effects of phallotoxins are not reversible and phallotoxins are often cytotoxic although that effect may not be direcdy related to its effects on actin as cells exposed to cytochalasins (see below) do not die (Cooper, 1987). Live cells loaded with phallotoxin demonstrate delayed spreading and compromised motility. Their actin staining shows replacement of parallel actin cables by disordered actin aggregates (Serpinskaya et al., 1990) but phallotoxins can stabilize F-actin against depolymerization by cytochalasins and elevated temperatures (Product Monograph MP354, Molecular Probes, Inc, Eugene, Oregon).  b. Cytochalasins (1) Mechanisms of Action Cytochalasins are fungal metabolites that have variable but reversible effects on actin and cell . behaviours. The effects of cytochalasin vary with the cell type (Holtzer and Sanger, 1972) and the mechanisms of cytochalasin action result from both high-affinity and low-affinity interactions with actin (Pollard 1990). The effects of cytochalasin also vary with the concentration (see also Chapter 3, Section I.B.2.b.) and length of exposure to cytochalasin (Yahara et al., 1982). In addition, different cytochalasins possess different affinities for actin as well as non-actin directed effects (reviewed by Yahara et al., 1982 and Cooper, 1987). For example, cytochalasins A and B (CB) inhibit monosaccharide transport by competing with hexose for high-affinity sites on the plasma membrane but this effect is independent from its effects on actin (Yahara et al., 1982). Cytochalasin D (CD) is thought to be specific for actin, although it can inhibit protein synthesis (Ornelles et al., 1986).  In vitro studies indicate that high affinity interactions of cytochalasins with actin occur with actin  73 filaments. Cytochalasins bind to the barbed end, one cytochalasin per filament, and prevent further polymerization as well as depolymerization at that end and CD is about 10 times more effective than C B . It is also possible that cytochalasins can bind to a subunit within a filament and sever or break the filament, with subsequent capping of the fragments (Cooper, 1987).  In vitro studies suggest that low affinity interactions of cytochalasins with actin occur with actin monomers but cytochalasin is a poor nucleator. Cytochalasin can form dimers with G-actin but hydrolysis of the bound ATP to A D P causes the dimers to dissociate and the cycle to repeat. As well, because cytochalasin should cap the barbed end of the monomer, growth could only proceed from the pointed end (reviewed by Cooper, 1987).  In fibroblasts, cytochalasin does not depolymerize actin filaments. In fact, the ratio of F-actin:Gactin is not changed. Instead, cytochalasin appears to disrupt the supramolecular organization of actin filaments because filaments persist albeit in dense focal accumulations (Schliwa, 1982). It is also possible that cytochalasins compete with other capping proteins for the barbed end. If capping proteins (see Chapter 1, Section V.C.3.c.ii.) are also involved in the three-dimensional organization of the filaments, then displacement of the capping proteins would disrupt the network, possibly causing breakage of the filaments which are then contracted into dense aggregates (Schliwa, 1982). Furthermore, stress fibers are observed to contract in the presence of cytochalasin, behaviour that would be expected with disruption of cross-links, breaking of filaments and weakening of the actin gel structure (Kolega et al., 1991). As stress fibers disappear, they appear to lose their membrane attachments and in fact, focal contacts are not observed in the presence of cytochalasin (Bliokh et al., 1980; Domnina et al., 1982).  (ii) Correlation between Cell and Cytoskeletal Behaviours in the Presence of Cytochalasin Low doses of cytochalasin as well as short times of exposure to cytochalasin result in a decrease or cessation of ruffling behaviour and cell migration (see Chapter 1, Section VI). However, gross  74 cell morphology is not affected. With increased length of exposure as well as exposure to higher concentrations of cytochalasin, all motility ceases, the cell body rounds up and numerous processes or branches are extended, giving the cells a stellate or arborized shape (eg. Sanger and Holtzer, 1972; Bliokh et al., 1980; Domnina et al., 1982; Schliwa, 1982; Yahara et al., 1982).  This cell behaviour corresponds to the formation of actin condensations or foci, first in the cell periphery and then, along with dissolution of stress fibers, throughout the cell body (eg. Schliwa, 1982). As well, the cortical layer of MFs disappears completely (Domnina et al., 1982). The arborized processes contain MTs and IFs (Bliokh et al., 1980) which conform closely to the morphology of the processes. MTs in cytochalasin-treated cells are straight (Heath and Holifield, 1992) and traverse, uninterrupted, to the distal ends of the arborized processes (Weber et al., 1976) whereas MTs in non-cytochalasin-treated cells bend along the plasma membrane to conform to cell contours (Weber et al., 1976; see also Chapter 1, Section V.A.5.c.i).  The development of arborized cell processes can be explained by the cytochalasin-induced disruption of the cell cortex. That is, the spatial arrangement of cytoskeletal elements such as MTs within cells is constrained by the cell cortex which imparts tensile strength to the plasma membrane (Petersen et al, 1982). The actin network which comprises the cortex is more resistant to deformation than MTs (Janmey et al., 1991) and therefore MTs, unable to extend to their full length, curve along the cell contours (Weber et al., 1976). When the cortical actin network is destroyed by cytochalasins, the restraining influence of the cortex is reduced and MTs are able to push out the pliable plasma membrane and form straight protrusions (Edson et al.; 1993). Under cytochalasin conditions, MTs also become laterally aggregated, into bundles, and this may be the result of M A P s or other proteins that are released from the actin networks (see Chapter 1, Section V.A.5.c.ii.).  75 D.  Cell Attachments to the Substratum  As discussed earlier (Chapter 1, Section V.C.), actin is localized in the cytoplasm and in the actin cortex which interacts with the plasma membrane. Most commonly, the actin cytoskeleton associates with the plasma membrane indirectly through a special group of proteins. In fibroblasts, bundles of actin filaments or stress fibers (see Chapter 1, Section V . C A a . ) interact with the plasma membrane at specialized sites which facilitate not only mechanical anchorage to the substratum but also facilitate signal transduction across the plasma membrane (eg. Werb et al., 1989; reviewed by Horwitz and Thiery, 1994; Schaller and Parsons, 1994).  The attachment of the ventral surfaces of cells to their substratum has been examined in several ways and has produced a diverse nomenclature such that attachment sites can be described on the basis of their appearance in light and electron microscopy and on the basis of their purported macromolecular structure.  1.  Descriptions of Cell Attachment Sites Based on Appearance in Light and  Electron  Microscopy  Transmission electron microscopy revealed that electron-dense "plaques", associated with actin filaments, were much closer to the substratum than other areas (Abercrombie et al., 1971). Interference reflection microscopy (IRM) revealed three types of regions depending upon the distance between the substrate and the plasma membrane (Izzard and Lochner, 1976):  a. Extracellular Matrix Contacts Extracellular matrix contacts (ECM contacts) appear light grey or white in IRM indicating distances of 100-140 nm separation from the substratum. They are typically located under the center of the cell (Izzard and Lochner, 1976). Electron microscopy revealed that E C M contacts are composed of colinear transmembrane associations of actin filaments and strands of extracellular matrix (see also Section V.D.3 below). E C M contacts also contain a fibronexus which is a very close  76 transmembrane association of actin filaments, vinculin (see Section V.D.2 below) and fibronectin (FN; see also Section II.A.2. below) fibers, localized about 15 nm from the substrate in stationary cells (Singer 1979,1982; see also Section V.D.3. below).  b. Close Contacts Close contacts appear as broad grey areas in IRM and are separated 30 nm from the substratum. They are diffusely distributed under the peripheral regions of the leading lamellae in spreading cells and focal contacts may be distributed within close contacts under the leading lamellae. Close contacts can also extend under the centre of the cell to varying degrees but typically, this area contains greater cell-substrate distances (eg. E C M contacts).  c. Focal Contacts: Focal contacts appear black in IRM and are separated 10-15 nm from the substratum. They are 210 |im long and vary in width between 0.25-0.5 (im (Izzard and Lochner, 1976) and 0.1-2 |im (eg. Opas, 1987). Focal contacts are located under peripheral sites of leading lamellae and near edges of non-spreading regions of the cell margins in moving and stationary cells (reviewed by Harris, personal communication).  The combined use of electron microscopy and IRM confirmed that adhesion plaques and focal contacts described the same structures (Abercrombie and Dunn, 1975; Heath and Dunn, 1978). Thus, the term "focal adhesion" (Burridge et al., 1988) was suggested because it conferred a specific function (adhesion) to a discrete (focal) region of the plasma membrane. Focal contacts have also been characterized as small, transient structures as compared to focal adhesions which are larger and more permanent structures (Opas, 1987) and although both structures appear black in I R M , the differences in size and stability may be related to the maturation of focal contacts (see Section V.D.3. below).  77 The functional significance of different types of cell contacts is unclear although focal adhesions appear to be the sites of strongest attachment between the cell and substratum (Burridge et a l , 1988). As well, different cell morphologies and different cell types are commonly associated with a particular attachment type. For example, focal contacts are not characteristic of highly motile cells (eg. reviewed by Burridge et al., 1988, Couchman and Rees, 1979) and in polarized, fibroblasts, focal and close contacts are closely associated at the base of advancing margins whereas only focal contacts are observed along trailing margins or retraction fibers (Harris, personal communication; see Chapter 1, Section VI.A.).  2.  Descriptions of Cell Attachment Sites Based on Macromolecular Structure  Close contacts have not been well-characterized on a molecular basis (eg. Izzard and Izzard, 1987) and therefore the following discussion is limited to focal contacts and focal adhesions. Focal contacts and focal adhesions have both been associated with bundles of actin filaments or stress fibers and while the exact nature of the interactions between actin filaments and the plasma membrane is unclear, it is known that interactions are mediated by a variety of cytoplasmic proteins and specific transmembrane adhesion receptors called integrins.  a. Integrins Integrins are a diverse family of heterodimers comprised of a and 8 subunits. Different combinations of subunits produce integrins with binding affinities for different extracellular ligands (reviewed by Hynes, 1992; Haas and Plow, 1994). One of the receptors for fibronectin (see also Chapter 1, Section II.A.2.) is 0:581 integrin. The 6 subunit appears to bind to cytoplasmic proteins whereas the a subunit serves a regulatory function related to the binding of the ligand (reviewed by Hynes, 1992). Binding of fibronectin to 0581 causes the integrins to be relocated from a diffuse distribution in the plasma membrane to a localized accumulation within focal adhesions in which the integrins become restricted in their mobility. As well, ligand binding appears to cause a conformational change in 61 so that interactions with cytoplasmic proteins and actin are possible .  Conversely, if the integrin and its ligand dissociate, the focal adhesion is lost as actin filaments and cytoskeletal proteins dissociate. Cytoplasmic events can also influence integrin affinity for its ligand, suggesting that not only can the structure of focal adhesions be controlled from both sides of the plasma membrane but integrin functions are modulated as well (reviewed by Hynes, 1992; Luna and Hitt, 1992). In fact, modulation of integrin activity forms the basis of signalling via integrins and it takes two forms: regulation of the affinity and conformation of the integrin from inside the cell (inside-out signals) and triggering of intracellular events by ligand binding to the integrin (outside-in signals), (reviewed by Hynes, 1992; Horwitz and Thiery, 1994).  b. Cytoplasmic Proteins Associated with Focal Adhesions Stress fibers are linked to integrins comprising the adhesion plaque by a complex assembly of proteins. For example, the cytoplasmic domain of Gi can bind to talin (reviewed by Beckerle and Yeh, 1990) and a-actinin. In turn, talin can bind to vinculin (reviewed by Otto, 1990) although talin can also nucleate actin assembly and self-associate. Vinculin can bind paxillin and a-actinin and to other vinculin molecules, a-actinin binds to and crosslinks actin and binds to vinculin, zyxin and the G-integrin. Tensin can bind vinculin and may bind and cap the plus ends of actin filaments. Radixin, ezrin, moesin as well as other proteins (reviewed by Luna and Hitt, 1992; Bretscher, 1993; Arpin et al., 1994; Hitt and Luna, 1994) and kinases (reviewed by Schaller and Parsons, 1994) have been localized to focal adhesions. The exact assembly or configuration of actin filaments and associated proteins within focal adhesions is unclear as evidenced by the diverse collection of schematic representations (eg. Opas, 1987; Burridge et al., 1988; Alberts et al., 1989; Uitto and Larjava, 1991; Luna and Hitt, 1992; Schaller and Parsons, 1994).  The functional significance of this complex chain of protein interactions at focal adhesions is also unclear but may be related to the observation that some of the protein interactions are of low affinity. Clustering and interactions amongst the protein components may stabilize their association with the adhesion site (reviewed by Haas and Plow, 1994; Hitt and Luna, 1994).  79 The temporal association of stress fibers and the various protein components within adhesion plaques also varies (see also Section V.D.3. below). For example, vinculin has been reported to be a late addition during adhesion formation (Woods and Couchman, 1992) and adhesion plaques can be maintained in the absence of vinculin (Herman and Pledger, 1985). However, the presence of vinculin at adhesion plaques can also precede the presence of stress fibers (eg. Herman and Pledger, 1985; Bershadsky et al., 1987; Feltkamp et al., 1991) as the association of vinculin with focal contacts is largely actin-independent (Avnur et al., 1983). Furthermore, it is also possible that integrins can aggregate and associate with actin independently, in the absence of both talin and vinculin (Samuelsson et al., 1993). One explanation for these disparate observations may be found in a discussion of the formation and maturation of adhesion sites.  3.  Formation and Maturation of Adhesion Sites  In vitro, three stages of adhesion have been described by Woods and Couchman (1988) and Bereiter-Hahn (1990) and although the respective stages are not identical, they share some features (see also Segel et al., 1983 and review by Brunette, 1986c).  Stage 1 of adhesion (Bereiter-Hahn, 1990) may be comparable to the attachment phase (Woods and Couchman, 1988). Primary adhesion to a solid substratum in Stage 1 does not require metabolic energy although blebbing of the round cells is evident. Neither stress fibers nor orientated fibrillar patterns are formed but close contacts or broad,smooth contact areas are evident (Bereiter-Hahn, 1990).  Stage 2 (spreading phase) and stage 3 (organization of the cytoskeleton phase) include further flattening and structural organization and both stages are dependent upon metabolic energy. In Stage 2 the peripheral zone of actin fibrils become organized and the basal cytoplasm becomes reinforced with stress fibers (Bereiter-Hahn, 1990). During this stage, the cells undergo radial spreading (eg. Abercrombie, 1980) and a regular periodic array of radial focal contacts are formed  80 at the cell periphery (Segel et al., 1983). In Stage 3, cells enter the phase of polarization; actin filaments and MTs become organized and lamellae are extended. Once the cell acquires a polarized shape and starts to move, the periodicity of the peripheral focal contacts is lost (Segel et al., 1983).  Some areas of the plasma membrane may be predestined to develop into focal contacts and subsequently into focal adhesions. For example, focal adhesions usually form within or just behind the ruffling lamellae of motile cells (eg. Izzard and Lochner, 1980) and in fact, ruffling may be a specialized activity concerned with the accumulation and deposition of components necessary for development of an adhesion site (Rinnerthaler et al., 1988). The formation of focal contacts may be preceded by motile structural precursors which consist of linear F-actin fibers oriented radially like ribs within a motile lamelhpodium (Izzard and Lochner, 1980; Izzard, 1988; see also Feltkamp et al, 1991). The precursor separates into two portions and the proximal or basal part becomes stationary. Focal contacts form beneath and parallel to the F-actin fibers in the proximal precursor, which persist at the new focal contact as the initial adhesion plaque. Subsequently, talin and vinculin accumulate at the new adhesion plaque and stress fibers develop centripetally through the leading lamella (Izzard, 1988). The distal portion of the precursor continues to advance with the leading edge (Izzard and Lochner, 1980; Izzard, 1988) and talin, but not vinculin, accumulates at the distal tips of the precursors and along nodes across the lamella suggesting that talin may be involved in development of the precursor and the maturation of the focal contact (Izzard, 1988).  Contact-forming units comprised of ruffling and microspike activity and contact proteins may also form short-lived primordial or immature contacts (eg. Geiger et al., 1984a; Rinnerthaler et al., 1988) that appear as small dot-like accumulations of vinculin and are not associated with actin bundles (eg. Bershadsky et al., 1987). Some of these focal adhesions disassemble and others enlarge and promote nucleation of associated stress fibers (reviewed by Opas 1987; Burridge et al., 1988). In fact, the association of MTs (Geiger et al., 1984b; Rinnerthaler et al., 1988) and IFs (Bershadsky et al., 1987) at these sites may increase stability and longevity of the adhesion site.  81 After vinculin is recruited into the initial focal contact area, F-actin accumulates to form M F bundles and these adhesion sites are usually stable for long periods of time (reviewed by Opas, 1987; Burridge et al., 1988).  Thus, the formation of substratum contacts proceeds through phases. That is, a fast, initial phase in which small immature contacts are formed at large distances of separation from the substratum, followed by a maturation phase in which contacts enlarge at reduced distances of separation (reviewed by Opas, 1987). Within mature focal adhesions, there is a slow exchange of actin, vinculin and a-actinin with the cytoplasmic pool of these proteins, although focal adhesions remain competent to nucleate stress fibers (reviewed by Burridge et al., 1988). For example, fluorescently-labelled actin microinjected into cells becomes concentrated at newly formed focal adhesions (reviewed by Burridge et al., 1988) and stress fibers appear to elongate from focal adhesions both in the front and rear of the cell (Wang, 1984). Most actin filaments terminate at focal contacts but bundles of MFs can also attach laterally at focal contacts so that their barbed ends are free to elongate without first detaching (Samuelsson et al., 1993).  Mature contacts demonstrate elongated, dash-like accumulations of vinculin that are associated with the ends of actin filament bundles (Bershadsky et al., 1987). Mature contacts may also demonstrate an "eye of a needle" distribution pattern of vinculin and the integrin receptor for fibronectin in which the vinculin fills the "eye" and the receptor outlines the "needle" (Damsky et al., 1985; reviewed by Opas, 1987). However, some investigators have described a more uniform distribution of the fibronectin receptor within the adhesion site and still others have reported that fibronectin receptors are redistributed over time (reviewed by Burridge et al., 1988).  In comparing data from various studies, the composition of the substratum must be considered or more correctly, for in vitro observations, the composition of the culture media must be considered. For example, serum contains both fibronectin (FN) and vitronectin (VN). V N is also known as  82 serum spreading factor and accounts for the majority of adhesion-promoting activity in serum (Hayman et al., 1985); therefore, cells spreading in the presence of serum, are actually spreading on a carpet of V N adsorbed onto the substratum (eg. Weiss, 1975). Some cells possess distinct receptors for F N (FNR; eg. 0:561) and V N (VNR; eg. 0^,65), (Hynes, 1992) and their localization within adhesions sites can differ. For example, in the presence of serum, both receptors can occur in the same focal contact. After prolonged culture, only the V N R is retained within focal adhesions because the FNR is diverted from focal contacts to sites of attachment on extracellular matrix cables (Singer et a l , 1988), the E C M contact (see Section V.D. l.a. above) which is composed of colinear transmembrane associations of actin filaments, FNR and FN-containing extracellular matrix fibers closely apposed to the substratum (Chen et a l , 1985). [ECM contacts also contain a fibronexus which is a very close transmembrane association of actin filaments, vinculin and F N fibers, localized about 15 nm from the substrate in stationary cells (Singer 1979, 1982). The formation of focal contacts precedes the appearance of both E C M contacts and fibronexuses (Singer, 1982)].  Similarly, the localization of F N and V N within focal adhesions can also differ and the presence of F N within adhesion sites may be another criterion for assessing the maturity of an adhesion site. However, the data is equivocal because the presence or absence of serum (or low serum) can affect the ability of cells to clear F N from focal adhesions over time. For example, in low serum, F N is retained and therefore detected within focal adhesions. In the presence of higher levels of serum, cells gradually remove F N from focal adhesions (Grinnell, 1986) but V N is retained and is therefore the major extracellular component of focal adhesions in most culture conditions (Burridge, et al., 1988).  The association of F N within focal contacts may be part of the maturation process of focal adhesions developing in low serum conditions. The youngest focal contacts which form at the distal cell (Izzard and Lochner, 1980) margins contain neither vinculin (Vnc) nor F N (Vnc-/ FN-).  83 Intermediate-stage focal contacts are located more centrally and are Vnc+/FN-. Fully mature contacts which are perinuclear, contain stress fibers and appear as black streaks in IRM, are Vnc+/FN+ (Singer, 1982). The presence of F N within adhesion sites con-elates with the accumulation of FNR and with appearance of the fibronexus. As contacts mature, the formerly smooth membrane surface of the focal contact becomes fibrillar; membrane-associated vesicles align at the borders of the focal contact suggesting focal contacts may provide a conduit for deposition of F N into the contact. The association of F N within adhesions may also represent functionally distinct contact sites. FN-negative focal contacts may be relatively unstable and characteristic of motile cells; FN-positive focal contacts may be characteristic of stationary cells (Singer, 1989).  4.  Functions of Focal Adhesions  Initially, focal adhesions were considered important in cell migration (Abercrombie et al., 1971) but in fact, they are most prominent in stationary cells and are absent in highly motile cells (reviewed by Burridge et al., 1988; Woods and Couchman, 1988). In addition to organizing stress fibers, providing transmembrane signalling links and possible conduits for F N deposition, focal adhesions are also sites of the strongest cell attachment to the substratum (Burridge et a l , 1988). The anchorage provided by focal adhesions permits the development of tension which promotes the formation of stress fibers which in turn, align parallel to the tension (Kolega, 1986). The development of tension may also lead to maturation of the focal contacts which in turn, reinforces M F bundling. As steady tension is developed between a few of the strongest sites of attachment, a spread cell becomes irregularly stretched between these strongest sites and the circumferential or radial array of focal contacts is lost (Segel et al., 1983; reviewed by Opas, 1987, 1994). Thus, the distribution of stress fibers partly reflects the lines of tension generated by the cell (Greenspan and Folkman, 1977).  The cytoarchitecture of cells is thought to reflect the mechanical equilibrium between the cell and its  84 substratum and this interaction is mediated by focal adhesions. For example, forces generated by the cell can remodel the extracellular matrix through purely mechanical activity and if the substratum is elastic, it can be deformed (reviewed by Opas, 1994, Harris, personal communication). Likewise, mechanical forces applied to integrins, cause a reorganization of the cytoskeleton (Wang et al., 1993).  In the tensegrity model (reviewed by Ingber, 1993) of cytoskeletal organization, integrins act as mechanical receptors or mechanochemical transducers (reviewed by Ingber, 1991) that transmit forces to MFs within focal adhesions (Wang et al., 1993). For example, mechanical deformation of the substratum can produce rapid and large increases in filamentous actin prior to changes in cell shape (Pender and McCulloch, 1991) and mechanical deformation or twisting of attachment sites causes cells to increase in stiffness and to increase their resistance to deformation (Wang et al., 1993).  Although actin MFs contribute the most to cell stiffness, MTs and IFs resist compression as a paired unit (Wang et al., 1993) and provide struts to balance the mechanical force of the contractile actin networks. As well, MTs and the substratum or extracellular matrix (ECM) share complementary compressive load-bearing functions that are mediated through focal adhesions. In fact, when MTs are depolymerized, increased loads are borne by the substratum (Dennerll et al., 1988; Danowski, 1989; Kolodney and Wysolmerski, 1992) making possible the attachment and spreading of cells in the absence of MTs.  In vitro , fibroblast behaviour which is interpreted as "locomotion" may in fact be a manifestation of the cells' attempts to mechanically reorganize the extracellular matrix or compress the substratum (Harris, 1994; personal communication). Because the artificial substrata are rigid, fibroblast attempts to contract or close wounds produce directional protrusions combined with strong pulling forces in the directions opposite to the protrusion, ie "locomotion". Locomotion is associated with  polarization or development of cell margins which advance in contrast to margins that trail behind or sag inwards and net locomotion simply reflects the imbalances between the different types of margins (Harris, personal communication). The following section will briefly correlate the morphologic events of polarization and locomotion with cytoskeletal behaviours.  VI.  Polarization and Locomotion of Tissue Culture Cells  Cells in suspension are under cortical tension and therefore adopt a spherical shape (eg. Bray and White, 1988). When the cells setde onto a rigid substratum and attach (see Chapter 1, Section V.D.3), they spread radially along their entire periphery and adopt a 'fried egg' shape. During the phase of polarization (eg. Abercrombie, 1980) only some regions continue to advance outwards whereas others retract. As a result, some cells develop only one prominent advancing edge or lamellipodia (see Section VI.A. below) and such cells can be described as monopolar (Glasgow and Daniele, 1994; reviewed by Alberts et al., 1989; Harris, 1994; personal communication). Typically, two or more advancing edges compete, stretching the cell into stellate shapes and thus, fibroblasts are best described as irregularly multipolar (Harris, personal communication).  The polarization of fibroblasts can be described as the development of directional protrusions combined with the exertion of strong pulling forces in the direction opposite to the protrusion (Harris, personal communication). Net locomotion occurs as a result of imbalances between advancing edges, such that the dominant or strongest margin becomes the leading edge of the cell and the rest of the cell appears to be pulled along behind (Harris, personal communication).  Net locomotion involves several processes that must be coordinated and repeated, either continuously or periodically. That is, cell locomotion involves interdependent processes that include the generation of force for displacement of the cell from one site to another, the development of biased protrusions in the direction of movement, the retraction of posterior processes and the formation and breakage of adhesive interactions (see Chapter 1, Section V.D.) between the cell and the substratum. In order for protrusion and locomotion to occur, not all parts of the cell can be pulled centripetally to the same degree and only some parts of the cell can protrude. Specifically, the cortical tension of limited regions of the cell surface must be selectively relaxed. The exact cytoskeletal and molecular mechanisms whereby protrusions develop, cells become polarized and net locomotion occurs are still unresolved but this section will briefly  87 introduce some models of these behaviours.  A.  Terminology  Harris (1994, personal communication) has reviewed and clarified the terminology used to describe structures characteristic of polarized and locomoting fibroblast-like cells.  Retraction fibers are long, flexible trailing extensions that represent the rear end of cells. Retraction fibers and retreating margins are firmly attached to the substratum at their distal ends via focal contacts, in the absence of close contacts. However, the appearance of close contacts at trailing margins may signal reactivation of the margin into an advancing margin (see below), (Harris, personal communication).  Cell protrusions can assume a variety of morphologies. Leading lamella or lamellipodia are broad, flattened, thin (about 0.1-lp.m) but stiff forward extensions or protrusions. Filopodia or microspikes are short, thin, rod-like and rigid protrusions that extend from or amongst lamellipodia. Both lamellipodia and filopodia extend out ahead and fold back in characteristic ruffling motions. Neither lamellipodia nor filopodia are attached to substratum at their distal ends; instead, focal contacts are located behind the leading edge of these protrusions, at the base of the lamellipodia. Typically, focal contacts are surrounded by close contacts and the disappearance of close contacts may forecast the transformation of an advancing margin into a trailing margin. (Harris, personal communication).  Ruffles, particles, antibodies or other materials attached to the outer cell surface move centripetally across the cell's dorsal surface in a process termed retrograde surface transport. Surface protrusions may also occur as small rounded blebs that can occur anywhere on the cell surface but are typically concentrated along advancing margins, intermingled with lamellipodia which they accompany rearwards.  88 B.  Role of the Cytoskeleton in Polarization, Formation of Protrusions and  Directed Motility Investigations into the role of the cytoskeleton in polarization, formation of protrusions and directed motility have not been limited to fibroblasts. In fact, amoeboid locomotion (reviewed by Harris, 1994) and leukocyte chemotaxis (eg. reviewed by Zigmond, 1989) have been favoured by many investigators and form the basis for several models, some of which may not be entirely relevant to events in fibroblasts, although, interestingly, locomotion of the slime mould Dictyostelium may be the most analogous to that of tissue culture cells (reviewed by Harris, 1994). A brief review of possible cellular mechanisms involved in polarization, formation of protrusions and directed migration is presented below. For the sake of clarity, the role of MTs and the actin system are discussed separately but should not imply that the two systems operate independently.  1.  Microtubules  The dependency of different cell types upon MTs for maintenance of polarity and motility varies (see also Chapter 1, Section V.A.7.). For example, PMNs (Keller and Niggli, 1993), macrophages (Glasgow and Daniele, 1994) and E cells (eg. Middleton et al., 1988) can maintain their polarity and motility after disruption of MTs (eg. by colchicine). In contrast, in eosinophils, if MTs are disrupted (Fay et al., 1989) or the MTOC is destroyed (eg. by laser microirradiation, Koonce et al., 1984) cell polarity is lost and migration becomes erratic and eventually ceases. In fibroblasts (previously discussed in Chapter 1, Section V.A.7.), disruption of MTs produces effects similar to those observed in eosinophils.  In eosinophils, the free ends of MTs become closely apposed to the cell edge before the changes in surface activity suggest that lamellipodia (pseudopod in leukocytes) formation will occur at that site. In some way, M T entry into the cell cortex triggers a reaction that proceeds for some time and ultimately results in lamellipod/pseudopod formation. As well, the density of MTs in regions that  89 have recently formed a pseudopod is routinely greater than in other regions of the cell (Fay et al., 1989).  In fibroblasts (Kupfer et al., 1982; Singer and Kupfer, 1986) and endothelial cells (eg. Gotlieb et al., 1981), wounding experiments have demonstrated that the M T O C and Golgi apparatus become positioned anterior to the nucleus, between the leading margin and the nucleus within 30 minutes to several hours (depending upon the cell type), before the onset of migration into the wound (see also review by Harris, personal communication). As well, stable (detyrosinated MTs, see Chapter 1, Section V.A.3.d.) MTs become oriented towards the wound, in parallel to the reorientation of the M T O C , prior to the onset of migration (Gundersen and Bulinski, 1988). The correlation between the reorientation of the M T O C to the front of a cell and the onset of migration suggests that orientation of the M T O C determines the direction of cell movement (eg. Gotlieb et a l , 1981). However, it also appears that the initial phase of biased lamellipod formation and directional movement can occur in the absence of a preferred MTOC position (Schutze et a l , 1991; Euteneuer and Schliwa, 1992). In fact, anterior location of the M T O C is attained gradually during the first 4 hours of directional migration (Schutze et al., 1991). Furthermore, for fibroblasts in collagen gels or on grooved surfaces, the M T O C remains in essentially random positions with respect to the nucleus although the cells themselves may become polarized (Schutze et al. 1991).  MTs may also be involved in polarization by directing intracellular transport (see Chapter 1, Section V.A.5.a.). Specifically, kinesin may be a mediator of polarization as antibodies specific to kinesin inhibit polarity very much in the same way as colchicine (Rodionov et al., 1993). Thus, polarity may not depend directly upon intact MTs but rather upon the transport of specific materials to certain intracellular sites. Nevertheless, directional intracellular transport and endocytosis may be involved in maintaining localized protrusions and directed locomotion. For example, in order to reconcile retrograde surface transport, membrane would have to be constantly reassembled or recycled along the leading margin. Although some models propose that tension generated by the  90 actin system is the driving force behind membrane flow, others suggest that the flow of membrane lipids through the cell is responsible (reviewed by Heath and Holifield, 1991; Harris, 1994). Such a directed intracellular flow could be mediated by an asymmetry in the endocytic cycle (eg. Singer and Kupfer, 1986; Bretscher, 1984,1987) and could also aid extension of the leading edge of migrating cells (Alberts et al., 1989). Oriented MTs and actin filaments in a polarized cell could not only orient directional motors but reinforce protrusions and cell locomotion along their long axes (reviewed by Alberts et al., 1989).  MTs and/or their associated proteins (MAPs; see Chapter 1, Section V.A.4.) may also be involved in controlling the locations and degree of actin polymerization and the distribution of myosin (Bornens et al., 1989; reviewed by Wang, 1991). These conclusions are reached by observing the effects upon the actin system when MTs are depolymerized (see also Chapter 1, Section V.A.5.C.). For example, in lymphocytes, MTs appear to control the recruitment of cytoplasmic myosin to the cortex, thereby 'toning-down' or interfering with the dynamic activity of the acto-myosin system which is forced to adopt a unidirectional activity (Bornens et al., 1989). Moreover, the presence of MTs may not only restrict the polymerization of actin and formation of protrusions to localized sites but MTs also may be necessary for development of maximal activity at those sites (Bershadsky et al., 1991). When MTs are depolymerized, polymerization of cytoplasmic actin becomes generalized as evidenced by ruffling around entire cell peripheries and, although actin contractility increases (eg. Danowski, 1989), polarity and motility are lost and the average rates of protrusion and retraction actually decrease, in comparison to control cells (Bershadsky et a l , 1991).  2.  Actin  In fibroblasts, the precise signals that determine the future site of a protrusion are not known but it is becoming evident that local biochemistry not only governs cytoskeletal behaviour but produces and regulates the forces for motility (Cooper and Mitchison, 1995). Several mechanisms  responsible for generating the force for protrusive activity have been proposed and include osmotic pressure, actin polymerization, myosin motors and combinations thereof (reviewed by Small, 1989; Condeelis, 1993; Cramer et al., 1994; Giulano and Taylor, 1995).  Models in which protrusive forces are derived from osmotic pressure appear to relate best to systems in which actin filaments are arranged as networks or gels rather than as filament bundles (Condeelis, 1993). Localized reactions at the plasma membrane and depolymerization of the actin network can increase the amount of particulate matter in the cytoplasm and trigger the influx of fluid from other parts of the cell into the area until the concentration of particles is equalized. The influx of fluid causes distension of the membrane which is subsequently stabilized by actin reassembly (Oster, 1988; 1989; reviewed by Stossel, 1990). The role of osmotic pressure may also augment the protrusive forces produced by actin polymerization in sperm acrosomal processes (Inoue and Tilney, 1982; Tilney and Inoue, 1982; reviewed by Condeelis, 1993) although actin polymerization alone can also deform artificial membranes (see Chapter 1, Section V.C.6.; Cortese et al., 1989; Miyata and Hotani, 1992).  Most commonly, protrusion models couple actin polymerization to myosin contraction and the simplest model is based on fish epidermal keratocytes. In these cells, shape and size appears to result from the same process that drives locomotion (Lee et al., 1993). That is, the protrusion of lamellipodia is always perpendicular to the cell edge and results from actin polymerization pushing the cell membrane forward. Extension at the front of the cell is balanced by myosin contraction in the rear of the cell which causes a graded retraction perpendicular to the rear cell margin (reviewed by Cramer et a l , 1994).  Some protrusion and migration models based on actin polymerization and myosin activity (reviewed by Small, 1989; Condeelis, 1993) are more complex. Within initial protrusions (see Chapter 1, Section V.A.5.c.i.), actin is diffuse, myosin II is absent but myosin I (see Chapter 1,  92 Section V.C.3.c.iii.c.) is found in either a punctate distribution or in a linear array, parallel to the direction of migration. Myosin I is also localized in filopodia, leading lamellae and in established protrusions (Conrad et al., 1993). This localization of myosin I appears to be suitable for its involvement in the extension and/or retraction of protrusions and in the transport of vesicles (Conrad et al., 1993). In fact, if the barbed ends of all filaments were adjacent to the cell membrane, sliding mediated by myosin I between parallel actin filaments could cause protrusion in the complete absence of actin polymerization (see also Cramer and Mitchison, 1993, below and review by Condeelis, 1993).  At the base of protrusions, just behind lamellipodia, foci of unpolymerized actin (Cao et al., 1993) are localized and actin subunits recycle (Wang, 1985, Theriot and Mitchison, 1991). As well, precursors of cortical actin filaments form (Giuliano and Taylor, 1990) and transverse fibers containing actin and myosin TJ (Conrad et a l , 1993) assemble and are transported rearwards to the perinuclear region. Most likely, rearward actin-filament flux is driven by myosin II in the cell cortex or myosin I in the lamellipodium although forces driving rearward flux are probably independent from those driving protrusions (reviewed by Cramer et al., 1994; see below). Interestingly, ruffles contain actin, a-actinin, caldesmon, talin and possibly myosin I (reviewed by Bretscher, 1991).  During contraction of actin fibers, myosin II is concentrated in the centre and tail regions of the cell (Conrad et al., 1993) where it may aid detachment of the cell's rear end allowing the cell to continue forward (Jay et al., 1995), help to maintain a polarized cell shape, maintain the direction of motility, maximize the rate of locomotion and/or aid in the delivery of cytoskeletal or contractile elements to the leading edge (Conrad et al., 1993). Furthermore, concentrations of C a  2 +  and  calmodulin are highest in tail regions which favours activation of myosin II and depolymerization of actin filaments by calcium-dependent severing and depolymerizing ABPs (reviewed by Giuliano and Taylor, 1995).  93 Thus, actin filaments undergo a continuous assembly along lamellipodia and filopodia, flow centripetally, are disassembled into actin monomers and the treadmilling cycle is repeated. Actin recycling appears to involve both cortical actin and stress fibers but it is unclear whether stress fibers are formed independendy or whether part of the cortical actin becomes reorganized into stress fibers and associated with myosin as it moves centripetally (reviewed by Harris, 1994). It is also unclear to what degree myosin I and/or myosin II contribute to the centripetal movement of treadmilhng actin. That is, can treadmilling of cortical actin alone provide traction forces for cell migration and to what extent is actin/myosin contraction of stress fibers necessary (Harris, 1994)? It appears possible that cortical actin treadmilling can provide traction and this process is made manifest by retrograde surface transport which is really a continuous retrograde flow of cortical actin rather than plasma membrane (Harris, 1994). In this model, a 'conveyer belt' of treadmilling cortical actin is mechanically coupled, at the front of the cell, to special integral membrane proteins called 'tine proteins' (perhaps integrins?). As these proteins are pulled, like the tines of a rake through the plasma membrane in rearward direction, they drag along particles large enough to become trapped between them (Harris, 1994). Thus, the treadmilhng actin cortex exerts traction through the ventral plasma membrane and retrograde surface transport is a side effect (Harris, 1994).  Protrusions and possibly motility in some cells may be mediated largely by myosin. In postmitotic spreading epithelial (kidney PtK2) cells, two populations of actin filaments exist (Cramer and Mitchison, 1993). One population remains stationary, with respect to the substratum and is localized to retraction fibrils. The other population of filaments moves forward with and at the same rate as the spreading cell margin. Furthermore, the actin filaments in some spreading edges not only move forwards but the fdaments can separate so that some filaments either move forward and some remain stationary or, some move forward and the others move rearwards . Most likely, the stationary filaments within retraction fibers and in spreading edges are fully immobilized within adhesion sites to the substratum. Filaments that move rearwards within the spreading margins may  94 not attach to the substrate continuously within the advancing edge. Filaments that move forward with the cell margins could represent more dorsally-located filaments that are not anchored at the ventral surface. The force for forward movement of the cell margin and the filaments may be generated by a barbed-end directed myosin motor such as myosin II pulling the moving filaments over the stationary ones (Cramer and Mitchison, 1993). Overall, populations of forward-moving and stationary filaments may be analogous to lamellipodia and retraction fibers in any motile system in that they provide an immobile base of filamentous actin and the rest of the cell moves over or through it by means of myosin motors and movement of other actin filaments (Cramer and Mitchison, 1993; reviewed by Cramer et al., 1994).  The function and role of some cytoskeletal and associated proteins in cell motility have been established by manipulating cytoskeletal genes in situ (eg. review by Titus et al., 1990). For example, it was long postulated that myosin U played a major role in generating cellular movements. However, mutant strains of the slime mould Dictyostelium deficient in myosin II are still capable of locomotion although at a reduced speed. Furthermore, elimination of actinassociating proteins such as oc-actinin or myosin I does not appear to have any significant effects on movements of Dictyostelium (reviewed by Titus et al., 1990).  C.  Factors that can Affect Protrusions and Locomotion  In contrast to isolated fibroblast-type cells (eg. see below; review by Harris, 1994), isolated E cells often display variable and generally limited abilities to move (Brown and Middleton, 1985) and this may be because E cells differ from fibroblasts in their social behaviours (Trinkaus, 1984). That is, the spreading of E cells is enhanced upon contact with other cells (Middleton, 1976,1977) as cells form stable lateral adhesions which unite E cells into a cohesive sheet that moves as a unit. In vitro, E cells migrating as a sheet develop focal contacts primarily at the leading front or periphery of the sheet (Brunette, 1986c), which is also the site of marked ruffling activity (DiPasquale, 1975). These observations suggest that the marginal cells provide the motile power for migration  95 of the sheet and that submarginal cells are sparsely adherent to the substratum and are passively stretched and tangentially oriented by the pull of the marginal cells (eg. DiPasquale, 1975; Trinkaus, 1984; Brunette, 1986c).  Other cell types such as neural crest cells also migrate poorly, if at all, when isolated, but can be stimulated to travel upon contact with migrating cells. This phenomenon is called contactstimulated migration and appears to activate and sustain the migration of the mass of cells in contact with one another (Thomas and Yamada, 1992).  Fibroblasts are minimally responsive to contact-stimulated migration (Thomas and Yamada, 1992) and in fact, fibroblasts display contact paralysis and contact inhibition of locomotion upon contact with another cell (reviewed by Harris, personal communication). When a locomoting fibroblast crawls into contact with another cell, ruffling and blebbing along the margins where the contact occurs stops and this behaviour is called contact paralysis. As well, forward locomotion decreases or temporarily ceases and this behaviour is called contact inhibition (Abercrombie and Heaysman 1953,1954; reviewed by Harris, personal communication). Depending upon the angle of contact between cells (eg. head-on or obliquely) the cells typically resume their locomotion, after variable time intervals, but in a different direction. Contact inhibition can occur between fibroblasts, between fibroblasts and epithelial cells (Brunette et al., 1989; Damji, 1992) and even between epithelial cells although the latter collisions result in the formation of intercell adhesions (Middleton, 1982; see above).  One explanation offered for this fibroblast behaviour is that locomotion is slowed and redirected because the polarity of movement has changed. That is, advancing margins are converted to retreating margins (Harris, personal communication). The altered polarity of movement may also be explained by the loss of stable detyrosinated MTs (see Chapter 1, Section V.A.3.d.) from the original leading edge after contact with adjacent cells (Nagasaki et al., 1992). When contact is  confined to a portion of the cell margin, stable MTs are rapidly cleared from the area behind the contact site yet remain oriented towards adjacent noncontacting and ruffling margins. Thus, the state of cell-cell contact is a contributing factor in establishing the distribution of stable MTs in fibroblasts and the contact-induced loss of stable MTs may be a marker for contact inhibition of locomotion (Nagasaki et al., 1992). Interestingly, fibroblasts deficient in MTs due to the action of colchicine can also display contact inhibition although it is described as altered or delayed because contact between adjacent cells persists for longer times than in cells with an intact cytoskeleton (Vasiliev et al., 1970).  D.  Summary  This review of the cytoskeleton and its role in determining cell shape and directed migration illustrates that a myriad of complex and still largely-unresolved processes and regulatory mechanisms are involved. Likely, an array of mechanisms rather than a single mechanism is operational or perhaps, different mechanisms dominate in different cell populations.  The available knowledge regarding polarization and cell migration has been obtained primarily from observations of cells on flat or smooth surfaces. To what extent these processes are modified when cells encounter anisotropic substrata is not known and the molecular details of those events are beyond the scope of this thesis. However, the overt responses of tubulin filaments, actin M F bundles and vinculin-staining plaques in relation to changes in the morphology of cells on anisotropic substrata form the basis of the investigations that comprise this thesis.  97 CHAPTER 2 STATEMENT OF T H E PROBLEM Contact or topographic guidance refers to the reactions of cells with the topography of their substratum and includes alterations in their shape, orientation and polarity of movement (Curtis and Clark, 1990; see Chapter 1, Section II.B.2.)- A variety of mechanisms for this phenomenon have been proposed (see Chapter 1, Section IV.) but current hypotheses focus on the cytoskeleton and suggest that cells align in a manner which may minimize distortions to their cytoskeletons. Actin microfilament (MF) bundles (Dunn and Heath, 1976; Dunn and Brown, 1986) and focal contacts (Ohara and Buck, 1979; Dunn, 1991) have received the most attention, and microtubules (MT) little, despite observations that orientation of MTs precedes a functional polarization of the cells that results in directed cell movement (reviewed in Chapter 1, Section VI.B.l). If one criticism can be directed towards previous studies of cells and their cytoskeletons on anisotropic surfaces, it is that most investigators have examined the cytoskeletons of cells that were already aligned with their substratum topography and therefore it was unknown how the cytoskeleton came to be accommodated to the underlying topography.  At the outset of this thesis, I wished to address this shortcoming in the observations upon which the present theories of contact guidance are based. I also wished to establish the relative role of the various cytoskeletal components in contact guidance as well as to test the current hypotheses of contact guidance (see Chapter 1, Sections B. and C ) . I hypothesized that there may be a primary or principal cytoskeletal determinant of cell orientation on micromachined grooves and predicted that the most important or principal cytoskeletal component involved in contact guidance could be identified by the following criteria:  1. The principal cytoskeletal determinant of contact guidance would be the first component to respond to the substratum topography. 2. The principal cytoskeletal determinant of contact guidance would be most sensitive to  topographic characteristics such as size and arrangement. 3. The principal cytoskeletal determinant of contact guidance would be able to exert its influence independendy from other cytoskeletal components. 4. If the principal cytoskeletal determinant of contact guidance was also the sole cytoskeletal component effecting contact guidance, then contact guidance would not occur in its absence. 5. The principal cytoskeletal determinant of contact guidance would also operate in other cell types.  Using these criteria, I designed the experiments that comprise this thesis to (1) identify a possible candidate(s) for the principal cytoskeletal component or determinant of contact guidance and (2) if that component could be identified, to direcdy test the role of that component in contact guidance.  To determine which cytoskeletal element was the first to become aligned with the substratum, I examined cytoskeletal alignment in relation to overall cell alignment in fibroblasts spreading on parallel sets of R grooves.  Secondly, I used micromachined substrata with sets of intersecting R grooves. These substrata enabled me to determine how cells and their cytoskeletal components adapted to conflicting topographies. Moreover, these experiments allowed me to compare cytoskeletal events within leading lamellae of polarized cells with cytoskeletal events in cells spreading at the groove intersections and thus test the hypothesis of Brunette (1988a; see Chapter 1, Section IV.C.) who suggested that leading lamellae contain the machinery responsible for contact guidance.  Thirdly, I evaluated the role of the principal cytoskeletal element(s) in cell orientation by following cell spreading, alignment and orientation in the presence of specific inhibitors of MTs (colcemid) and actin MFs (cytochalasin B and D).  99 Fourthly, I tested the role of MTs and MFs as principal determinants of topographic guidance by examining their behaviours on precisely-defined surfaces with topographic features that were similar in shape but differed in size (R, N and V N grooves; see Chapter 3, Section I.D.2.).  Finally, to test the generality of the findings observed in fibroblasts, I examined cytoskeletal and cell behaviours in porcine epithelial cells.  The experiments described in this thesis differed from earlier investigations of contact guidance because they benefited from the use of several protocols that were not previously available. As noted above, micromachining produced silicon substrata with precisely-controlled geometries and dimensions. As well, titanium-coating of the silicon surfaces ensured chemically-uniform substrata so that cell behaviours could be reasonably interpreted as responses to topographic cues alone. Secondly, the use of confocal microscopy (reviewed by Shotton, 1989) enabled the distribution of cytoskeletal elements to be accurately determined in three dimensions. Although it was not possible to track individual living cells and their cytoskeletons over time, the time course of the distribution of cytoskeletal elements was followed in a large population of cells and considered at the population level, consistent and clear-cut patterns of cytoskeletal development occurred over time.  CHAPTER 3 METHODS and MATERIALS  The methods and materials used in this thesis are presented in the following order:  I. Fibroblasts A . Cell Culture B . Culture Conditions C. Observations of Living Cells D. Micromachined Substrata E. Preparation of Samples for Examination by Epifluorescence Microscopy F. Epifluorescence Microscopy G. Statistics  II. Epithelium A . Cell Culture B. Epifluorescence Microscopy C. Statistics  I. Fibroblasts A.  Cell Culture  1.  Starting Cultures from Gingival Explants  Fibroblasts were isolated from human gingival explants as described by Brunette et al., (1976). Healthy gingival tissue obtained from young (<35 years), systemically-healthy patients was collected in a sterile culture tube containing 10 ml sterile medium (see Section I.A.2. below). As soon as possible after excision of the tissue, the following procedures were performed in a biohazard hood using sterile techniques.  a. The tissue was rinsed in fresh medium and placed in a 60 mm dish. The epithelium was removed from the connective tissue by excising an approximately-1.5 mm-thick section of tissue from the oral side of the specimen. Using 2 sterile scalpels, the remaining connective tissue was minced.  b. The minced tissue was collected into a tube of medium and centrifuged (5 minutes at 1500 rpm). The supernatant was discarded, the pellet resuspended and centrifuged.  c. Meanwhile, sterile 60 mm tissue culture dishes were prepared. Using toothpicks, 2 small dabs of sterile silicone grease were placed about 22 mm apart, on the bottom of the dishes.  d. Most of the supernatant from the recentrifuged samples was discarded and the pellet was mixed in the remainder of the medium in the tube. Using a sterile Paster pipette, a few explants were placed in each 60 mm dish, between the spots of silicone grease. The medium and any bubbles were carefully removed with the Pasteur pipette. A sterile glass coverslip was placed over the explants and pressed down onto the silicone-spots. 4.5 ml medium was added per dish and the dishes were placed into an incubator at 37° C in humidified 5% CO2.  102 e. After 1 week, each dish was examined for cell growth and feeding of the cultures began when growth was evident. When cells appeared to be well established on the coverslip and/or dish, the coverslip was removed and placed upside-down into another 60 mm dish in 4.5 ml medium. Both dishes were maintained and fed. Once growth was confluent, or nearly so, selective trypsinization began.  This culture technique results in the growth of both E cells and fibroblasts but fibroblasts will appear first, followed by sheets of epithelium. The two cell types can be separated by two techniques. One method is based upon the differential resistance of fibroblasts and E cells to detachment by trypsin. Fibroblasts are more easily detached from the growth surface by trypsin, requiring 5-10 minutes to round up and detach. E cells require longer (15-20 minutes) as well as extensive rinsing and repeated aspiration by a pipette (Brunette et al., 1976). Thus, selective trypsinization can be used to produce cultures that are predominandy one cell type (either fibroblasts or E cells) from cultures that originally contained both cell types.  The second technique used to separate fibroblasts from E cells is based on their differential preference for different serums in the growth medium (see Section I.A.2. below). Human gingival fibroblasts thrive in Calf Supreme (Gibco, Grand Island, New York) which is a supplemented bovine serum that favours human fibroblasts but not human E cells.  f. Trypsin solution [0.25% trypsin (Gibco, Grand Island, New York) and 0.1% glucose dissolved in citrate saline (pH 7.8)] was thawed. After removal of the medium in the 60 mm dishes, 2 ml trypsin were briefly added to each dish, rinsed and discarded. Another 2 ml trypsin were added for 5-10 minutes and incubated or until cells rounded up. The trypsin-and-cell suspension was added to tubes containing 4 ml medium, 2 dishes/tube, centrifuged and the pellet resuspended in medium and transferred to tissue culture flasks, 2 tubes/flask, containing a total of 20 ml medium per flask and incubated. These cells were predominantly fibroblasts. Cultures were  103 subcultured and divided twice more and then the surplus was frozen.  2.  Established Fibroblast Cultures  Fibroblasts between the 4* and 12 subculture were cultured in Alpha Minimal Essential Medium th  (MEM) (Terry Fox Labs, Vancouver, B.C.) supplemented with antibiotics (penicillin G (Sigma, St. Louis, Missouri) 100 u.g/ml, gentamicin (Sigma) 50 (ig/ml, amphotericin B (Fungizone, Gibco) 3 }ig/ml) and 15% bovine serum (Calf Supreme, Gibco) at 37° C in a humidified atmosphere with 5% CO2. Cells were removed from the growth surface using a trypsin solution, centrifuged and resuspended in medium. To determine the density (cells/ml) of the suspended cell solution, 0.5 ml of cell suspension was added to a cuvette containing 9.5 ml isoton (Isoton II, Coulter Electronics of Canada Limited, Surrey, British Columbia). The closest three counts out of 4 counts by a Coulter Counter (Coulter Electronics, Inc., Hialeah, Florida) were used. If counts exceeded 10,000, the values were corrected using the Coulter Counter Coincidence Correction Chart. Multiplying the Coulter count by 40 produced the density of cells/ml (40 was used because the counted cell suspension was a 1:20 dilution and the Coulter counts 0.5 ml).  Different cell densities were used in different experiments because experiments differed both in the length of time that the cells were cultured and in culture conditions. Because we wished to examine isolated cells without contact with adjacent cells, it was necessary to obtain cells which were spatially well-isolated at the different time periods. In the experiments oudined in this thesis, the following culture conditions were used:  B.  Culture Conditions  1.  Control Conditions  For control conditions, population densities of l x l O cells/ml were used for experiments lasting up 5  to 2 hours; l x l O cells/ml for those lasting 6 hours, and l x l O cells/ml for experiments lasting 24 4  3  and 72 hours. For controls used in experiments in conjunction with colcemid, cytochalasin or  104 cytochalasin/colcemid conditions, control medium included dimethyl sulfoxide (DMSO, Fisher Scientific, Ottawa, Canada), (<0.1%) (see also Section I.B.2.b. below). This concentration of D M S O had no observable effect on cell morphology or cell behaviour, similar to observations of other investigators adding D M S O to control media (eg. Yahara et a l , 1982).  2.  Drug Conditions - Experiments using Cytoskeleton-Perturbing Agents  a. Colcemid Conditions A stock solution of 100 u.g/ml colcemid (Sigma), solubilized in D M S O was prepared in medium so that the final concentration of DMSO was less than 0.1%. A n appropriate volume of stock solution, stored at 4° C, was added to the medium used for resuspension of the trypsinized cells so that a final concentration of 1.0 |ig/ml colcemid (Hollenbeck et al., 1989) was obtained. For colcemid conditions, cell densities of l x l O cells/ml were used for experiments lasting 2 hours and 5  l x l O cells/ml were used for experiments lasting longer than 2 hours. 4  b. Cytochalasin Conditions As noted in Chapter 1, Section V.C.7.b., the effects of cytochalasin can vary with the concentration of cytochalasin used. One source of disparate results amongst different studies is that some studies used culture media to which cytochalasin was added directly (eg. Bliokh et al., 1980; Domnina et al., 1982) and other studies (eg. Sanger and Holtzer, 1972; Weber et al., 1976) added a low concentration of a solubilizing agent. Solutions of cytochalasins prepared in aqueous media form fine dispersions because cytochalasins are not soluble in water. Therefore, solutions of cytochalasins are more accurately prepared if cytochalasin is solubilized in dimethyl sulfoxide (DMSO) or dimethylformamide (DMF), (Technical Information Bulletin No. AL-126, Aldrich Chemical Company, Inc., Milwaukee, Wisconsin, USA). At low concentrations (<0.1 M-g/ml), D M S O has no apparent effect on cells or their cytoskeletons (eg. Yahara et a l , 1982).  Stock solutions of 1.0 mg/ml cytochalasin B (CB) (Sigma) or cytochalasin D (CD) (Sigma),  105 solubilized in DMSO, were prepared in medium so that the final concentration of D M S O was less than 0.1%. A n appropriate volume of CB or C D stock solution, stored at 4° C , was added to the medium used for resuspension of the trypsinized cells so that either final concentrations of C B ranging from 0.3 fig/ml to 200 (ig/ml or final concentrations of C D ranging from 0.02 |ig/ml to 20 Ltg/ml were obtained. Lower concentrations of CD were used as this agent is about 10 times more effective than C B (Cooper, 1987). CB was used for time-lapse observations and for cytoskeletal investigations in order to compare results to earlier reports of fibroblasts in the presence of C B (Weber et al., 1976; Bliokh et al., 1980; Domnina et al., 1982). C D was used as an additional control to verify that the cytoskeletal patterns elicited in the presence of either cytochalasin B or D were due to their action on the actin system and were not related to the interference with sugar transport associated with C B (Yahara et al., 1982; Cooper, 1987).  For C B or C D conditions, cell densities of 1x10 cells/ml were used for experiments lasting 2 s  hours and l x l O cells/ml were used for experiments lasting longer than 2 hours. 4  c. Cytochalasin and Colcemid Conditions Appropriate volumes of the colcemid stock solution were added to the CB- or C D - containing medium used for resuspension of the cells so that a final concentration of 1.0 fig/ml colcemid was obtained. C B / C was used for time-lapse observations and for cytoskeletal investigations in order to compare results to earlier reports of fibroblasts in the presence of CB/C (Bliokh et al.,1980; Danowski, 1989). As noted above, CD/C was used as a control to verify that similar cytoskeletal responses occurred in the presence of either CD/C or CB/C.  For CB/C or CD/C conditions, cell densities of l x l O cells/ml were used for experiments lasting 2 5  hours and l x l O cells/ml were used for experiments lasting longer than 2 hours. 4  C.  Observations of Living Cells  Time-lapse observations were used to observe cell spreading and migration because these behaviours are too slow to be observed in real time. Specialized chambers and microscopic equipment (see Section I.C.3, Section I.C.5 and Section I.C.6 below) are utilized in time-lapse observations and the first task is to transfer the cells from their culture dishes onto the desired substrata and into the chambers.  1.  Seeding the Substrata  Cells were trypsinized and seeded onto smalltitanium-coatedsilicon wafers, ranging in average size from 1.0-1.5 cm by 1.0 cm, that contained the smooth and/or grooved substrata (see Section I.D. below). The cells were seeded onto the substrata either as cell suspensions or as droplets of cell suspension. Cultures used in time-lapse observations were buffered in 10 m M Hepes (Sigma). a. Cell Suspensions Cell suspensions of l x l O cells/ml were used for all time-lapse observations of fibroblasts, 5  regardless of culture conditions or length of observation. The cell suspensions were injected into a Pentz chamber (Bachofer, Reutlingen, Germany) in which the substrata had been mounted.  b. Droplets of Cell Suspension This method was used for observation of directed fibroblast migration along the grooves leading into the groove intersection (see Chapter 4, Section H ) . 10 | i l droplets of a cell suspension of l x l O cells/ml were plated onto the grooves leading into the intersection. The substratum was then 6  placed into a sterile culture dish in an incubator for 1 hour before dishes were flooded with medium. Within 6-12 hours, the substrata were mounted onto glass slides (see Section I.C.3 below) and the slide inserted into a Pentz chamber.  107 2.  Effects of gravity  It should be noted that the effects of gravity upon cell attachment, spreading and migration were negligible, as expected theoretically (Albrecht-Buehler, 1990). Identical behaviours occurred when cell droplets were seeded onto the substratum wafers that were suspended from pedestals (ie in upside-down position) or when wafers, seeded with a cell suspension, were inverted and suspended from pedestals (turned upside-down) 20 minutes after plating, or when wafers were seeded and maintained in right-side-up position (ie. not upside-down). Therefore, time-lapse observations could be made using either an upright or inverted microscope.  3.  Assembly of Pentz Chambers  In order to obtain the correct focal plane between the wafer in the Pentz chamber and the 8x objective, it was necessary to raise the height of the wafer within the Pentz chambers. This was accomplished by mounting the substrata onto pedestals which were mounted on glass slides in the Pentz chamber.  a. Fabrication of pedestals: i. Dl-1.2 mm gaskets (Bachofer) were lubricated with silicone grease and placed onto lightlylubricated (silicone) glass slides which were placed into petri dishes. To ensure a tight seal between the gasket and slide, the dish/slide/gasket assembly was placed into a vacuum for 1 minute.  ii. A n epoxy (Epotek 302-3, Epoxy Technology, Billerca, Massachusetts, USA) mixture of 10:4.5 weight-ratio of resins A:B was prepared and placed into a vacuum for 5 minutes.  iii. The epoxy resin was drawn into a 10 cc syringe and the epoxy was injected into each gasket, almost to the top.  108 iv. Epoxy solution was also drawn into a lcc syringe which was used to top off the resin levels in each gasket, level to the top of gasket with a flat or very slightly positive meniscus. The epoxy was initially polymerized for either 1 hour at 60° C or for 24 hours at room temperature. The gaskets could be removed at this stage. Final polymerization of the epoxy pedestals was ensured by incubation at 60° C for 3 days.  v. The pedestals were gently removed from the glass slides and the excess silicone grease wiped off. The pedestals were cleaned by ultrasonication in a detergent formulated for tissue culture (IX, ICN Biomedicals, Inc., Costa Mesa California), copious rinsing and ultrasonication in deionized distilled water. After air-drying, pedestals were autoclaved.  b. Mounting of pedestals and wafers Using indelible ink, clean glass slides were marked with an " X " about 1 inch from one end of the slide and the slides were autoclaved along with the Pentz chambers, Bachofer gaskets and Bachofer cover slips. Sterile silicone grease was prepared by filling a 10 cc syringe (needle/cap end removed) with silicone grease, reattaching the needle/cap and autoclaving the assembly.  Using sterile technique, the glass slide was placed into a petri dish. A dab of sterile silicone grease was applied onto the " X " marked on the glass slide and a pedestal was placed onto the grease dab over the " X " on the slide. A second small dab of grease was applied to the centre of the pedestal and the wafer [either freshly-glow discharged - (see Section I.D.3.b below) or containing attached cells (see Section I.C.I.b above)] was placed onto the pedestal.  c. Final Assembly of the Pentz chamber for use in upright microscope: In this assembly, the bottom of the glass slide rests against the bottom of the chamber and the Bachofer coverslip rests between the gasket and lid of the chamber.  109 The lid of the Pentz chamber was raised so that the lid faced towards the operator. The slide/pedestal/wafer assembly was inserted into the base of the Pentz chamber, with the " X " labelled end of the slide inserted first. Next, the Bachofer gasket was placed onto the slide, around the pedestal, with the wide side of the gasket facing towards the slide. Finally, the Bachofer slide was inserted under the sliding lock of the chamber and onto the top of the gasket. The lid of the Pentz chamber and the sliding lock were then closed.  d. Final Assembly of the Pentz chamber for use in inverted microscope: In this assembly, the bottom of the glass slide rests against the lid of the chamber and the Bachofer coverslip rests between the gasket and bottom of the chamber.  The lid of the Pentz chamber was opened so that the lid faced away from the operator and the bottom of the chamber faced the operator. With the wafer facing towards the operator and the bottom of the glass slide facing the lid of the chamber, the " X " -labelled end of the slide/pedestal/substratum assembly was inserted into the Pentz chamber. The Bachofer gasket was inserted onto the glass slide, between the slide and bottom of the chamber, and around the pedestal. The wide side of the gasket faced the glass slide. The Bachofer coverslip was carefully inserted between the gasket and the bottom of the chamber. The lid of the chamber was closed against the bottom of the glass slide and the sliding lock closed.  4.  Injection of cell suspension or medium  The plungers of two 10 cc syringes were loosened. One syringe (loading syringe) was used to take up 8-10 ml of either cell suspension (see Section I.C.I.a above) or medium (droplet technique, Section L C . l . b above). Holding the assembled and locked Pentz chamber sideways, with the wafer facing towards the operator, the needle of the second syringe (air syringe) was inserted through the upper right corner of the Bachofer gasket. The needle of the full loading syringe was inserted into the lower left corner of the gasket and the plunger was gently depressed  110 to inject its contents into the chamber. The chamber was tilted to raise upper right corner and force remaining air or bubbles into the air syringe. When the Pentz chamber was full, the air syringe was removed first, followed by the loading syringe. The specimens were now ready for microscopic observation or the loaded chambers containing the specimens could be stored in the incubator for later observation.  5.  Documentation of Cell Behaviours  The Pentz chamber was placed on the stage incubator (Bachofer) on either an upright microscope (Reichert, Vienna, Austria) or inverted microscope (Zeiss) equipped with reflected Nomarski differential interference contrast (DIC; see Section I.C.6. below) optics and an Epiplan 8x objective (Zeiss). Three methods were used to document cell behaviours:  a. Cells were photographed (Reichert camera used on the upright Reichert microscope; Contax 167 M T (Zeiss) used on inverted Zeiss microscope; Kodak Technical Pan film, Rochester, New York) at 5-15 minute intervals over variable time periods.  b. Cinematography: A cinemicrographic system (Opti-Quip, Highland Mills, N.Y.) controlled the timing and exposure of the 16-mm film (Kodak 2415). This system was generally used on the upright microscope although it could be used on the inverted microscope.  c. Videomicroscopy: A television camera, image processing system (Hamamatsu model C2400; Hamamatsu Photonics K. K., Hamamatsu City, Japan) and a time-lapse video recorder (Panasonic 8050, Panasonic Video Systems) were used. This system was used on the inverted microscope.  Time-lapse observations were used because cell motions were too slow to be observed in real time. Both time-lapse cine and videomicroscopy used exposures of either 1 frame/minute or 1 frame/2  Ill minutes and the successive frames, separated by the above-noted time delays, were recorded. When the film or video was played back at normal speed, cell behaviours were greatly speeded up.  Both cine and videomicroscopy have advantages and disadvantages. Use of cine film is considerably more expensive andtime-consumingin that it requires processing and drying of the film before it can be viewed. However, superior detail was obtained with cine film as compared to the video recordings and images captured on cine film could be retrieved and printed as a hard copy. Although images recorded on a video system were readily accessible for viewing, the Hamamatsu video system described above did not permit retrieval of images for publication purposes and the recorded detail was inferior to that obtained with cine film  6.  Differential Interference Contrast (DIC) Microscopy  Nomarski differential-interference contrast (DIC) microscopy exploits the interference effects produced when the phase of a light wave is changed as it passes through a cell. Thick or dense parts of the cell retard a tight wave and therefore the phase of this tight wave is shifted, relative to the phase of the tight wave that passed more easily and faster through a thinner part of the cell. When these two sets of waves recombine after passing through the cell, the interference effects between the disparate phases of the two waves creates an image of the cell's structure.  DIC was used to observe living cells and their movements but because the titanium-coated substrata were opaque, the technique was adapted for use with reflected light. Images of cells on smooth titanium surfaces were easier to obtain than images of cells on grooved titanium and in particular N grooves (6-9(im-pitch grooves). However, it was determined that if the long axis of the grooves was positioned horizontally rather than vertically in the field of view, then images were easier to obtain. This observation was not unexpected given that DIC utilizes polarized light and objects with a substructure of parallel units, such as fibers, muscles and microelectronics circuits (eg. micromachined substrata, see Section I.D. below), can interfere with the polarized incident  112 light waves (James, 1976; Foster, 1988). In fact, in order to produce the best DIC images, specimens with specific directionality should be rotated so that their long orientation is perpendicular to the direction of the shear (Foster, 1988).  D.  Micromachined substrata  1.  Micromachining  Micromacmning is a technique that allows wafers of single-crystal silicon to be chemically etched into a variety of three-dimensional shapes including pits, holes, trenches, walls, pyramids, hemispheres, etc. (Angell et al., 1983). Titanium-coated micromachined n-type (1-0-0) silicon wafers (Virginia Semiconductor Inc., Fredericksburg, Va, USA) were produced as described previously by Brunette (1986a, b). The micromachined wafers were produced in the laboratories of the Center for Microelectronics, Department of Electrical Engineering, University of British Columbia, directed by Dr. N . Jaeger. The technique was originally developed by Camporese et al. (1981), Department of Electronic Engineering, University of British Columbia, for the fabrication of photomasks for solar cells. Briefly, the technique comprises the following steps (Chehroudi, 1991): a. Cleaning: n-type (1-0-0) silicon wafers (Virginia Semiconductor Inc., Fredericksburg, Virginia) were used in these studies. The wafers were 5 centimeters in diameter, 200-350 iim thick, and had polished front surfaces and bright-etched rear surfaces. Thorough cleaning of the silicon wafers was accomplished by i. 10 minutes in a solution of H 0 (300 ml), H 0 (60 ml), and N H O H (60 ml) at 2  2  2  4  followed by 10 minute rinse in distilled water ii. 30 seconds in 10% HF and 10 minute rinse in distilled water iii. 10 minutes in a solution of H 0 (300 ml), H C L (60 ml); H 0 (60 ml) 2  followed by 10 minutes rinse in distilled water iv. 4 minute immersion in isopropyl alcohol v. blow drying in filtered nitrogen  2  2  75-85o  C  113 b. Oxidation: Growth of a 0.6-(im-thick silicon dioxide layer on both surfaces of the wafer was achieved by using wet oxygen at 1150° C for 2 hours in a furnace (two-inch tube furnaces [no. 7], Fairchild Semiconductor Corporation, USA). c. photolithography. A pattern was produced on a photomask by a computerized optical pattern generator and the pattern was etched in chromium-gold. The front surface of the silicon wafer was coated with negative photoresist (Microposit S-1400 Series (Shipley), Newton, Massachusetts) and the photomask was positioned onto the silicon wafer and exposed with a 320 nm wavelength UV-light through the photomask, developed and baked at 160° C. Precise alignment of the photomask on the wafer was critical for obtaining grooves of precise depth and spacing because misalignment of the photomask results in less predictable etching. d. Oxide patterning. The oxide layers that were unprotected from the UV-light exposure were removed by buffered HF. Next, the remaining developed layer was removed in microstrip (NMP) solvent (Microstrip 2001, Olin Hunt Specialty Inc., West Paterson, NJ, USA). These procedures produced a silicon wafer whose front surface was patterned with the oxide layer. e. Final etching. Etching in 19% K O H at 80° C etched away the silicon, leaving the oxide layer intact. f. Titanium coating: The micromachined grooved substrata were evaporatively coated with 50 nm of titanium. Smooth (unmicromachined) silicon wafers were also coated with titanium and were used as control surfaces. The titanium-coating protocol was adopted by Brunette (1986a, b) in order to eliminate possible heterogeneities in surface chemistry and surface energy between etched and unetched regions. Thus, the micromachined substrata employed by Brunette (eg. 1986a, b) and in the present studies were chemically homogeneous (Singhvi et al., 1994).  2.  G r o o v e Patterns  The geometry and dimensions of the grooves can be precisely controlled. The pitch (combined width of one groove and one ridge) of the grooves is incorporated in the design of the master pattern. The depth of the grooves can be controlled by the time of etching but is also restricted by  114 the pitch and the crystal orientation (Ghandhi, 1983). For example, the speed of etching varies according to the crystal arrangement of the silicon wafer; on 1-0-0 wafers the etch rate is typically 1.4 (im/minutes but this rate is decreased 300 times on 1-1-1 wafers. The crystal orientation of the silicon wafer also determines the shape of the grooves. For example, the n-type 1-0-0 wafers used in these studies produce "V"-shaped grooves whereas 1-1-0 wafers produce vertical-walled grooves.  The experiments described in this thesis employed three different groove patterns. A l l of the grooves had a truncated V-shape in cross-section and the walls of the grooves formed an angle of 55 degrees with the top of the ridges. However, the three groove patterns differed in pitch and depth. a. R grooves: The first pattern consisted of a series of 3 |im deep grooves with a 30-|Xm pitch comprising a 15-um-wide groove and 15-um-wide ridge (Fig. 1). One series of parallel grooves intersected at right angles to a second series in the centre of the wafer. For ease of discussion, grooves leading into the intersection were designated as " X " grooves; those grooves intersecting at right angles with the " X " grooves were designated as " Y " grooves. This pattern was considered the standard or reference pattern and was therefore termed the reference (R) groove pattern.  b. N grooves: The second pattern consisted of a series of 3-^m-deep grooves with a pitch that ranged from 6-9 |im in different areas of the silicon wafer. The areas of different pitch were separated from each other by a 25-p.m-wide ridge-free section which was considered as a wide groove separating ridges of different pitch. This pattern was designated as the narrow (N) groove pattern.  c. V N grooves: The third pattern consisted of 0.5-^m-deep grooves with a 1.0 ^tm pitch and was termed the very narrow (VN) grooved pattern. At this narrow pitch, geometrical and micromachining constraints prevented fabrication of grooves deeper than 0.5 Jim (Ghandhi, 1983).  115  15 (im  I A.  15 nm <  J  >  v_  10.7 (im  Figure 1. Schematic cross-section drawing of the R grooves. The R-groove pattern consisted of a series of 3 pirn deep grooves with a 30 \im pitch comprising a 15-|im-wide groove and 15-|imwide ridge. In cross-section, the grooves had a truncated 'V-shape and the walls of the grooves formed an angle of 55 degrees with the top of the ridges.  116 3.  Final Preparation of Substrata  Typically, a wafer contained several micropatterned areas that were surrounded by smooth titanium. Using a diamond-tipped glass cutter, the newly-fabricated titanium-coated silicon wafers were scored and gendy separated into smaller pieces, ranging 1-1.5 cm by 1 cm. Whenever possible, each wafer was sectioned into small pieces of substrata that contained both a grooved micropattern and a smooth, non-patterned perimeter. Wafers containing both smooth and grooved portions facilitated observation of cell behaviours on both substrata simultaneously.  a. Washing: The titanium-coated substrata were collected into a small beaker and cleaned by ultrasonication for 10 minutes in a detergent formulated for tissue culture (7X, I C N Biomedicals, Inc., Costa Mesa, California) followed by copious rinsing and ultrasonication in filtered deionized water and air-drying in a laminar-flow hood.  After completion of an experiment, substrata were also reused or recycled. In that instance, the wafers were placed into the trypsin solution for at least 30 minutes at 37° C prior to cleaning in detergent and washing as described above.  b. Radio-frequency glow-discharge treatment: Final cleaning and sterilization of the titaniumcoated substrata were accomplished by 3-minute argon-gas radio-frequency glow-discharge (RFGD) treatment (Baier and Meyer, 1988). In RFGD-treatment, plasma clouds of argon gas bombard the surface of the wafers and ash away organic contaminants to render the wafers sterile, highly hydrophilic and very receptive for adhesion (Doundoulakis, 1987; Gombotz and Hoffman, 1987). For experiments in which substrata were seeded with cell suspensions, the substrata received RFGD-treatment just prior to seeding with the cell suspension. For experiments in which droplets of cell suspension were plated onto the wafers, the substrata were sterilized with RFGD treatment at least 48 hours prior to plating the substrata with the droplets. During the interim, the substrata were stored in a sterile 60 mm dishes. This delay was necessary in order for the  117 hydrophilicity produced by RFGD-treatment to diminish enough to maintain a positive contact angle between the droplet and substratum; that is, to maintain the cell suspension within the droplet at the site it was placed. If the substratum was too hydrophilic, the droplet immediately spread along the direction of the grooves and cells in the suspension were swept into or closer to the groove intersection than desired. Alternatively, if the droplet shape was maintained, cells attached at the site of droplet plating and the cells were retained at these sites when the wafers were subsequently flooded with medium, an hour later.  118  P R O T O C O L FOR FIXATION AND STAINING OF SPECIMENS  FIX - *  PERMEABILIZE  INDIRECT I M M U N O F L U O R E S E N C E  F L U O R E S C E N T PHALLOTOXINS  M O U N T SPECIMENS  * CONVENTIONAL EPIFLUORESCENCE MICROSCOPY and PHOTOGRAPHY and/or C O N F O C A L L A S E R SCANNING F L U O R E S C E N C E MICROSCOPY, I M A G E PROCESSING, DIGITAL G R E Y S C A L E IMAGES A N D PRINTS Figure 2. Flow chart of the sequence of procedures used in the preparation of specimens for conventional and confocal epifluorescence microscopy.  119 E.  Preparation of Samples for Examination by Epifluorescence Microscopy  After time-lapse observations, samples were prepared for examination by epifluorescence. A flowchart summary of the sequence of events in the preparation of samples for examination by epifluorescence is presented in Fig. 2. Some steps in the staining of the cell membrane are shared in common with procedures used in staining of the cytoskeleton, but staining protocols for each are described separately below.  1.  Staining to Visualize Cell Morphology  Because the titanium-coated substrata were opaque, cell morphology was visualized using epifluorescence. In this technique, cell-surface glycoconjugates were oxidized in order to form aldehyde groups. Specifically, mild periodate treatment modifies sialic acid residues to yield aldehydes (Wilchek et a l , 1980) which can then react with fluorescently-labelled aldehyde-reactive probes (Molecular Probes) such as hydrazides. The following protocol was used:  a. Wafers were rinsed with warm (37° C) cytoskeletal stabilizing (CS) buffer (Opas, 1989) containing 0.1 M Pipes (Sigma), 1 raM E G T A (Sigma), 4% (w/v) polyethylene glycol 8000 (Sigma); pH 6.9 (achieved by titration with K O H (Sigma))  b. Wafers were fixed 10 minutes in fresh 3.7% formaldehyde (BDH Inc., Toronto, Canada) in warm (37° C) phosphate-buffered saline (PBS) and then washed in PBS buffer  c. Samples were oxidized in fresh 4.2 m M sodium periodate (BDH Inc.) in PBS (.0089 gr NAIO4 in 10 ml PBS) for 30 minutes at 4° C, and then rinsed 3 times with PBS.  d. Excess PBS was blotted from the wafers by holding (with forceps) the wafers sideways and placing one edge of the wafer against an absorbent surface (eg. tissue). The surface of the wafer onto which the cells had been seeded was not blotted. Depending upon the size of the wafer to be  120 stained, 20-50 (il droplets of a 10 m M fluorescent aldehyde-reactive probe (Molecular Probes Inc.) solution (see below) were dispensed into a small culture dish. The wafer was gendy placed onto the droplet of stain, cell-side down. In order to maintain a humid environment, moist filter paper was placed into the bottom of a second larger culture dish or petri dish. The culture dish containing the stain and wafer were placed on the filter paper in the second dish. The entire dishassembly was covered and incubated for 2 hours at 37° C. Specimens were protected from light for all subsequent procedures by covering dishes with aluminum foil.  e. Wafers were rinsed well in PBS. In most instances the wafers could be gendy lifted off the culture dish and placed into PBS. However, if the wafer had dried-down onto the floor of the culture dish, PBS was added directly to the dish, around the wafer so that the wafer floated off the bottom of the dish.  Samples were either mounted (see Section I.E.3. below) and examined immediately or they could proceed with permeabilization and staining of the cytoskeleton. Prepared samples were stored protected from light at 4° C.  Several fluorescent aldehyde-reactive probes were utilized (listed below) but fluorescein-5thiosemicarbazide (Molecular Probes Inc.) produced the most favourable results with human gingival fibroblasts and porcine epithelium, in conjunction with labelling of the cytoskeleton. Fluorescent aldehyde-reactive probes examined for use in these experiments included: (i) . fluorescein-5-thiosemicarbazide (Molecular Probes Inc.) (ii) . lucifer yellow (Molecular Probes Inc.) (iii) . Texas Red Hydrazide (Molecular Probes Inc.)  A l l aldehyde-reactive probes were prepared at a concentration of 10 m M (Wilchek et al., 1980) and the prepared solutions were stored at 4° C and protected from light.  121 2.  Staining of Cytoskeletal and Associated Elements  a. Staining of Microtubules. Kinesin. Vimentin Intermediate Filaments. Vinculin-containing focal contacts To prepare samples for staining of cytoskeletal and associated elements, two fixation/permeabitization protocols were used and the two protocols used similar but not identical buffers. Nevertheless, both protocols produced final results that were similar in their staining of the cytoskeleton and kinesin.  In the first or standard protocol, samples were fixed and then permeabilized. In the second protocol, samples were permeabilized and then fixed. The second protocol was adopted specifically to comply with the protocol described by Ffollenbeck (1989) for the staining of kinesin.  The objective of fixation is to preserve antigenicity and to preserve cell height, three-dimensional spatial relationships and structural details without introducing artifacts. The objective of permeabilization is to permit and optimize antibody penetration into the cells, preserve antigens and prevent extraction of desired antigens. The most desirable sequencing of these procedures is dependent upon the specific agents used, the cell type and the particular proteins or structures of interest. Because publications report a variety of protocols differing in buffers, detergents and extraction times, a comparison screening of results with different protocols is generally recommended (Bacallao et al., 1990)  As noted above, both protocols produced similar results. Initial rinsing or washing of samples in either CS buffer (Opas, 1989 or Hollenbeck, 1989) produced superior results to initial rinsing in PBS, particularly in preservation of microtubules. Formaldehyde fixation was used in these studies because samples were subsequently investigated with confocal microscopy, a technique sensitive to distortion of three-dimensional structure. Formaldehyde causes less than a 5%  increase in cell height whereas methanol or acetone can cause a 20-50 % decrease in cell height (Bacallao et al., 1990). Glutaraldehyde was not considered as a fixative because it often compromises antigenicity and results in high levels of background fluorescence (reviewed by Bacallao et al., 1990).  (i) . Standard Protocol (a) . Wafers were gently washed or rinsed in warm (37° C) CS buffer (Opas, 1989; see Section I.E.I.a above) (b) Samples were fixed 10 minutes in fresh 3.7 % formaldehyde (BDH Inc.) in warm (37° C) PBS. (c) . Samples were permeabilized for 3 minutes in 0.5% Triton-X (Sigma) in CS buffer followed by a wash in CS buffer.  (ii) . Protocol adopted from Hollenbeck (1989) (a) . Wafers were gently rinsed or washed in 0.02% saponin (Sigma) in warm (37° C) CS buffer (Hollenbeck, 1989) containing 0.1 m M Pipes (Sigma), 5 m M M g S 0 (Sigma), 10 m M E G T A 4  (Sigma), 4% polyethylene glycol (Sigma) (b) . Samples were fixed 10 minutes in fresh 3.7% formaldehyde (BDH Inc.) in warm (37° C) PBS or warm (37° C) 0.02% saponin in CS buffer (Hollenbeck, 1989) (c) . Samples were washed in PBS plus 0.05% Triton X-100 followed by a wash in PBS.  (iii) . Common Protocol After this point in each of the above protocols, fixation/permeabilization or vice-versa, and washing were completed and the two protocols continued along identical paths.  (d) . Samples were quenched with 0.05% sodium borohydride (BDH Inc.), prepared fresh in PBS. This step was included as a precaution against autofluorescence, although we observed that  123 omission of this step did not adversely affect staining or visualization of the cytoskeleton.  (e) . Samples were blocked with 1% bovine serum albumin (BSA; Sigma) in PBS for 30 minutes at 37° C.  This step was included to minimize or eliminate any nonspecific binding of the antibodies that could contribute to background labelling and fluorescence. 1% B S A was used to saturate nonspecific protein binding sites prior to application of the primary antibodies and as a washing solution. A 0.1% B S A solution was used to dilute antibodies (primary and secondary) and to dissolve phallotoxins (see Section I.E.2.b. below).  (f) . Cytoskeletal elements, with the exception of actin filaments, were stained by using indirect immunofluorescence. In this technique, the primary antibody binds to the antigen to form an antigen-antibody complex. A secondary antibody, labelled with the fluorophore, is directed against the first antibody. This "sandwich technique" amplifies the original signal and therefore indirect immunofluorescence produces brightness between 4 and 10 times (Osborn and Weber, 1982) that of direct immunofluorescence in which the primary antibody is labelled directly with a fluorophore.  Both monoclonal and polyclonal primary and secondary antibodies were used. Recommendations of the manufacturers were used to dilute lyophilized antibodies but titrations of both the primary and secondary antibodies were performed in order to establish ideal working concentrations for use in these experiments. The specificity of the monoclonal antibodies was determined by either preincubation with antiserum of the same specificity but raised in a different species or by incubation with normal IgG (normal IgM in the case of kinesin) from the animal used to produce the antiserum. In addition, more than one antigen was commonly stained for on the same specimen and if the primary antibodies were produced in the same species, then these stains acted  124  as negative controls for each other because their respective patterns of staining were different. As well, system controls of nonspecific binding of the primary and secondary antibodies were performed by substituting the primary antibody with normal serum from the same species, omitting the primary and/or secondary antibodies and by incubating a sample in diluent only (Dimension Laboratories Inc., 1992; Mississauga, Ontario). Antibody staining was also examined by incubating substrata that contained cells as well as substrata that had not been seeded with cells.  If recommended by the manufacturer and after titration assays determined the ideal working dilutions, appropriate volumes of the primary antibodies were aliquoted into Eppendorf tubes and stored at -80° C. Final dilutions of the frozen aliquots with 0.1% B S A in PBS were completed just prior to incubation of the specimens with the antibody.  When indirect immunofluorescence for tubulin, vimentin, vinculin or kinesin was combined with staining for actin, indirect immunofluorescence was always completed first. That is, incubations with both primary and secondary antibodies were completed prior to staining for actin (see Section I.E.2.b. below).  i. Double-labelling with two primary antibodies: In order to visualize two cytoskeletal elements in the same specimen, it was sometimes necessary to double-label specimens with two primary antibodies. If one of the primary antibodies was a monoclonal antibody (eg. an antibody raised in mice) and the second primary antibody was a polyclonal antibody (eg. an antibody raised in rabbits such as the anti-vimentin from Calbiochem Corporation in ii.c. below), then incubation with both of these primary antibodies could be performed at the same time. Likewise, incubations with both of their respective secondary antibodies could be performed simultaneously. As well, if two polyclonal antibodies were raised in two different animals, then both primary antibodies and their respective secondary antibodies could applied simultaneously.  125 However, if it was necessary to label specimens with two monoclonal or two polyclonal primary antibodies that were raised in the same animal (eg. both antibodies were raised in mice or rabbits, respectively), then it was critical to ensure that cross-over staining with their respective secondary antibodies did not occur. In double-labelling samples with two primary antibodies that were raised in the same animal, cross-over was eliminated as follows: (a) specimen was incubated with first primary antibody (b) specimen was incubated with the first secondary antibody (c) specimen was washed on a rotating stage in 1% B S A solution and the wash solution was changed frequentiy (d) any remaining unbound mouse antibody sites were saturated by incubating the sample with goat anti-mouse IgG (whole molecule, No. M-98645, Sigma). The lyophilized mouse IgG antibody was reconstituted according to the instructions of the manufacturer and was not diluted further. Incubations were performed for a minimum of 1 hour at 37° C and preferably overnight at 4° C (e) the specimen was washed on a rotating stage in 1% B S A with frequent changes of wash solution (vi) specimen was incubated with second primary antibody (f) specimen was washed on a rotating stage in 1% B S A with frequent changes of the wash solution (g) specimen was incubated with a second secondary antibody, ensuring that this antibody was conjugated to a different fluorophore from that used previously in (b) above  The sequence or order in which each of the primary antibodies was applied was important. For example, for double-labelling with vinculin and tubulin, it was necessary to complete the incubations for vinculin first, followed by the incubations for tubulin. If the order was reversed (tubulin first, followed by vinculin), it was not possible to obtain definitive vinculin staining, possibly due to steric interference by previous staining of the microtubule networks. However, in  126 these experiments, staining interferences were apparent only between tubuhn and vinculin.  ii. The following primary antibodies were used: a. mouse-anti-kinesin (clone fBII, ascites (K1005), Sigma). A working dilution of 1:400 was used (1 uT antibody and 400 | i l 0.1 % B S A in PBS).  b. mouse anti-beta tubulin (No. 1111 876, Boehringer Mannheim GmbH, Mannheim, Germany). A final working concentration of 3 Hg/ml was used; 3 (il of antibody were aliquoted, stored at -80° C and diluted with 40 | i l 0.1 % B S A in PBS just prior to incubation with the specimen.  c. anti-vimentin. Both monoclonal and polyclonal antibodies were used. mouse anti-vimentin (Clone V9, No. 814 318, Boehringer Mannheim GmbH). A working dilution of 1 in 4 was used (eg. 10 (il antibody and 30 | i l 0.1% B S A in PBS) rabbit anti-human vimentin (No. 677105, lot 180491, Calbiochem Corporation, L a Jolla, California). A working dilution of 1 in 4 was used; 10 | i l of antibody were aliquoted, stored at -80° C and diluted with 30 ^il 0.1 % B S A in PBS just prior to incubation with the specimen.  d. mouse anti-vinculin (No. 1174843, Boehringer Mannheim GmbH). A final working concentration of lO^g/ml was used; 10 (il of antibody were aliquoted, stored at -80° C and diluted with 40 | i l 0.1% B S A in PBS just prior to incubation with the specimen.  Incubations of primary antibodies were performed either for 1 hour at 37° C or overnight at 4° C. As described above for application of the membrane stain, 20-50 (il of antibody solution at the desired working dilution were dispensed into small culture dishes. After they had been carefully blotted, the wafers were placed, cell-side down, onto the droplets of antibody. In order to  127 maintain a humid environment, moist filter paper was placed into the bottom of a second larger culture dish or petri dish and the culture dish containing the antibody and wafer were placed into the second dish, covered, and incubated at either 37° C for 1 hour, or at 4° C overnight.  iii. Secondary Antibodies After they had been washed in 1% B S A in PBS and carefully blotted, the wafers were placed, cellside down, onto 20-50-pJ-droplets of secondary antibody. Incubations with secondary antibodies were performed for 1 hour at room temperature. Specimens were protected from light with aluminum foil.  The secondary antibodies, either anti-mouse or anti-rabbit IgG antibodies, were selected on the basis of the fluorophores to which they were conjugated. Specifically, the excitation and emission spectra of the fluorophores were considered in relation to the availability of appropriate filter sets for conventional epifluorescence as well as the availability of appropriate lasers and filter sets for confocal laser scanning fluorescence microscopy.  The confocal laser scanning microscope (CLSM 10, Zeiss) available for use was equipped with an argon laser with maximum intensity at 488 nm and a second band near 514 nm. As well, a heliumneon laser with maximum intensity at 543 nm was available. The Zeiss C L S M 10 was equipped to detect only one emission spectrum at a time and therefore it was necessary to select fluorophores that were discrete in their excitation profile with little overlap of their emission spectra. As well, if there is overlap between the emission spectrum of one fluorophore (eg fluorescein or FITC) and the excitation range of another (eg rhodamine or TRJTC), then fluorescence energy transfer may be a concern in that the light from the FTTC could excite the rhodamine.  In experimentation with specimens double-labelled with FTTC and rhodamine, overlap was common in both conventional and confocal epifluorescence microscopy. However, the excitation  128 and emission spectra of FITC and Texas Red are better-separated than the spectra of FITC and rhodamine. In addition, the use of the helium-neon laser (543 nm) for Texas Red labels eliminates excitation of the FITC label. When FITC (or Bodipy fluorescein for actin, see Section I.E.2.b. below) and Texas Red were used in double-labelled specimens, no crossover was observed; specimens labelled with FITC (or Bodipy fluorescein) were not detected in the Texas Red channel and specimens labelled with Texas Red were not detected in the FITC channel.  Consideration was also given to which fluorophore was used to label a particular antigen. Factors to consider include the observation that FITC appears to produce images with better contrast than Texas Red and therefore structures with fine detail yield better images when stained with FITC. Texas Red conjugates also tend to have a lower quantum yield or less efficient fluorescence than fluorescein conjugates. Therefore, in double-labelled samples, it may be advantageous to enhance the Texas Red label by using it to label that antibody that has the greater number of antigen sites in the cell (Bacallao et al., 1990).  The following secondary antibodies were used: a. sheep anti-mouse IgG conjugated to Texas Red (Molecular Probes Inc.) or fluorescein isothiocyanate (FITC; Boehringer-Mannheim GmbH), b. goat anti-rabbit IgG-Texas Red (Calbiochem Corporation) or goat anti-rabbit IgG-FITC (Sigma).  A l l secondary antibodies were used at a 1:40 dilution (eg. l(il antibody and 40 pJ 0.1% B S A in PBS). Secondary antibodies were protected from tight and stored at 4° C.  b. Staining of Filamentous Actin To stain filamentous actin (F actin), specimens were fixed and permeabilized, or vice versa, as described above (see Section I.E.2.a. above). As noted previously, when actin was stained in  129 double-labelled specimens, actin staining followed staining of either the membrane or other cytoskeletal elements with indirect immunofluorescence. Indirect immunofluorescence was not used to stain for F-actin because the protocols described above for indirect immunofluorescence produced unacceptable results with monoclonal anti-actin antibody (No. 1 378 996, Boehringer Mannheim GmbH). Manufacturers of this antibody recommend cold methanol fixation; however, this agent is inappropriate for use in samples prepared for examination with confocal microscopy (Bacallao et al., 1990).  In preparation of double-labelled specimens, samples that were already labelled with Texas Red via indirect immunofluorescence were subsequendy stained for F actin with Bodipy fluorescein phallacidin (Molecular Probes Inc.). In comparison to fluorescein conjugates, Bodipy fluorescein is more photostable, has a brighter fluorescence intensity and has a narrower Stokes shift resulting in less overlap with rhodamines or Texas Red. However, Bodipy fluorescein phallacidin was considerably less photostable than rhodamine phalloidin (Molecular Probes Inc.) and therefore rhodamine phalloidin was used in specimens stained for F-actin alone.  Instructions from the manufacturer (Molecular Probes Inc.) were followed for staining with the fluorescent phallotoxins. Lyophilized phallotoxins were dissolved in methanol, producing a 6.6 pjvl stock solution which was stored at -20° C, protected from light. Just prior to incubation of the specimen with the fluorescent phallotoxin, a desired volume was withdrawn, dispensed into a micro-centrifuge tube (National Scientific Supply Co, Inc., San Rafael, California), dried down on the sides of the tube by evaporation of the methanol with a gentle stream of air and redissolved in an appropriate volume of PBS. The volume ratio of stock phallotoxin to PBS was 1:2, yielding a final working concentration of 3.3 |J.M. The specimens were incubated with the redissolved phallotoxin, typically 20-50 (J.1 per substrata, for 20-30 minutes at room temperature, protected from light, rinsed and mounted.  130 Specificity of the phallotoxin staining was confirmed by an experiment in which each of 4 specimens received one of the following treatments: i) incubation of specimen with Bodipy fluorescein phallacidin or rhodamine phalloidin ii) pre-incubation of specimen with unconjugated phalloidin or phallacidin, respectively, to block F-actin binding sites, followed by incubation with Bodipy fluorescein phallacidin or rhodamine phalloidin, respectively iii) incubation of specimen with unconjugated phalloidin or phallacidin only iv) incubation of specimen with PBS only  3.  Mounting the Stained Samples  Once the samples were stained and rinsed, the wafers were blotted and placed onto glass microscope slides. Glass coverslips were mounted onto the substrata with 1:1 glycerokPBS solution containing 0.02% azide (Sigma) and l,4-diazabicyclo[2.2.2]octane (DABCO, 100 mg/ml; Sigma). Samples were placed in slide trays covered with aluminum foil, stored at 4° C and examined within 48 hours of preparation.  The mounting solution described above was used because it rendered the substrata easily retrievable and because a 50% glycerol and PBS solution produces the most favourable results in terms of preventing cell shrinkage and preserving cell height (Bacallao et al., 1990). D A B C O was used as an anti-bleaching agent as it markedly inhibits bleaching of fluorescein and rhodamine (Small et al., 1988). Some yellowing of specimens was noted after about one month; however, if these specimens were gently washed and remounted with fresh solution and a new coverslip, the images produced from the samples were not adversely affected.  F.  Epifluorescence Microscopy  Cells and cytoskeletal elements were examined through a microscope equipped with both conventional and confocal epifluorescence microscopy (Confocal Laser Scanning Microscope  ( C L S M 10), Zeiss). For double-labelled specimens, the FITC or Bodipy fluorescein signal was examined first, before the rhodamine or Texas-Red signal as the latter fluorophores bleach less rapidly than FTTC or Bodipy fluorescein, even in the presence of anti-bleaching agents (Bacallao et al., 1990).  1.  Cell Morphology and Quantitation  Cells stained with fluorescein thiosemicarbazide were examined with epifluorescent illumination (HBO 50 lamp, C L S M , Zeiss) and photographed (Olympus PM4T camera, Kodak, Tmax 400) using either a 20x (NA 0.50) or 40x (NA 0.75) pan neofluar objective (Zeiss). Fields were selected randomly and the only criterion for inclusion of a cell in the sample was the absence of contact with other cells.  a. Cell Height After photography, the same cells were examined by confocal microscopy using a 63x oilimmersion plan-apochromatic objective, N A 1.4 (Zeiss) and an argon laser (\ =48%). By using max  the phi Z function, it was possible to measure the fluorescent signal in the z-axial direction in a desired xy line. Phi Z section planes were taken over cell nuclei. For cells on smooth surfaces, the direction of the xy line was a line with randomly-selected orientation and maximum cell height was measured from the top of the nuclei to the plane of the substratum. For cells on grooved substrata, the direction of the xy line was perpendicular to the direction of the grooves so that aspects of the cells on the ridge and groove could be included. Maximum cell height was measured from the ridge or groove located immediately below the nuclei. The number of grooves and ridges spanned by each cell was also recorded.  b. Computer-assisted Morphometry: Projected Area and Form Factor. Form Ell Images of the photographed cells were projected (Durst photographic enlarger system, 50mm lens, condenser 130/85, convex to convex, magnification lcm=7cm) and cell outlines as well as  132 corresponding groove-ridge patterns, where applicable, were traced onto paper. The tracings were retraced on the digitizing tablet of a morphometries system (Videoplan, Zeiss) which calculated the projected cell area and a measure of cell shape, Form Ell.  Form Ell is a ratio of the minor to major axes of the cell. A value of 1.0 corresponds to a circular cell and values less than one indicate elliptical or spindle-shaped cells. The longest diameter between two edges within the cell was considered the major axis and the minor axis was considered the longest axis perpendicular to the major axis. Form Ell was used as a measure of shape for fibroblasts and E cells (see Section II below) under drug-free conditions on smooth and grooved surfaces. It was also used for fibroblasts under colcemid, C B and C B / C conditions.  c. Dmax The maximum cell diameter (D max) was the major axis of the cell. Dmax was measured with a ruler as the longest distance within the cell borders.  d. Orientation and Polarity The angle formed by the major axis (Dmax) to either the grooves or a line with randomly-selected orientation for smooth surfaces was termed the orientation angle and was measured with a compass. For measurement of cell orientation, no attempt was made to distinguish cell polarity and therefore the maximum angle possible was 90 degrees, with a value of 0 degrees indicating a cell that was perfectly aligned with the direction of the grooves. Cells were considered aligned when their orientation angle measured less than 10 degrees to the direction of the grooves (Clark et al., 1990). Cell polarity was used to describe cell shape, similar to Glasgow and Daniele, (1994). When a single front or head of a cell was clearly identifiable, as in cells with a well-defined leading lamella, the cells were termed monopolar. Two distinct head regions described bipolar cells, and apolar cells were round. Cell alignment thus was not contingent upon cell polarity and cell polarity did not imply alignment.  133 2.  Microscopy of Cytoskeletal Elements  Cytoskeletal and associated elements were examined with both conventional and confocal epifluorescence microscopy using a 63x oil-immersion plan-apochromat objective, N A 1.4 (Zeiss) on the C L S M (Zeiss). A n argon laser (A.max=488nm) was used for FITC- or Bodipy fluoresceinlabelled cells in confocal microscopy; a helium-neon laser (Xmax=543nm) was used for rhodamine or Texas-Red labels. For double-labelled specimens, the FITC or Bodipy fluorescein signal was examined first, before the rhodamine or Texas-Red signal. Digital grey-scale images obtained by the C L S M were printed using a video printer (UP-5000, Sony Canada, Richmond, British Columbia). The prints were labelled and stored in manila file folders, protected from light, at room temperature.  Confocal microscopy was used to examine the cytoskeleton in these experiments because this technology produces optical sections of intact specimens at the light microscopic level (reviewed by Shotton, 1989). Precise focal planes can be selected and because out-of-focus information is eliminated from the final image, confocal imaging of cells on grooves makes it possible to distinguish the distributions of cytoskeletal elements on top of the ridges and within the grooves.  The C L S M (Zeiss) is a scanned-beam system in which the specimen is stationary and is scanned by a focused laser light beam that moves laterally in a regular two-dimensional raster across the sample, in a given focal plane. Once a desired focal plane is identified as the initial z plane for a series of optical sections, this initial z plane is automatically relocated at the end of the scan. Because the Zeiss C L S M is presently not equipped for dual detection of two emission spectra, recovery of the initial z plane is particularly useful for double-labelled specimens. For doublelabelled specimens, the lasers and their respective filters were changed without altering the focus, thus ensuring that subsequent optical sections with the second laser were made in the same focal planes as the first series.  134 a. Cytoskeletal Patterns - Descriptions and Criteria for Alignment Cytoskeletal elements were examined for their overall morphology as well as their orientation with respect to the direction of the groove/ridge pattern. Because individual cells spread both on ridges and within grooves, morphology and orientation on the ridge and groove levels were assessed independently. Cytoskeletal elements were classed as aligned when their long axes made an angle less than 10 degrees to the direction of the grooves, similar to the criterion used to determine cell alignment (Clark et al., 1990). When individual cells were examined, the ridge and groove morphology of each cytoskeletal element were tabulated according to one of the pattern categories. Description of the patterns was approached in a manner similar to that of Hynes and Destree (1978b), in which the distribution of the cytoskeletal patterns with time was described by the percentage of cells in the sample that exhibited each pattern. (See Appendix 1 for tables of cytoskeletal patterns in fibroblasts under control conditions on smooth and grooved substrata). i. actin The actin patterns in cells on smooth surfaces were adapted from categories described by Hynes and Destree (1978b). However, several categories had to be added to reflect cell behaviour on grooved surfaces. In particular, ruffles and circumferential fibers were evident in cells on the grooved surfaces, but the circumferential fibers were limited to the ridge level and often spanned over the grooves. In addition, actin condensations along the ridge/groove interface were specifically noted, as were aligned microfilament (MF) bundles in any location. Typically, alignment of M F bundles was an all-or-nothing affair; when cells contained aligned M F bundles, the bundles were prominent.  ii. Vinculin Observations on vinculin staining were classed as diffuse, or as punctate, or as discrete focal contacts. In addition, the predominant orientation of the vinculin in stained structures was noted. Where vinculin staining indicated focal contacts with parallel, perpendicular and intermediate angulations to the direction of the grooves, the distribution was described as radial. However,  135 when the majority of structures were oriented at angles less than 10 degrees to the direction of the grooves, the distribution was described as aligned.  iii. Microtubules When microtubules (MT) were distinguishable, they generally exhibited either an overall radial or stellate pattern, or alignment. Although most MTs could not be traced from their point of origin to termination, the free ends could usually be discerned. When the majority of M T segments detected within a groove or on a ridge were aligned less than 10 degrees to the direction of the grooves, M T orientation was designated as aligned. As well, the alignment of M T segments with respect to the groove and ridge edges was noted. A distinct microtubule-orienting center (MTOC) could sometimes be determined from the distribution of MTs, and its location as to ridge, groove or both was recorded.  iv. Intermediate Filaments The patterns and distribution of intermediate filaments (IF) were described by the same criteria as outlined above for MTs. However, IFs appeared to encircle or originate around the nucleus without evidence of a discrete organizing center.  G.  Statistics  Data were entered into a mainframe computer (IBM) and analyzed using SPSS-X (SPSS Inc., Chicago, Illinois). Projected area, maximum cell diameter (D max), cell height, form factors, orientation angle and the effects of and the interaction between both time and surface were analyzed by two-way analysis of variance (ANOVA). In addition, some data were analyzed by StudentNewman-Keuls (SNK). Because the data on the percentage of aligned cells were not normally distributed, they were analyzed using the Kruskal-Wallis one-way analysis of variance.  The proportions of cells exhibiting particular characteristics such as alignment (eg. chapter 4,  Section I) or arborized processes (chapter 4, Section IV) or specific patterns of cytoskeletal elements (chapter 4, Section I and Appendix 1) per time period on smooth and grooved surfaces were expressed as percentages.  137 II. Epithelium The experimental protocols for E cells was very similar to those described above for fibroblasts. However, there were some differences unique to epithelium in the methodology used to establish the E cultures and to assess cell morphology and cytoskeletal patterns. These differences in methods and materials are described below. A.  Cell Culture  Periodontal ligament (PDL) contains nests of undifferentiated E cells known as epithelial rests of Malassez (Gaunt et al., 1971). These E cells were cultured from porcine periodontal ligament as described by Brunette et al. (1976). 1.  Starting Cultures from Periodontal Ligament  Freshly-slaughtered porcine jaws were obtained from Olympic Meat Packers, Pork Department, Vancouver, B.C. The following procedures were performed:  a. The teeth were brushed with bleach.  b. The 1 and 2 st  n d  molars were extracted by incising the gingiva around the teeth with a scalpel.  After each tooth was broken in half (mesio-distally) with an elevator, each tooth segment, including the root, was removed with #2 forceps. Only teeth with intact root apices and small amounts of adhering gingiva were selected. Each tooth segment was placed into a 100 mm culture dish (1 dish per pig) containing medium (see below) and stored in an incubator while other dishes were prepared and processed.  c. Subsequent procedures were performed in a culture hood using sterile techniques. One tooth fragment was placed into a 60 mm dish and the crown of the tooth was held with large forceps while the root was scraped with a sterile scalpel. Only the middle third of the root was scraped as  138 only P D L cells were desired. The edge of the scalpel blade was used to scrape along the long axis of the root. The PDL scrapings were washed from the scalpel blade in a tube containing 10ml medium (1 tube per pig). Tubes were stored in the incubator while the remainder of the teeth were prepared. d. From this point the protocol was similar to that already described for the preparation of gingival explants (see Section L A . 1 above). As this technique yielded both E cells and fibroblasts the two cells types had to be separated. Separation of porcine E cells and fibroblasts could be accomplished through their different detachment susceptibilities by trypsin (E cells are more resistant to detachment by trypsin, Brunette et al., 1976) as well as by their differential preference for culture medium supplements. Porcine E cells thrived in medium supplemented with foetal bovine serum (Cansera International Inc., Rexdale, Ontario, Canada), porcine gingival fibroblasts do not. 2.  Established E Cultures  Barely-confluent E cells between the 2  nd  and 6 subculture were cultured in Alpha Minimal th  Essential Medium (MEM) (Terry Fox Labs, Vancouver, B.C.) supplemented with antibiotics (penicillin G (Sigma, St. Louis, Missouri) 100 lig/ml, gentamicin (Sigma) 50 (ig/ml, amphotericin B (Fungizone, Gibco) 3 |ig/ml) and 15% foetal bovine serum (Cansera International Inc.) at 37° C in a humidified atmosphere with 5% CO2. Cells were removed from their growth surface by a trypsin solution. Population densities of l x l O cells/ml were used for experiments lasting up to 3 5  hours, and l x l O cells/ml for 15 or 24 hours. For time-lapse observations, cell suspensions of 4  2 x l 0 cells/ml were used. 5  B. 1.  Epifluorescence Microscopy Cell Morphology and Quantitation  Cells stained with fluorescein thiosemicarbazide were examined with epifluorescent illumination  139 (HBO 50 lamp, C L S M , Zeiss) and photographed (Olympus PM4T camera, Kodak, Tmax 400) using either a 20x (NA 0.05) or 40x (NA 0.75) pan neofluar objective (Zeiss). Fields of E cells were selected randomly and included clusters of cells (>3 cells), pairs of cells and single cells. Images of the photographed cells were projected, traced and digitized as described above for fibroblasts.  a. Single cells and Cells of a Pair or Cluster For cells within a pair or cluster, cell response was considered in two ways. First, the entire unit, that is, the entire pair or cluster, was measured. Then, individual cells within the pair or cluster unit were measured. Thus, the sample included single cells, individual cells within a pair or cluster as well as the entire pair or cluster as a unit. It was not possible to establish whether the pairs and clusters observed in fixed samples were seeded onto the substratum as such, or whether these groupings resulted from single cells spreading in close proximity. For single cells, the only criterion for inclusion in the sample was the absence of contact with other cells. For cells within a pair or cluster, the proportion of each individual cell's perimeter in contact with adjacent cells was estimated and ranked as <25%, 26-50%, 51-75% or >76%.  b. Computer-assisted Morphometry - Projected Area and Form Factors  Projected areas and form factor, Form Ell, were calculated. The longest diameter within the cell or within the entire pair or cluster was considered the major axis or maximum diameter (Dmax) and the minor axis was considered the longest axis perpendicular to the major axis.  c. Orientation and Polarity  Two measures of orientation were used and neither required an attempt to distinguish polarity. The first was Orientation Angle which was the angle formed by the major axis either to the direction of the grooves or, for smooth surfaces, to a line with randomly-selected orientation. Cells or pairs  140 and clusters of cells were considered aligned when their orientation angle measured less than 10 degrees to the direction of the grooves or line (Clark et al., 1990), and a value of 0 degrees indicated a cell or pair/cluster that was perfectly aligned with the direction of the grooves or tine. The second measure of alignment was Orientation Index (01), defined as LAV where L is the length of the longest segment of the cell or pair or cluster in the direction of the grooves or line and W is the width at right angles to the grooves or the tine at the widest point of the cell or pair/cluster (Brunette, 1986b). A n OI value of 1.0 indicates the absence of a preferred orientation in relation to the direction of the grooves or line; an OI>1.0 indicates preferential orientation in the direction of the grooves or line. Alignment of cells was not contingent upon cell polarity and polarity did not imply alignment.  C.  Statistics  For E cells, analysis of variance (ANOVA) and Scheffe's test were used to analyze the effects of surface and percentage of adjacent cell contact for projected area, Form Ell, OI and log OI, the most suitable variate for statistical analysis.  141 CHAPTER 4 RESULTS  Introduction and Overview This chapter is divided into six sections which correspond to different experimental conditions. Sections I-V comprise experiments with human gingival fibroblasts. Sections I and II deal with fibroblasts under control conditions on smooth or R grooves (see Chapter 3, Section I.D.2.). Sections HJ, IV and V describe experiments in which cytoskeleton-perturbing agents were used and Section V also includes data of cell behaviours on N and V N grooves. In order to facilitate comparisons between cells under control and drug conditions, there is some overlap in presentation of data on controls as well as referencing to earlier sections. It should also be noted that controls were performed in conjunction with each of the experiments using cytoskeleton-perturbing agents and that under control conditions, cell and cytoskeletal behaviours in the various experiments were consistent. Section VI comprises observations of epithelial (E) cells, under control conditions, on smooth and R grooves.  Results of fibroblasts under control conditions on R grooves are presented first (Sections I and II). In Section I, I described cell behaviours and the development of cytoskeletal patterns in cells spread on smooth titanium and on R grooves. Observations ranged from the time the cells were seeded onto the substrata, to 6 hours and beyond (up to 72 hours) when polarization and alignment (on grooved surfaces) were completed and migration occurred. In Section U , I described cell and cytoskeletal behaviours of fibroblasts in response to conflicting topographic cues, such as intersecting R grooves. Based on observations in these sections, I suggested that MTs may determine cell orientation in response to uniform and conflicting topographic cues.  I performed experiments with cytoskeleton-perturbing agents in order to examine the effects on cell  142 orientation on R grooves when the influence of MTs, actin MFs or both were eliminated. The effects of colcemid, cytochalasin B (CB), or both (CB and colcemid, CB/C) on cell and cytoskeletal behaviours on R grooves are presented in Sections III and IV respectively.  In Section III, I established that if MT-deficient cells were provided with enough time, an appropriate substratum such as R grooves could compensate for the M T deficiency and I termed this process "topographic compensation".  I observed that CB-treated cells were not motile due to destruction of the actin network (Section IV). However, MT-containing processes aligned with the R grooves within a time frame and with topographic responses similar to those observed in controls, supporting the role of MTs as the principal but not sole cytoskeletal mechanism effecting topographic guidance.  In Section V , I examined cell behaviours on N and V N grooves under control, C B , colcemid and C B / C conditions and compared them to the previous observations (Sections I-IV) on R grooves and smooth surfaces. Alignment responses of MTs to specific topographic features were consistent on R, N and V N grooves and not surprisingly, alignment of MTs increased as the pitch of the grooves decreased. In fact, the enhanced alignment of CB-treated cells on N and V N grooves lent further support to the role of MTs as a substratum-sensing mechanism. However, V N grooves failed to enhance topographic guidance of controls and failed to effect topographic compensation (Sections i n , IV), suggesting that alignment of both MTs and MFs was required for optimal guidance of fibroblasts on micromachined grooves.  In Section VI, I examined the cell and cytoskeletal behaviours of porcine E cells spread on R grooves. Although E cells differed from fibroblasts in their social and locomotory behaviours as well as in cytoskeletal organization, similarities to fibroblasts in alignment behaviours on R grooves were evident. Alignment of both actin M F bundles and M T was initially and most  143 consistently observed along the walls and ridge/groove edges and overall, cytoskeletal alignment preceded and exceeded alignment of E cells as a whole. However, it was also apparent that local topographic effects on the cytoskeleton could be overridden by adjacent contacts between E cells.  In each section, numerous figures have been included to illustrate specific features that are presented in the results. To prevent disruption of the readers' train of thought throughout the results, the figures that pertain to each section are presented in order, together at the end of each section.  144 I.  The Sequence of Alignment of MTs. Focal Contacts and Actin M F Bundles in Fibroblasts Spreading on Smooth and R-grooved Substrata  The appearance of fibroblasts cultured 24 hours on smooth and R-grooved substrata is illustrated in Figure 3. The experiments in this section follow the morphologic and cytoskeletal changes through the early stages of cell spreading and alignment that eventually result in the appearance of the polarized cells depicted in Figure 3. This experiment was designed to identify a candidate for the principal cytoskeletal determinant of cell alignment by determining which cytoskeletal component was first to respond to and align with the substratum topography (see Chapter 2).  Morphometric and cytoskeletal data are presented for cells on smooth surfaces first, followed by data for cells on R grooves. On each surface, the distribution of cytoskeletal components (see Chapter 3, Section I.F.2.) is described as a function of time. Tables and detailed descriptions of the developing cytoskeletal patterns are presented in Appendix 1.  The morphometric measurements in this section were based on between 93 and 262 cells examined, per time point, on the smooth or R-grooved surfaces at 20,40, 60, 120, 180, 240 and 360 minutes. The proportions of cells exhibiting the cytoskeletal patterns for actin M F bundles, vinculin, MTs and M T O C were based on between 100 and 155 cells per cytoskeletal element per time period on smooth and R-grooved surfaces. The proportions of cells with aligned structures were expressed as percentages.  MTs, actin M F bundles and focal contacts were selected for examination in these experiments because they form the basis of current, accepted hypotheses of contact guidance (see Chapter 1, Section IV). IFs were included for the sake of completeness and the distribution of kinesin was included because kinesin has been imphcated as a mediator of polarization (Rodionov et al., 1993). In this thesis, the cytoskeleton and associated elements were examined with conventional and confocal epifluorescence microscopy and the criteria used to describe the morphology and  145 distribution of these elements has already been described in Chapter 3, Section I.F.2.  It should also be noted that the role of individual actin filaments in cell orientation could not be discerned through the light-microscopic examinations that are described in this thesis. Therefore, all comments regarding actin filaments in the experiments that comprise this thesis are made in reference to actin M F bundles or stress fibers. It should also be reiterated that the most widelyaccepted hypothesis on the mechanism of contact guidance (reviewed in Chapter 1, Section IV.B.l.) is based on stress fibers and not individual filaments and therefore observations made in this thesis can be legitimately compared to other reports that utilized light-microscopy (eg. Hynes and Destree, 1978b).  A.  Smooth Surfaces  1. Cell Morphology and Quantitation (see Chapter 3, Section I.F.I.) Over 6 hours, cells spreading on smooth surfaces gradually increased in projected area (Fig. 4) and decreased in nuclear height (Fig. 4). Cell shape gradually became less circular and more elliptical or irregular (Fig. 5) During the course of spreading, the orientation angle did not differ significantly from 45 degrees (Fig. 5) and the percentage of aligned cells did not increase with time. This was expected as no known orienting influences were present on the smooth surfaces.  2.  Cytoskeleton and Associated Elements  a. Actin In agreement with Hynes and Destree (1978b), the distribution of actin staining through the course of cell spreading progressed through a series of patterns. The distribution of actin patterns with time is given in Appendix 1, Table A . l . At early times, actin was found predominantly in ruffles. By 6 hours, well-developed M F bundles were evident and the distribution of actin M F bundles reflected the shape of the cells. At 24 and 72 hours, actin patterns were similar to those observed at 6 hours and in most cells, the leading edges of the cells could be determined from the  146 distribution of the actin M F bundles (Fig. 6a).  b. Focal contacts (Vinculin Staining) Distinct focal contact localizations were evident in 19% of the cells after 20 minutes of spreading on smooth surfaces. By 60 minutes, over 90% of cells exhibited distinct focal contacts, most commonly located at the cell perimeter. At 6 hours and later, focal contacts were still most prominent at the cell edges but some were found within central areas of the cell as well (Fig. 6c). In general, focal contacts were closely associated with the termination of actin MFs.  c. Microtubules At 20 minutes, MTs were not distinguishable in almost half of the cells, but by 40 minutes over 80% of cells demonstrated MTs in a radial or stellate pattern. The distribution of MTs reflected the shape of the cells, and in most cells, the leading edges of the spread cells could be determined from the distribution of MTs. The MTOC could be inferred from the pattern of the MTs in most of the cells (Fig. 6b).  d. Vimentin Intermediate Filaments By 40 minutes, IFs were evident in almost half of the cells and by 60 minutes, the majority of cells demonstrated distinct IFs. At these early times, IFs did not extend to the cell peripheries as MTs did (Fig. 7) and hence the distribution of IFs did not reflect the shape of the cells. By 6 hours, the distribution of IFs was similar to the distribution of MTs although IFs were not prominent in lamellae (Fig. 8), similar to the observations of Hynes and Destree (1978a) and Terasaki et al. (1986).  e. Kinesin In agreement with others (Hollenbeck 1989; Hollenbeck et al., 1989 and Okuhara et al., 1989), the distribution of kinesin staining was similar to the distribution MTs, although their respective  147 distributions differed in some areas (Fig. 8).  3.  Time Lapse Observations on Smooth Surfaces  Cell spreading, polarization and locomotion on smooth titanium were similar to reports of cells spreading on other flat surfaces (eg. Abercrombie 1980). The phase of radial spreading was followed by the phase of polarization in which ruffling and protrusions became limited to a leading edge. Locomotion was characterized by extension of a protrusion and general stretching of the cell, followed by a sudden recoil. Intermittent repetition of this sequence resulted in cell motion that has been described as a modified random walk (Gail and Boone, 1970).  B.  R Grooves  1.  Cell Morphology and Quantitation  Significant differences (P<0.001) in projected area were observed between cells spreading on grooved and smooth substrata with respect to both surface and time (Fig. 4). There was considerable variability in cell size and shape on both grooved and smooth surfaces. At all time intervals, some cells had spread while others remained spherical, but the relative proportions of each changed with time. Projected cell areas were less for cells on grooved surfaces although cell spreading did not appear to be restricted to the direction of the grooves as cells clearly were able to span laterally over several grooves and ridges. Typically, up to 40 minutes, cells were spherical with no preferred orientation and were localized primarily within one groove with some lateral spreading onto ridges (eg. Figs. 9,10). With increased time, the area occupied by a cell increased so that several ridges and grooves were spanned. Lateral spreading (i.e. number of grooves and ridges spanned) was maximal at 3 hours and had decreased by 6 hours so that the cells had become more elliptical and were once again located primarily within one groove. However, some of these cells had cell processes at one or both ends of the cell that spanned one or two ridges (eg. Fig. 11).  Although there was no significant difference between the maximum diameter of cells on grooved  148 and smooth surfaces, the orientation of fibroblasts differed significantly (P<0.001). With increasing time, cells on grooved surfaces increased their orientation (Fig. 5) and likewise, the percentage of aligned cells also increased with time (Fig. 12).  A t 20 minutes, Form Ell was similar for cells on both grooved and smooth substrata (Fig. 5). By 6 hours, significant (P<0.001) differences were evident, with cells on grooves being more elliptical and less circular, and these differences continued to be significant even at 72 hours. Cells on grooves also demonstrated a significandy (P<0.001) greater height than cells on smooth surfaces (Fig. 5).  2.  Cytoskeleton and Associated Elements  a. Overview of Alignment of Cytoskeletal Elements on R Grooves Figure 12 summarizes and compares the percentages, over time, of cells displaying aligned focal contacts, MTs and actin M F bundles (see Appendix 1 for Tables of cytoskeletal patterns). The percentages of aligned cells at each time interval were included to permit comparison between cytoskeletal alignment and cell orientation.  At 20 minutes, no actin M F bundles, aligned or otherwise, were evident in any of the cells examined. However, discrete vinculin staining indicative of focal contacts was apparent in 47% of cells including 14% of cells that demonstrated focal contacts aligned to the direction of the grooves (Appendix I, Tables A.2, A.3). In 55% of cells, MTs (Appendix I, Table A.4) within the grooves were aligned to the direction of the grooves even though the cells themselves were not oriented to the direction of the grooves. For example, Figure 9 shows a cell stained for both actin and MTs. The only structures staining for actin were ruffles, yet the MTs were aligned to the direction of the grooves and were located close to the walls and floor of the groove. Even through MTs were aligned, the cell itself showed no orientation with the grooves.  149 The percentage of cells demonstrating aligned MTs increased with time. B y 6 hours all cells demonstrated M T alignment. Actin M F bundles were not aligned in the majority of cells (69%) until 2 hours, but by 6 hours, 97% of the cells displayed aligned M F bundles. Aligned focal contacts were evident prior to the detection of M F bundles and the percentage of cells displaying them actually exceeded the percentage of cells with aligned M F bundles up until 60 minutes. By 3 hours, a majority of cells had aligned focal contacts and there was a further increase by 6 hours (Fig. 12).  In summary, for cells spreading on a grooved substratum, considered at the population level, MTs were first to align with the direction of the grooves and were then followed by focal contacts and actin M F bundles. A detailed description for each cytoskeletal component is given below:  b. Development of Cytoskeletal Orientation on R grooves i. Actin Like cells on smooth surfaces at early times, cells on grooved surfaces displayed ruffles only or ruffles in combination with circumferential fibers (Appendix 1, Table A. 2). Circumferential fibers were limited to the ridges and with time spanned over the grooves (Figs. 13,14). Actin condensations (Fig. 15) along the ridge/groove edge were evident until 2 hours, but no actin M F bundles aligned to the direction of the grooves were evident until 40 minutes when a small percentage (10%) of cells displayed some aligned M F bundles along the wall and in the groove. Figure 16 shows a cell that is representative of a small subset (4%) of cells observed between 2 and 6 hours that displayed ruffles and circumferential fibers in conjunction with M F bundles aligned on the ridge, along the wall and within the floor of the groove. The overall shape of these cells was generally circular without orientation to the grooves.  By 2 hours, circumferential bridging fibers at the ridge level, in association with M F bundles aligned within the groove, became the predominant actin pattern. Initially, more M F bundles were  150 observed along the walls of the groove than on the floor in the centre of the groove (eg. Fig. 14). Over the next 4 hours, an actin distribution began to predominate that comprised aligned M F bundles on the ridge as well as in the groove. Many cells also exhibited actin M F bundles oriented perpendicular to the direction of the grooves in processes at the ends of the cell at the ridge level (Fig 1 la). By 6 hours, cells were typically elliptical, aligned within a single groove, and they demonstrated aligned M F bundles. A similar distribution of actin M F bundles was observed at 24 hours and beyond and the actin distribution reflected the shape and alignment of the cells (Fig. 17a).  ii. Focal contacts (Vinculin) In general, more focal contacts were found on the ridges than in the grooves. When located within grooves, the focal contacts were frequently located at the ends of the cells (Fig. 10).  At 20 minutes, discrete vinculin staining was evident at both the ridge and groove levels in 47% of cells (Appendix 1, Table A.3). Over the 6 hour time span, focal contacts were radially distributed in grooves and on ridges, but the predominant pattern between 40 minutes and 4 hours combined focal contacts with radial orientation on the ridges but with alignment to the direction of the groove in the grooves. By 6 hours, focal contacts aligned with the direction of the grooves in both the grooves and on the ridges had increased but did not predominate. Despite these general patterns, focal contact orientation could vary both on the ridges and within the grooves and remained variable over time. For example, Figures 10,17b and 18 illustrate focal contacts in a groove; some focal contacts were oriented parallel and others were oriented perpendicular to the groove direction. Clearly the groove width was sufficient to accommodate both orientations.  Focal contacts could also be observed along the walls of the groove (Figs. 10, 18). Between 40 minutes and 6 hours, 10% of cells exhibited focal contacts which appeared along the walls but also appeared to bend around the edges of the ridge and groove and between the floor and wall of the  151 groove.  The relationship between vinculin and actin M F bundles changed with time. As indicated in Figure 12 and Tables A.2 and A.3 (Appendix I), focal contacts were evident prior to the presence of actin M F bundles. In fact, focal contacts were aligned on both ridges and grooves prior to and without association with the M F bundles. Likewise, focal contacts on the walls were not always observed in conjunction with M F bundles.  iii. Microtubules Almost half of the cells at 20 minutes exhibited either very dense staining, in which distinct structures were not evident, or MTs that were arranged in a radial pattern at all levels (Appendix 1, Table A.4). In the balance of the cells, MTs were distributed radially at the ridge level or above, but within the grooves, MTs were aligned to the direction of the grooves, along the walls and floor of the grooves (Fig. 9). During the early stages of spreading, the course of individual MTs could not be readily distinguished in many cells but some cells displayed collections or small bundles of MTs that emerged from the MTOC in a direction parallel to the grooves and these MTs aligned along the walls, along the length and depth of the grooves (Fig. 19).  The pattern of radially oriented MTs at the ridge level and aligned MTs within the grooves predominated until 3 hours. With increased time of spreading, cells spanned a number of grooves and ridges and aligned MTs could be found in each groove (Fig. 14, 20, 21). In instances where several grooves were spanned, some MTs appeared to align within the grooves from their initial point of contact with the wall of the groove (eg. Fig. 21). That is, those MTs that emerged from the M T O C in non-parallel directions to the groove, came to align within the groove from the point of apparent contact with the wall of the groove. In contrast, the alignment of MTs on the intervening ridges was variable but the radial pattern of M T orientation at the ridge level was gradually replaced by a progressive alignment of the MTs along the ridge, with alignment on the  152 ridge being apparent first at the edges of the ridges (Figs. 13, 20). Thus, the alignment of MTs by grooves appeared to be under local control of both the walls of the grooves and the ridge edges. Alignment of MTs within the grooves and along the ridge edges also appeared to override the penetration of MTs into some cell processes that were oriented in non-parallel directions to the direction of the grooves. For example, the cell in Figure 13 demonstrates two processes, identified by phalloidin staining, that were oriented perpendicular to the direction of the grooves within the grooves. MTs on the adjacent ridges were aligned along the edges of those ridges but MTs were not detected within the processes. It is unlikely that the ridge edges could have presented a physical barrier to M T penetration into the processes within the adjacent grooves, yet, nevertheless, the MTs were confined to and aligned along the ridge edges. Overall, 10% of cells between 20 minutes and 6 hours demonstrated cell processes, oriented in non-parallel directions to the grooves, that were devoid of detectable MTs.  By 4 hours, aligned MTs on the ridge and within the groove were the predominant pattern (Fig. 1 lb). By 6 hours, all cells demonstrated aligned MTs within the grooves and only 14% of cells did not exhibit MTs aligned at the ridge level. At 24 and 72 hours, the distribution of MTs closely reflected the shape and alignment of the cells to the direction of the grooves (Fig. 17c).  iv. Microtubule Organizing Center (MTOC) In 73% of all cells observed on grooved surfaces, the location of the M T O C could be determined from the overall distribution of the MTs (eg. Figs. 9f, l i b , 13d, 20, 21).  Of the MTOCs visible in cells spreading on the grooves, 72% were localized within a groove (eg. Fig. 21), 15% were on a ridge (eg. Fig. 20) and another 13% occurred over both the ridge and groove. 87% of cells with an MTOC on a ridge concurrendy exhibited MTs aligned to the direction of the groove within the groove. For example, Figure 20 clearly illustrates a M T O C over a ridge with clearly aligned MTs in the two adjacent grooves and along the edges of two adjacent  153 ridges.  The relationship between M T O C location, nuclear position and M T alignment was variable. The M T O C was classed as anterior or posterior to the nucleus as judged from the probable direction of cell movement suggested by cell shape. Some MTOCs were located anterior to the nucleus (eg. Fig. l i b ) whereas others were located posteriorly, and still others laterally (eg. Fig. 13). In some instances, the nucleus and M T O C were located in a groove that also contained aligned M T (eg. Fig. 13). However, in Figure 20, the nucleus was located within the lower groove and the M T O C on the adjacent ridge. The MTs were aligned within both grooves, yet neither contained the M T O C and only one of the grooves contained the nucleus. Thus the positions of the M T O C and nucleus did not appear to determine M T alignment within the grooves. By 6 hours and beyond, most cells were aligned within one groove and were monopolar and in those instances, the M T O C was typically located anteriorly to the nucleus (Fig. l i b ) .  v. Vimentin Intermediate Filaments and Kinesin In cells on grooved substrata, the time course of the codistribution of IPs (Fig. 22) and MTs was similar to that observed in cells spread on smooth surfaces. During the first 4-6 hours of spreading, IFs were distributed close to the nucleus and IPs did not codistribute with M T to the cell peripheries. However, even at these early times of spreading, IFs located within the grooves were aligned parallel to the direction of the grooves. In most areas of the cells spread on grooves, the distribution of kinesin staining was similar to the distribution of MTs.  3.  Time Lapse Observations on R grooves  For cells spreading on R grooves, radial spreading was not prevented by the grooves as cell peripheries extended laterally across several grooves and ridges. Cells gradually became bipolar and elongated in a direction parallel to the grooves and came to be aligned within one groove. As monopolar shapes developed, lateral extensions across the ridges became limited to leading edges.  154 Cells locomoting on the grooved surfaces appeared similar to cells moving on smooth surfaces, but their motion was directed along the grooves which resulted in a stronger directional persistence of locomotion and larger displacement over a given time relative to cells on a smooth surface (Damji, 1992).  C.  Summary  This experiment examined the temporal relationship between cytoskeletal alignment and overall cell orientation on R grooves. Using confocal microscopy, the distribution of cytoskeletal elements was determined in three dimensions and although individual living cells and their cytoskeletons could not be tracked over time, the data, considered at the population level, demonstrated consistent patterns of development. MTs were the first cytoskeletal element to become aligned to the direction of the grooves, followed by actin M F bundles, focal contacts and lasdy, IFs.  M T alignment occurred predominantly along the length and depth of the grooves and in particular, along the walls and floor of the grooves. Some MTs emerged from the M T O C in directions parallel to the direction of the grooves and these MTs were generally oriented along their entire length along the walls and grooves. Other MTs emerged from the M T O C in nonparallel directions to the grooves but these MTs appeared to become aligned within the grooves from the points of apparent contact with the walls of the grooves. Alignment of non-parallel MTs also occurred along the ridges, specifically along the ridge/groove edges but in general, alignment of MTs at the ridge level was observed later than alignment of MTs within the grooves. Alignment of MTs thus appeared to under local topographic control and was unrelated to the location of the M T O C .  Significantiy, development of actin patterns was also influenced by topographical features as actin condensations were observed along the ridge edges and later, aligned actin M F bundles were observed along the walls and in the grooves. However, poor agreement between aligned actin M F bundles and cell orientation was observed up until 6 hours, at which time cell spreading and  155 alignment were largely completed. Considered at the population level, the data do not support actin M F bundles as a principal determinant of cell alignment on R grooves. Instead, the data indicate that MTs may be the principal cytoskeletal element effecting alignment on R grooves.  156  Figure 3. Light micrographs of fibroblasts 24 hours after plating on smooth (a) and R-grooved (b) substrata. Thin lamellar and tail regions of cells stain less brighdy than the body of the cells. Cells on both smooth and grooved surfaces were monopolar but cells on the grooved surface were more spindle-shaped. Cells on grooves (b) were also aligned to the direction of the grooves (double-headed arrow in b) whereas cells on the smooth surface (a) demonstrated no preferred orientation.  157  Area, Smooth Area,  Grooves  Height, Smooth • -<>-•  Height, Grooves  10000H  Time,  minutes  Figure 4. Area and height of fibroblasts spread on smooth and R-grooved substrata are plotted against time. Error bars are not included but analysis of the data by two-way analysis of variance (ANOVA) indicated significant (P<0.001) effects of both time and surface. The interaction between time and surface was also significant (P<0.001).  158  Ell, Smooth Ell, Grooves Angle, Smooth Angle, Grooves  .-0--  h 60  H m  0.75 A  h 40  u  CO  u_  o LL  0.5 H  c  < c o CO  20  0.25 H  c O  OH  h 0  I  1  1  1  1  1  0  60  120  180  240  300  1—  360  Time, minutes Figure 5. Form factor Ell of fibroblasts spread on smooth and R-grooved surfaces is plotted against time. Analysis of the data by two-way analysis of variance (ANOVA) indicated significant (P<0.001) effects of both time and surface. The interaction between time and surface was also significant (P<0.001). Orientation angles (degrees) of fibroblasts spreading on smooth and Rgrooved substrata are plotted against time. Analysis of the data indicated significant (p<0.001) effects of both time and surface.  Figure 6 CLSM-generated images of polarized fibroblasts 24 hours after plating on smooth substrata.' A cell was double-stained for actin (a) and tubuhn (b), and a cell stoned for vinculin (c). Both cells were monopolar and the distribution of the cytoskeletal elements reflected the overall shape of the cells.  160  m  W  o ©  |  g©pap  <3  fiflSffi  Figure 7. CLSM-generated images of fibroblasts 1 (a, b) and 4 (c, d) hours after plating on smooth substrata. Cells stained for tubulin (a, c) and vimentin (b, d). Near the cell centers, the distribution of vimentin IFs was similar to the distribution of MTs. However, in the cell peripheries (arrows), the distributions of IFs and MTs differed.  (o]  1  —  Figure 8. CLSM-generated images of polarized fibroblasts 16 (a, b) and 24 (c-f) hours after plating on smooth substrata. Cells were double-stained for tubulin (a) and vimentin (b), tubulin (c) and kinesin (d), and vimentin (e) and kinesin (f). The distributions of MTs and kinesin (c, d) were similar throughout the cell, including the lamellae (arrowheads) but differed in some areas. Distributions of MTs and IFs (a, b) and IFs and kinesin (e, f) were similar but did not con-elate (an-ows) as closely as MTs and kinesin (c, d); particularly in parts of the lamellae. The nonidentical distributions of tubulin, vimentin and their respective kinesin patterns support the specificity and non-cross-reactivity of the respective antibodies.  162  Figure 9. CLSM-generated images comprising 7 optical sections, 1 Ltm apart (a) and 13 optical sections, 0.5 Ltm apart (f) of a fibroblast 20 minutes after plating on R grooves. The cell was located primarily within the groove (g) with lateral spreading onto the ridges (r). Orientation angle is 82°; Form Ell is 0.82. The cell was stained for actin (a-e) and tubulin (f-j). Only MTs showed alignment within the groove. Optical sections located approximately 3 Ltm above the level of the ridge (b, g); optical sections located just above the ridge (c, h); optical sections located just below the ridge (d, i); optical sections located at the bottom of the groove (e, j). Composite actin image (a) comprised of 7 optical sections: 1 (b), 3 (c), 5 (d), 7 (e) are each separated by a distance of 2 Ltm and a ruffled actin pattern was evident at all levels. Composite M T image (f) comprised of 13 optical sections: 2 (g), 6 (h), 10 (i) 13 (j). A distance of 2 Ltm separates optical sections 2 and 10; 1.5 Ltm separates optical sections 10 and 13. The M T O C (arrow) and nucleus (n) were evident over the groove. At the ridge level (g, h), MTs were radially distributed. MTs were aligned along the walls of the groove (i) and within the groove (j).  164  Figure 10. CLSM-generated images of fibroblast 40 minutes after plating on R grooves. The cell was stained for vinculin (focal contacts) and actin (not shown; the actin pattern included ruffles and a few circumferential fibers at the ridge level). Orientation angle is 26°; Form Ell is 0.69. Optical section at the ridge level (a) showed a radial pattern of focal contacts. Arrowheads indicate focal contacts on the ridge level continuing along the wall of the groove (g). Optical section in the groove (b), 2.95 \im below the optical section shown in (a). In die groove (b), the predominant orientation of the focal contacts was parallel to the grooves and the pattern was designated as aligned. Arrowheads indicate focal contacts on the wall (w) near the bottom of the groove.  165  166  Figure 11. CLSM-generated images each comprising 4 optical sections, 1 Ltm apart, of fibroblasts 4 hours after plating on R grooves. A cell stained for actin (a) was located primarily within one groove (g) with lateral spreading to the adjacent ridges (r). The orientation angle is 7°; Form Ell is 0.28. Aligned actin M F bundles were evident within the groove and along portions of the ridge. The processes at both ends of the cell demonstrated M F bundles oriented perpendicular to the grooves on the ridges and other M F bundles bridging over the groove. A second cell stained for tubulin (b) was located primarily within one groove with lateral spreading onto the ridges near the leading of the cell. The orientation angle is 2°; Form Ell is 0.18. The MTOC (arrow) was located anteriorly to the nucleus (n) and both were located in the groove.  167  Cells • • • •f^^ • • • Actin Filament Bundles a  —o--.  Focal  tm m  Microtubules  ^^BH  •  mm  Contacts  100-. 80•D  o> 60 H < •c <i) u  40H  °-  20H  1 0  1 60  1 120  1 180  1 240  1 300  1 360  Time, minutes  Figure 12. Percentages of aligned fibroblasts and percentages of fibroblasts demonstrating aligned M F bundles, aligned focal contacts and aligned MTs are plotted against time for fibroblasts spread on R grooves.  168  Figure 13. CLSM-generated images comprising 5 optical sections, 0.5 Ltm apart (a) and 11 optical sections, 0.5 Ltm apart (d) of a fibroblast 60 minutes after plating on R grooves. The cell spread over 2 ridges (r) and 3 grooves (g) with an orientation angle of 8°; Form E l l is 0.71. The cell was stained for actin (a-c) and tubulin (d-f). Only the MTs within the groove demonstrated alignment The arrowheads in (a) indicate portions of the cell within the upper and lower grooves where MTs in (d) were not detected. Optical sections at the ridge level (b, e); optical sections at the groove level (c, f). Composite actin image (a); optical sections 1(b) and 4 (c) are separated by a distance of 1.5 Jim. On the ridge level (b), circumferential fibers bridged over the grooves; in the groove (c), no aligned M F bundles were evident Composite M T image (d); optical sections 5 (e) arid 9 (f) are separated by a distance of 2 Ltm. The MTOC (arrows in e, f) was located over the groove and the nucleus (n) was visible to the left of the MTOC, in the same groove. On the ridge, MTs were radially distributed with some alignment (white arrowheads in d) along the edge of the upper ridge and along the edge of the lower ridge. Within the groove (f), MTs were aligned and the M T O C was evident within the groove.  169  170  Figure 14. CLSM-generated images comprising 6 optical sections, 1 nm apart (a) and 8 optical sections, 0.5 \im apart (c) of a fibroblast 100 minutes after pladng on R grooves. The cell spread over 2 grooves (g) and 3 ridges (r) with an orientation angle of 3°; Form Ell is 0.68. The cell was stained for actin (a, b) and tubuhn (c-g) and both elements demonstrated alignment within the grooves. Composite actin image (a) demonstrated circumferential fibers with bridging fibers over the grooves. A t the groove level (b), composite image of optical sections 5 and 6 demonstrated aligned M F bundles in the center of the upper groove and along the wall (w) of the lower groove. The M T O C was not evident in the composite M T image (c). Optical sections 2 (d), 4 (e), 6 (f) and 8 (g) are each separated by a distance of 1 um. At the ridge level (d, e), MTs demonstrated a radial orientation. In the groove (f, g), MTs were aligned along the wall of the groove as well as within the bottom of the groove.  171  172  Figure 15. CLSM-generated images of a fibroblast 20 minutes after plating on R grooves Cell stained for actuL Optical sections 1 (a), 2 (b), 3 (c) and 4 (d) are each separated by a distance of I Mm. Distinct M F bundles were not evident at either the ridge level (a) or in the groove (b-d) but condensations (arrows) of actin stain were evident along the ridge/groove edge and along the walls ot the groove.  173  Figure 16. Composite CLSM-generated image comprising 8 optical sections, 1.5 |im apart, of a fibroblast 120 minutes after plating on R grooves. Orientation angle is 70o; Form Ell is 0.94. The cell was stained for actin. Ruffles were evident at the cell perimeter; circumferential and bridging fibers were evident in a concentric pattern. Aligned M F bundles were evident on the ridge (r), along the walls and within the lower groove (g).  174  Figure 17. CLSM-generated images of aligned and monopolar fibroblasts 24 hours after plating on R grooves. Cells were stained for actin (a), vinculin (b) and tubulin (c). With the exception of the leading edges, the cytoskeletal elements reflected the overall alignment of the cells. A l l cells were aligned within a groove (g) although not all the focal contacts (arrow in b) within a groove were oriented parallel to the direction of the grooves.  175  •  IK •  Z = 1 pci  9  (?  ggg j • 5 -  ••*'  Figure 18. CLSM-generated images of a fibroblast 180 minutes after plating on R grooves. The orientation angle is 3°; Form Ell is 0.24. Composite vinculin image (a) comprising 9 optical sections, 0.5 |im apart showed aligned focal contacts on the ridge (r). In the groove (g), both parallel and perpendicular orientations were evident (small arrows). The material containing vinculin was also visible along the wall (w). The magnified imaged (b) of the area indicated by the large arrow in (a) is seen in optical section 3 (b), 1 Jim below the ridge level. The arrowheads indicate vinculin-containing structures at the junction of the ridge and wall of the groove.  176  Figure 19. CLSM-generated images of a fibroblast 40 minutes after plating on R grooves. Composite tubulin image (a) comprising 9 optical sections, 0.5 Ltm apart. Individual MTs were difficult to discern but collections or bundles of M T outgrowths (arrows) were evident along the walls (w) of the groove (g). Optical sections 2-9 (b-i) demonstrated that the M T outgrowths were not evident above (b) or at a level above the ridge (r)(c, d) but were evident at the top of the groove (e), throughout the depth of the groove (f-h) to the bottom of the groove (i).  177  Figure 20. CLSM-generated image comprising 10 optical sections, 0.5 Ltm apart, of a fibroblast 100 minutes after plating on R grooves. The cell spread over 2 grooves (g) and 3 ridges (r). Orientation angle is 43°; Form Ell is 0.80. Cell stained for tubulin. The MTOC (large arrow) was located on a ridge; the nucleus (n) was located in the lower groove as evident in other optical sections, but not shown. MTs were aligned in both grooves (small arrows). Some MTs at the ridge level emerged from the MTOC in nonparallel directions to the grooves and were dispersed radially. However, some of these MTs (white arrowheads) appealed to change their orientation from the point of apparent contact with the edges of the ridges.  178  Figure 21. CLSM-generated image comprising 10 optical sections, 0.5 Ltm apart, of a fibroblast 4 hours after plating on R grooves. The cell spread over 3 grooves (g) and 2 ridges (r). Cell stained for tubulin. Some MTs (arrowheads) emerged from the MTOC (arrow) parallel to the direction of the grooves and these MTs remained aligned within the groove along their lengths. Some MTs (asterisks) emerged from the MTOC in non-parallel directions to the grooves and appeared to change their orientation from the point of apparent contact with the walls of the grooves.  179  Figure 22. CLSM-generated images of fibroblasts 4 (a, b) and 6 (c, d) hours after plating on R grooves. Cells stained for tubulin (a, c) and vimentin (b. d). Vimentin IF and MTs were codistributed near the cell centers but not along the cell peripheries (arrows), including lamellae.  180 II. Behaviours of Fibroblasts and their Cytoskeletons in Response to Conflicting Topographic Cues The experiments described in this section were also designed to determine the first cytoskeletal element to become oriented to the topography during cell alignment. In these experiments, conflicting rather than uniform topographic cues were used and therefore it was also possible to observe cytoskeletal behaviours in response to grooves of similar size but with a different arrangement. As well, changes in cell morphology and in the distribution of cytoskeletal elements were examined in (i) fibroblasts that spread at or near groove intersections and (ii) in polarized fibroblasts that migrated along set of grooves towards the groove intersection, entered the intersection and negotiated the intersection to orient with the second set of perpendicular grooves. The observations of polarized cells encountering the groove intersections facilitated testing of the hypothesis that lamellae contain the machinery responsible for contact guidance (Brunette, 1988a; see Chapter 1, Section IV.C.).  The behaviours of cells travelling towards and encountering the groove intersection are presented first. Descriptions of the cytoskeletal patterns of cells spreading at or travelling towards and through the groove intersection follow. In these experiments, R and N grooves were used. Grooves leading into the intersection were designated " X " grooves; those grooves intersecting at right angles with the " X " grooves were designated as "Y" grooves.  A.  Changes in Cell Morphology as Migrating Cells Approached the Groove Intersections  Polarized cells travelling along the 30-Ltm-pitch R grooves exhibited a variety of behaviours as they approached and negotiated the groove intersection. Cell behaviours at the groove intersection were broadly categorized on a morphological basis; however, these categories were not meant to imply different mechanisms of turning.  181 1. A few cells approached the groove intersection but did not proceed into or through the intersection. Instead, these cells paused at the intersection, often after contact with the far wall of the " Y " groove. Some cells remained stationary and others retreated back along the same " X " groove.  2. Some cells entered a corner of the intersection and gradually turned their entire cell body around the corner of the " X " groove to enter the "Y" groove. Typically, these cells hugged one wall of the " X " groove and followed this wall around the corner into the intersection (Fig. 23).  3. As many as 80-90% of cells travelling towards the groove intersection extended leading lamellae or cell process ahead of their cell bodies (Figs. 24-28). In these instances, the leading lamellae or filopodia entered the intersection and "Y" groove ahead of the cell body which was aligned within an " X " groove some distance away from the intersection. As a result, the leading lamellae or process typically came to be aligned perpendicular to the cell body such that the leading edges formed either a "sideways L"- or "hook-" shape (Figs. 24, 26a) or a "sideways T"- shape (Figs. 27, 28) with respect to the cell bodies. "T" configurations occurred more frequently than "L" shapes. Near the intersection, portions of the cells often spread onto the ridges of both the " X " and " Y " grooves as the walls of the " X " and "Y" grooves did not appear to present absolute physical barriers to spreading onto the ridges (Figs. 25,26a) and as a result, alignment of the cells within the groove intersection was not absolute (Fig. 26b).  As the cells moved through the intersection, "L"-shaped leading edges advanced further along one direction of the " Y ' groove and the cell bodies gradually followed the leading edges into and through the intersection and into alignment with the "Y" groove. "T"-shaped leading edges eventually developed into "L"-shapes as one end of the "T" became the leading edge. Occasionally, leading edges spread along the " Y " grooves, turned back into a second " X " groove (Figs. 26a, 27), thus effecting a "U" turn. When both ends of the "T"-shaped leading edges were  182 active, the cells became very long and thin as they were stretched along the " Y " groove in two directions yet part of the cell remained aligned in the " X " groove (Fig. 28). Eventually, one end of the "T"-shaped leading edges predominated and the residual lamella and cell body followed the course of the new leading edge.  On 6-9-Ltm-pitch N grooves, cells travelling along the " X " grooves encountered a 25-|im-wide " Y " groove and then a second series of " X " grooves (Fig. 29). Some cells travelled through the intersection, without changing their direction and continued along the second series of " X " grooves. Other cells developed "T"-shaped leading edges at the junction of the " X " grooves with the wide " Y " groove (Fig. 29)  B.  Cytoskeletal Behaviours of Cells Spreading at Intersecting Grooves  Cytoskeletal patterns were similar between cells spreading at or near groove intersections and cells travelling along the grooves into the intersections. As well, cytoskeletal patterns were similar between cells encountering groove intersections on 30-|im-pitch R grooves and 6-9-Ltm-pitch N grooves. In all instances, cytoskeletal alignment appeared to be controlled locally by the topography of the substratum and included the following features:  1. Cell morphology did not always reflect cytoskeletal orientation, particularly within the groove. Cell boundaries often extended across one or more " X " grooves and ridges as well as into the " Y " groove and onto the " Y " ridge. However, neither M F bundles nor MTs were consistently located near the cell boundaries at either the ridge or groove levels (Figs. 30, 31, 32).  2. Alignment of actin M F bundles and MTs was different, particularly within the groove. MTs demonstrated more consistent and greater alignment than M F bundles to the " X " and " Y " grooves of the intersection. Typically, M F bundles were not adapted to or around the corners of the intersection. If M F bundles were aligned within the " X " and " Y " grooves of the intersection, the  183 bundles were separate and perpendicular to each other without apparent continuation or bending around the edges of the grooves (Figs. 29, 31). In contrast, MTs closely reflected the groove contours of the intersection and MTs appeared to exhibit considerable flexibility as they bent around the corners of the grooves at the groove intersection (Figs. 29-34). In fact, cells spreading at the intersections of 6-9-Lim-pitch grooves appeared to become trapped at the intersection because of the close adaptation of MTs to the grooves (Fig. 34). Typically, M T alignment with the " X " and " Y " grooves resulted in a "sideways T" configuration, similar to the shape assumed by leading edges extended into the intersection by cells travelling towards the intersection (Figs. 27, 28).  3. M T alignment occurred without a consistent relationship to the location of either the nucleus or microtubule-organizing centre (MTOC). MTs were aligned in one or more grooves in instances where the M T O C was located anterior to (Fig. 29), posterior to (Fig. 31) or beside (Figs. 32, 33) the nucleus and where the M T O C was located on a ridge (Figs. 29, 32, 33) or in a groove (Fig. 31).  4. MTs were aligned along both the length and depths of the grooves. Within the grooves MTs demonstrated two main patterns or distributions which implied certain M T behaviours:  i. Barrier Behaviour MTs aligned along the far wall of the "Y" groove (Figs. 31, 32). It appeared that if MTs were long enough and oriented in a direction so that they encountered the far wall of the " Y " groove, then M T alignment occurred at and continued from the point of apparent contact with the wall of the "Y" groove. I termed this pattern of M T orientation "barrier behaviour".  ii. Wall-hugging Behaviour MTs near the edges of the " X " grooves followed the walls of the " X " grooves into the intersection, around the corner and occasionally back into adjacent " X " grooves (Figs. 30, 31, 32). I termed  184 this pattern of M T orientation "wall-hugging behaviour". When these wall-hugging MTs reached the intersection, the wall-hugging MTs often did not maintain the "X-wall"-orientation they had displayed within the " X " groove. Instead, the wall-hugging MTs stayed close to the wall and bent around the corner to follow the wall into the intersection and sometimes, back into an adjacent "X" groove.  C.  Summary  The experiments in this section demonstrated that MTs in cells encountering conflicting topographic cues responded similarly in some aspects to MTs of cells presented with uniform topographies (Section I). That is, M T alignment to the conflicting topographies of intersecting grooves was greater and more consistent than alignment of actin M F bundles, and M T alignment preceded morphological alignment of the cells themselves. Moreover, MTs consistendy and predominandy aligned along the walls of the grooves, irrespective of whether they were parallel or intersecting sets of grooves. This alignment response of MTs also appeared to be independent of the location of the M T O C (Section I) as it appeared that M T alignment was controlled locally by topographic features of the substratum.  The experiments in this section also demonstrated similarities in cytoskeletal behaviours between cells spreading at or near groove intersections and polarized cells travelling into the intersections. These experiments supported the hypothesis that the leading lamellae of cells may contain the machinery responsible for contact guidance (Brunette, 1988a).  185  Figure 23. Image obtained by reflected Nomarski DIC optics of fibroblast on micromachined substratum comprised of two sets of R grooves (g) that intersect at right angles. Cell travelled along the "X" groove to approach the intersection with the "Y" groove. The cell adapted its bod^ closely to the wall and corner of the "X" groove to follow the wall into the intersection  186  Figure 24 Time-lapse sequence obtained by reflected Nomarski DIC optics of fibroblast on micromachined substratum comprised of two sets of R grooves that intersect at right angles Cell body arge arrow in a, b) was situated some distance from the intersection and the leading edge of the cell process (small arrow) which entered the intersection (a), encountered the far wall of the Y groove (b), and bent to align (asterisk) along the wall of the " Y " groove (b, c) Time given in  187  Figure 25. Image obtained by reflected Nomarski DIC optics of fibroblast on micromachined substratum comprised of two sets of R grooves (g) that intersect at right angles. Cell travelled along the "X" groove to enter the "Y" groove where the leading lamellae (small arrows) spread laterally along the " Y " groove to form a "sideways T" shape. A small portion of the leading edge spread onto the "Y" ridge (white arrowhead). The cell also extended a long cell process (large arrow) diagonally across the "Y" groove to ruffle along the wall and ridge (white arrowhead) of the " Y " groove.  188  Figure 26. Images obtained by reflected Nomarski DIC optics of fibroblasts on micromachined substratum comprised of two sets of R grooves (g) that intersect at right angles. (a) . Cell travelled along the " X " groove into the intersection where the leading edge spread into the " Y " groove in an "inverted L"-shape with respect to the cell body (large arrow). A small portion of the lamella spread onto the " Y " ridge (large arrowhead) but most of the lamella was located within the " Y " groove (asterisks). A portion of the leading edge spread back into an adjacent " X " groove (small arrowhead). (b) . Cell travelled along an " X " groove but as it approached the intersection, cell oriented diagonally across several ridges (large arrowheads) and grooves (small arrows). Thus, neither grooves nor ridges presented absolute banners to cells spreading across them, but such behaviour was infrequent.  189  Fig. 27. Time-lapse sequence obtained by reflected Nomarski DIC optics of fibroblast on micromachined substratum comprised of two sets of R grooves that intersect at right angles. Cell body (large arrow) was situated some distance from the intersection (a, b) but the leading lamellae (small arrows) spread onto the "X" ridges near the intersection and entered the intersection where they spread laterally to form a "sideways T"-shape with respect to the cell body. With increased time, the cell body approached the intersection and the leading edge spread laterally along the far wall of the " Y " groove without spreading onto the "Y" ridge (asterisks in c, d). A portion of the lamella spread back into an adjacent "X" groove (arrowhead). Time given in hours:minutes.  190  Figure 28. Image obtained by C L S M of fibroblast on micromachined substratum comprised of two sets of R grooves (g) that intersect at right angles. Cell travelled along " X " groove to enter intersection and spread along "Y" groove. Two leading edges (small arrows) formed "sideways T"-shape with respect to the cell body (large arrow).  191  Fig. 29. Composite CLSM-generated images comprising 6 optical sections each, 1 Ltm apart, of fibroblast that travelled along the 9-ixm-pitch " X " grooves (small arrows) towards the intersection with the wide " Y " groove (G). The cell was double-stained for actin (a) and tubulin (b). (a) . M F bundles (large arrows) were aligned along the lengths and depths of several of the narrow " X " grooves and along the floor of the " Y " groove. However, the " X " - and "Y"-oriented M F bundles were separate and the two sets of MFs did not appear to continue or bend around the edges of the grooves. (b) . MTs were aligned along the lengths and depths of several " X " grooves and appeared to follow the edges of the grooves, bend around the corners (arrowheads) and align around the corner and along the edge of the " Y " groove. The MTOC was located anterior to the nucleus (n) and both were located at the ridge level.  192  Figure 30. CLSM-generated images of fibroblast 60 minutes after plating near the intersection of two sets of R grooves. The cell was double-stained for the cell membrane (a) and tubulin (b). (a) . Thin cell processes stained less brighdy with the membrane stain but it was evident that the cell had spread into both " X " and " Y " grooves (g), onto both " X " ridges (r) as well as onto part of the " Y " ridge (white arrowheads). (b) . Composite M T image comprising 8 optical sections, 0.5 Ltm apart. MTs were evident on the both " X " ridges but MTs did not appear on the "Y" ridge (white arrowheads), although a cell process was present on the " Y " ridge in (a). MTs were aligned within and along the depths of both the " X " and " Y " grooves, resulting in a sideways "T" configuration. MTs bent around the corners (arrowheads) throughout the depth of the groove and appeared to turn back (small arrows) into the top and bottom " X " grooves so that MTs appeared to run in a parallel direction to the MTs in the middle " X " groove.  193  Figure 31. CLSM-generated images of fibroblast 180 minutes after plating near the intersection of two sets of R grooves. The cell was double-stained for actin (a, b) and tubulin (c, d). (a) . Composite actin image comprising 9 optical sections, 0.5 Ltm apart Circumferential MFs at the ridge (r) level with MFs bridging over the grooves (g) were evident. (b) . Composite actin image comprising optical sections 8 and 9 in the groove. Two separate and aligned M F bundles (large arrows) were present within and aligned along the floor of the "X" and " Y " grooves but there did not appear to be any continuation of the M F bundles around the edges of the grooves. (c) . Composite M T image comprising 9 optical sections, 0.5 Ltm apart. The M T O C (large arrow) was located posterior to the nucleus (n) and both were located in grooves. (d) . Composite M T image comprising optical sections 8 and 9 in the groove. MTs were aligned within and throughout the depths of both the " X " and "Y"grooves, resulting in a sideways "T" configuration. MTs appeared to follow the edges of the groove and bend around the corner of the grooves (arrowheads). Some MTs (asterisks) encountered the far wall of the " Y " groove and MTs appeared to change direction at the point of apparent contact with the wall and align along the wall of the " Y " groove. Other MTs (small arrows) in the top and bottom " X " grooves appeared to ran in parallel direction to the MTs in the middle "X" groove.  194  Figure 32. CLSM-generated images of fibroblast that travelled along the " X " groove, into the intersection of the " X " and "Y" R grooves. The cell was double-stained for the cell membrane (a) and tubulin (b-d). (a) . The cell appeared to have extended two leading edges into the intersection with the " Y " groove and onto the " Y " ridge. (b) Composite M T image comprising 10 optical sections, 0.5 pm apart. The M T O C (arrow) was located on an " X " ridge (r) and beside the nucleus (n) which was located in an " X " groove (g). (c) . Optical section number 2 located at the ridge level. MTs in the lower leading edge did not extend onto the "Y" ridge (white arrowhead) although the cell process was located on the "Y" ridge at this location (white arrowheads in a). (d) . Optical section number 8 located in the groove. MTs were aligned within and throughout the depths of both the " X " and " Y " grooves, resulting in a sideways "T" configuration. MTs appeared to bend around the comers (arrowheads). Some MTs (asterisks) appeared to encounter the far wall of the " Y " groove and appeared to bend and change direction to align with the wall at the point of apparent contact with the wall of the "Y" groove. Other MTs (small arrow) appeared to turn into the top " X " groove in parallel direction to the MTs in the middle "X" groove.  195  Figure 33. CLSM-generated images of fibroblast 3 hours after plating near the intersection of two sets of R grooves. The cell was double-stained for the cell membrane (a) and tubulin (b-d). (a) . The cell had spread into two " X " grooves (g) as well as into the " Y " groove. (b) Composite M T image comprising 4 optical sections, 1 um apart. The M T O C (medium arrow) was located on an " X " ridge (r) and beside the nucleus (n) which was located in an " X " groove (g). (c) . Optical section number 2 at the ridge (r) level. (d) . Optical section number 4 in the groove. MTs were aligned within and throughout the depths of both " X " grooves and in the " Y " groove. MTs appeared to bend around the corners (arrowheads) of the grooves and align with the "Y" groove.  196  Figure 34. CLSM-generated images of fibroblast 24 hours after plating at the" Y " groove which intersected with a set of 8- and 9-p.m-pitch " X " grooves. The cell was stained for tubuhn. (a) . Composite image comprising 14 optical sections, 1 |im apart. (b) . Composite image comprising optical sections 6-8 at the ridge level. (c) . Composite image comprising optical sections 12-14 at the groove level. The wide " Y " groove (G) separated two sets of " X " grooves (small arrows) of different pitch on each side of the " Y " groove. MTs adapted closely around the corners (asterisks) of the grooves and aligned within and throughout the depths of both " X " and " Y " grooves. It appeared the cell became trapped in the intersection due to the close adaptation of its cytoskeleton to the topography.  198 III.  The Effect of Colcemid on Topographic Guidance by R Grooves  The observations in Sections I and II suggested that MTs were involved in fibroblast orientation on R grooves. If MTs were the principal as well as sole cytoskeletal determinant of cell orientation, then contact guidance was not expected to occur in the absence of MTs (see Chapter 2). To directly test the possibility that MTs were the principal determinant of fibroblast orientation on Rgrooved substrata, colcemid was used to eliminate the influence of MTs. Colcemid interferes with the polymerization of tubulin and therefore filaments of MTs are not observed in the presence of colcemid (see Chapter 1, Section V.A.6.b.i. and Section V.A.I, and Chapter 3, Section I.B.2.a.).  Morphometric measurements in this section were based on a mean number of 80 cells examined for both the smooth and grooved surfaces, under control and colcemid conditions at 2, 6 and 24 hours. At 72 hours, sample size was reduced to a mean of 38 cells because proliferation of control cells and the development of large, multinucleated cells under colcemid conditions limited the number of isolated, contact-free cells.  A. 1.  Smooth Surfaces Cell Morphology and Quantitation  The appearance of cells cultured on smooth and grooved surfaces in the presence or absence of colcemid is illustrated in Figure 35.  With increased time of spreading, cells in the presence of colcemid increased in projected area (Fig. 36). At 72 hours, large multinucleated cells were evident and the mean projected area of colcemidtreated cells was significantly (P<0.05) greater than the projected area of controls. On smooth surfaces, cells in the presence of colcemid remained round over 72 hours (Fig. 37). Colcemid had little effect on orientation as the orientation angle for cells on the smooth surface did not differ greatly from 45 degrees, the angle expected for cells orienting randomly on a surface (Fig. 38).  199 2.  Cytoskeleton and Associated Elements  a. Actin and Vinculin Patterns and distribution of actin and vinculin staining were related to the overall degree of cell spreading on smooth surfaces. Actin and vinculin staining in colcemid-treated cells (Fig. 39) were comparable to and indistinguishable from that observed in control cells (see Fig. 6 in Chapter 4, Section I) with similar shapes. The development of the pattern of cytoskeletal elements was delayed in colcemid-treated cells but eventually became similar to that observed for cells under control conditions.  b. Microtubules As expected, the fibrillar network of tubulin staining was not observed in the presence of colcemid (Fig. 39a).  c. Vimentin At early times, a dense vimentin stain concentrated around the cell centres, was the predominant feature. By 24 hours, bundles of vimentin filaments coiled near or around the nucleus were evident in some cells, similar to the observations of Hollenbeck et al., (1989).  d. Kinesin The distribution of kinesin varied with time of spreading and cell shape. At early times, kinesin staining, like vimentin, was evident but could not be localized to specific structures. By 24 hours, however, kinesin staining codistributed with vimentin bundles. In addition, kinesin staining was observed in fibrillar patterns that codistributed with some of the prominent actin M F bundles in nonpolarized cells on smooth surfaces (Fig. 40)  3.  Time Lapse Observations on Smooth Surfaces  Cell behaviours on smooth titanium agreed with descriptions of cells on other smooth substrata  200 under colcemid or colchicine conditions (eg. Vasiliev et al., 1970; Goldman, 1971; see also Chapter 1, Section V.7.) in that cells attached and spread, but never attained the same degree of polarization as control cells. Moreover, cell shapes remained irregular and polygonal; well-defined leading edges were not established even after 72 hours as many cells continued to display ruffling around entire cell peripheries. Some cells extended, retracted and re-extended protrusions so that cell shape changed continuously, ranging from apolar to bipolar to brief periods of monopolar and reverting to apolar. Although most cells remained stationary, some cells were displaced short distances to adjacent locations on the substratum but this behaviour was clearly different from the locomotion observed for cells under control conditions.  B. 1.  R Grooves Cell Morphometry and Quantitation  The projected area (Fig. 36) of colcemid-treated cells was similar on grooved and smooth surfaces and in both instances, colcemid-treated cells displayed significantiy (<0.05) greater values in projected area than controls.  On grooves, cells in the presence of colcemid were significantiy more elliptical (less round) than cells spread on smooth surfaces under control or colcemid conditions (Fig. 37). However, cells on grooves in the presence of colcemid required 24-72 hours before they displayed comparable shape changes to controls spread on grooves. As well, cells spread on grooves in the presence of colcemid demonstrated significant alignment (P<0.05) along the axis of the grooves by 24 hours (Fig. 38).  2.  Cytoskeleton and Associated Elements  a. Actin and Vinculin Patterns and distributions of actin and vinculin staining were related to the overall degree of cell alignment and polarization. The development of actin patterns was delayed in colcemid-treated  201 cells (Fig. 41) but eventually became similar to that observed for cells under control conditions (see Fig. 17 in Chapter 4, Section I).  On grooved surfaces, alignment of M F bundles and focal contacts preceded alignment of the cells as a whole under either control or colcemid conditions. At early times, cells spreading on grooves in the presence of colcemid developed narrow finger-like projections along the groove/ridge edge (Fig. 42). These projections stained for vinculin and actin condensations, actin ruffles and M F bundles. With increased time of spreading, aligned M F bundles were first evident along the groove/ridge edge. Some cells also displayed short M F bundles with oblique or nearperpendicular orientations within the grooves but, these orientations were observed only during the early phases of spreading.  b. Microtubules As expected, the fibrillar network of tubulin was not observed in the presence of colcemid (Figs. 41a, 43d).  c. Vimentin and Kinesin In cells on grooves, the vimentin and kinesin (Figs. 43a, b, c; 44; 45b, c) patterns were similar to those observed in cells spread on smooth surfaces. However, in well-polarized cells aligned on grooves, accumulation of kinesin was often evident within the cell protrusions and leading edges characteristic of ruffling borders (Fig. 45).  3.  Time Lapse Observations on R grooves  The spreading and polarization behaviours of colcemid-treated cells on R grooves were similar to those of cells spreading on grooves under control conditions (see Chapter 4, Section I.B.3.). However, significantly more time was required for cells under colcemid conditions to align with the direction of the grooves and to develop monopolar morphologies. On the grooved surfaces,  202 cells in the presence of colcemid (Fig. 46) exhibited directed locomotion with ruffling and protrusions limited to the leading edge of polarized cells, like cells moving under control conditions and directed locomotion along the grooves for considerable distances was observed. In those instances where the cells exited from the grooves onto a smooth portion of the substratum, the previously monopolar, directionally-locomoting cells became apolar and exhibited the behaviour already described for cells on smooth surfaces in the presence of colcemid. Moreover, when the apparently-random protrusions of a cell spreading on the smooth surface contacted the grooved portion of the substratum, the cell became monopolar and travelled along the grooves (Fig. 47).  C.  Summary  Although delayed, colcemid-treated cells aligned with the R grooves and their focal contacts and actin M F reflected this alignment. Furthermore, these bipolar aligned cells polarized to monopolar cells and migrated on the grooved substrata, behaviours that were not accomplished on smooth control surfaces. In the colcemid-treated cultures, kinesin codistributed with some prominent actin filaments and was distributed within some protrusions. These observations might indicate that association of kinesin, either directly or indirectly, with an aligned network of actin filaments may be involved in effecting polarization of ahgned bipolar cells spread on R grooves in the presence of colcemid. However, the grooved topography was requisite for organizing and maintaining the ahgned actin fdament framework and in essence, the topography could compensate for the role of MTs. I have coined the term "topographic compensation" to describe the process whereby cells deficient in M T align, polarize and migrate along micromachined grooves.  203  Figure 35. Light micrographs of fibroblasts spread on smooth (a, c) and R-grooved (b, d) substrata in the absence (a, b) and presence of 1.0 [ig/ml colcemid (c, d). The double arrow shows the direction of the grooves. Cells were fixed at 24 hours. Thin lamellar and tail regions (a, b) stain less brightly. Control cells have polarized but show no particular orientation on smooth (a) surfaces, whereas cells on grooves (b) show alignment to the direction of the grooves. Under colcemid conditions, cells on smooth surfaces (c) failed to polarize but cells on grooves (d) ahgned to the direction of the grooves and some have also polarized.  204  10000H  7500 A  —0—  Smooth, Control  — • —  Grooves, Control  --©--  Smooth, Colcemid  --M--  Grooves, Colcemid  T  CM  E it  5000 A  CO CD  2500 A  OA  ~T~  0  I  i  24  48  72  Time, hours  Figure 36. Projected area plotted against time for fibroblasts spread on smooth and R-grooved substrata in the absence or presence of 1.0 Ltg/ml colcemid (error bars, S.E.M.). At 72 hours, cells spread in the presence of colcemid demonstrated significantly (P<0.001) greater values in projected area than control.  205  —•—  Smooth, Control Grooves, Control  --U--  Smooth, Colcemid Grooves, Colcemid  —O— H  ID  E k-  0.5 H  o  OH "T~  0  24  i  48  —r—  72  Time, hours  Figure 37. Form Ell plotted against time for fibroblasts spreading on smooth and R-grooved substrata in the absence or presence of 1.0 p.g/ml colcemid (Error bars S.E.M.). Form Ell, a ratio of the minor and major axes of a cell, is an indicator of cell shape in which a value of 1.0 corresponds to a circular cell and values less than one indicate elliptical or spindle-shaped cells. Cells on smooth surfaces in the presence of colcemid remained circular whereas cells in the absence of colcemid became ellipsoid. Cells on grooves in both the presence and absence of colcemid became very spindle-shaped but in the presence of colcemid, cells required more time to attain a comparable shape.  206  Time, hours  Figure 38. Orientation angle plotted against time for fibroblasts spreading on smooth and Rgrooved substrata in the absence or presence of 1.0 Ltg/ml colcemid (Error bars S.E.M.). Cells were considered aligned when their orientation angle measured less than 10 degrees to the direction of the grooves or, for smooth surfaces, an arbitrarily-selected line. In the absence of any orienting influence, cells on smooth surfaces in the presence or absence of colcemid failed to align. On grooves, cells in the presence and absence of colcemid aligned with the grooves, but in the presence of colcemid, cells required more time to align.  207  Figure 39. CLSM-generated images of apolar fibroblasts 24 hours after plating on smooth substrata in the presence of 1.0 Lig/ml colcemid. Cell double-stained for tubulin (a) and actin (b), cell stained for vinculin (c). Tubulin staining showed that the vast majority of the M T system had been disrupted with only a small cluster of filaments evident near the nucleus. The actin and vinculin patterns reflected the shape of the cells and were comparable to those observed in cells with similar shapes spreading on smooth surfaces in the absence of colcemid.  208  Figure 40. CLSM-generated images of an apolar fibroblast 6 hours after plating on smooth substrata in the presence of 1.0 ug/ml colcemid, cell double-stained for actin (a) and kinesin (b). The airowhead illustrates a concentration of kinesin in one protrusion. Arrows illustrate colocalization of kinesin fibrils with some prominent actin M F bundles. The interior of the cell stained heavily with kinesin but no specific pattern was evident; this distribution was commonly observed during early times of spreading.  209  Figure 41. CLSM-generated images of aligned and monopolar fibroblasts 24 hours after plating on R grooves in the presence of 1.0 (ig/ml colcemid. Both cells are aligned within a groove (g). Cell double-stained for tubulin (a) and actin (b); cell stained for vinculin (c). Tubulin stain is diffuse and without structure. With the exception of the leading edge, actin and vinculin distributions reflected the overall alignment of the cell. Actin and vinculin patterns were comparable to those observed in cells spreading on grooves in the absence of colcemid.  210  Figure 42. CLSM-generated images of two nonaligned, apolar fibroblasts 2 hours after plating on R grooves in the presence of 1.0 p.g/ml colcemid. At this early time, the majority of cells were not aligned or polarized. Both cells were stained for actin. Arrowheads indicate finger-like processes along the groove/ridge edge. The projections were associated with ruffles (a) or aligned M F bundles (b). Small arrow in (b) illustrates oblique and almost perpendicular orientations of actin filament bundles within the groove (g). Ahgned M F bundles (large arrows) were evident along the groove/ridge edge and were observed prior to alignment of the cell as a whole (b). Both cells were spread laterally across the groove and ridge (r) with Form Ell values of .83 (a) and .92 (b) and orientation angles of 86 degrees (a) and 71 degrees (b).  Figure 43. CLSM-generated images of 3 fibroblasts 24 hours after plating on R-grooves (a-d) and smooth substrata (e, f) surfaces in the presence of 1.0 Hg/ml colcemid. The first cell was monopolar and aligned within a groove (g), double-stained for vimentin (a) and kinesin (b). Arrowheads indicate colocalization of vimentin (a) and kinesin (b) coils posterior to the nucleus (n). Another cell aligned within a groove (g) and polarized, double-stained for kinesin (c) and tubulin (d). Kinesin distribution is in tight coils (c); tubulin stain was diffuse without filamentous structure (d). A third cell is apolar and double-stained for kinesin (e) and tubulin (f). Kinesin distribution was in bundles (e), tubulin stain was diffuse without filamentous structure (f).  212  s-V-  ^  x  -  ^ 7  -  ;  g  g  Figure 44. CLSM-generated images of a fibroblast 16 hours after plating on R grooves in the presence of 1.0 pg/ml colcemid. Cell was aligned within a groove (g) and double-stained for actin (a, b) and kinesin (c, d). Frames (a) and (c) are composite images each comprising four optical sections, 1.0 pm apart; (b) and (d) are each composite images of optical sections 3 and 4 within the groove. Arrowheads illustrate kinesin accumulation at one end of the cell in the composite image (c) although the cell is not clearly polarized as a monopolar cell. Arrows illustrate colocalization of kinesin fibrils (d) with prominent actin M F bundles (b) within the groove.  213  Figure 45. CLSM-generated images of two aligned and monopolar fibroblasts 16 (a, b) and 24 (c) hours after plating on R grooves in the presence of 1.0 Ltg/ml colcemid. One cell doublestained for actin (a) and kinesin (b). The arrowheads illustrate colocalization of kinesin fibrils with some actin M F bundles. Arrows show kinesin accumulation in the lateral protrusions. A second cell stained for kinesin (c). Arrows illustrate kinesin accumulation in the lateral ruffling borders of the leading edge. Kinesin stain was also colocalized with vimentin bundles that were tightly coiled anterior to the nucleus (n).  214  Figure 46. Time-lapse sequence of fibroblasts spreading in the presence of 1.0 P-g/ml colcemid on R grooves. The light and dark shading is due to reflected Nomarski DIC optics which were used to visualize cells on the opaque and reflective titanium surfaces. This sequence starts at 20 hours (frame a) after cells were seeded; time given in hoursrminutes. The sprocket hole in the upper right corner of each frame is a stable reference point. The double arrow in frame (a) shows the direction of the grooves. Several cell behaviours are evident. One cell (large arrow) travelled unimpeded along the grooves for several cell lengths. Ruffling was limited to the leading edge of this aligned, monopolar cell (eg. frame e, f, g). A second cell (small arrow) is visible in frame (a) but did not locomote until frames (f-h). Some cells travelled short distances but collided with other cells; arrowheads indicate 2 pairs of colliding cells. Some cells were aligned and polarized but did not locomote (double asterisks) and one cell failed to do either (single asterisk).  216  Figure 47. Time-lapse sequence of a fibroblast spreading in the presence of 1.0 Ltg/ml colcemid was filmed using reflected Nomarski DIC optics. The cell initially spread on the smooth portion of the substratum and was observed for 25 hours. During this time the cell alternatively ruffled and extended protrusions but never polarized or translocated. This sequence (a-j) illustrates changes in the cell's shape and behaviour after it extended a protrusion onto the R grooves and exhibited contact guidance. The asterisk is a fixed reference point at the junction of the smooth and grooved surface. The double arrow in frame (a) shows the direction of the grooves. Arrowheads indicate the cell. Time given in hoursrminutes.  0:00  V. 3 8  0:4  0:24  0:56  1:19  1:39  2:33  3:34  3:44  218 IV.  The Effects of Cytochalasin B and Cytochalasin B/Colcemid on Topographic Guidance by R Grooves  The experiments in Sections I, II and HI illustrated that although MTs were involved in contact guidance, MTs were not the sole determinant of contact guidance (see also Chapter 2). Another approach to testing the identity or candidacy of the principal determinant of cell alignment on grooves was to examine the behaviours of the cytoskeletal components independently. That is, to observe the M T system independently from the influence of MFs and to observe the actin system independently from the influence from MTs (described in Chapter 4, Section III). In addition, the alignment behaviours of cells lacking M F bundles would directly test the role of M F bundles in topographic guidance. If M T were indeed the principal determinant of cell orientation, I predicted that MTs, in the absence of M F bundles, would demonstrate comparable alignment to the grooves, over a comparable time course, to control cells with an intact cytoskeleton. However, if actin M F bundles were requisite for cell orientation on grooves, then alignment should be inferior to the alignment of control cells with an intact cytoskeleton. Thus, both predictions were tested simultaneously by observing cell and cytoskeletal behaviours in the presence of C B . Furthermore, I predicted that cells lacking both MTs and M F bundles, would fail to align on grooved surfaces and this prediction was tested by observing cells in the combined presence of C B and colcemid (CB/C; see Chapter 3, Section I.B.2.C.).  Morphometric measurements in this section were based on a mean number of 62 cells examined for each of the smooth and R-grooved surfaces, under control, CB and C B / C conditions at 24 hours. The proportions of cells exhibiting arborized processes or M F bundles were based on a minimum of 100 cells per concentration of CB or CB/C at 24 hours. The proportions of cells exhibiting these structures were expressed as percentages.  As noted above, C B (see Chapter 1, Section V.C.7.b. and Chapter 3, Section I.B.2.b.) was used to disrupt the actin M F system so that the behaviours of MTs could be examined in the absence of  219 the influence of M F bundles. As well, it was of interest to determine the role of M F bundles in cell alignment on grooved substrata because one major theory (Dunn and Brown, 1986) of contact guidance is based on stress fibers (actin M F bundles) and topographic compensation of MTdeficient cells appears to involve the actin system (see Chapter 4, Section III.). In fibroblasts, cytochalasin does not depolymerize actin filaments; rather, cytochalasin appears to disrupt the supramolecular organization of actin filaments because filaments persist albeit in dense focal accumulations (Schliwa, 1982) that stain with fluorescently-labelled phallotoxins (see Chapter 1, Section V . C . 7.a.). Because the effects of cytochalasin can vary with the concentration of cytochalasin used, it was necessary to determine the effects of different C B concentrations upon actin M F bundles and upon cell morphology and behaviour of human gingival fibroblasts. The results of the titration experiments with CB are presented below.  I also predicted that cells lacking both M T and M F would fail to align on grooved substrata and therefore, the combined presence of CB and colcemid was used to disrupt both filament systems. Again, a dose/response titration was required to determine the appropriate concentration of CB to be used in the presence of 1.0 |ig/ml colcemid as it has been previously reported that MT-inhibitors can increase filamentous actin and its organization (see Chapter 1, Section V.A.5.C.). The results of the titration experiments with CB/C are presented below.  Only data for cells spread in the presence of CB or CB/C are presented in order to compare these results to earlier reports of cells in the presence of CB (Weber et al., 1976; Bliokh et a l , 1980; Domnina et al., 1982) or CB/C (Bliokh et al., 1980). However, like Bliokh et al., (1980), I also observed that cells in the presence of C D or CD/C showed the same responses as cells in the presence of C B or CB/C although lower concentrations of C D were required (Yahara et al., 1982).  A.  Time Lapse Observations and Morphometry  Figures 48 and 49 illustrate the appearance of cells spread on smooth surfaces in the presence of  220 C B (48a-e) or C B / C (49a-e), respectively and may be compared to cells in Figure 3a (Chapter 4, Section I) which illustrate the appearance of cells spread on smooth surfaces under control conditions.  1.  Smooth Surfaces  a. C B At the lowest C B concentration used (0.3u.g/ml), round or polygonal apolar cells predominated as fewer polarized and directionally locomoting cells, relative to control conditions, were evident. At 0.5 pLg/ml C B , the majority of cells had ceased ruffling and locomotion and most cells were still round or polygonal and apolar. However, a few cells developed arborized shapes as they extended a variable number of processes with no preferred orientation, but neither ruffling nor motility accompanied extension of the processes. At 1.0 pg/ml CB, no cells displayed ruffling or motility and cells were apolar and either polygonal or arborized in shape.  b. C B / C In the presence of < 1.0 pg/ml CB in combination with colcemid, cells spread radially with intermittent and variable degrees of ruffling. However, cells remained apolar and round and locomotion was not observed. Arborized processes such as those described for CB-treated cells did not develop in the presence of CB/C, in the range of C B concentrations examined.  Although cells attached and spread in the presence of either CB or CB/C on both smooth and grooved surfaces, at 24 hours the projected areas of control, C B - or CB/C-treated cells were comparable only at concentrations of CB<1.5 (ig/ml. As C B concentrations increased above 1.5 pg/ml, alone or in combination with colcemid, cells were more round (Form Ell) and projected areas decreased in the range of 10-60%, depending upon the concentration of C B .  On smooth surfaces, cells in the presence of CB or CB/C were significantly (P<0.05) more round  221 and less spindle-shaped than controls, and as the concentration of C B increased, cells became increasingly rounder (Form Ell, Fig. 50). On smooth surfaces, cells under all treatment conditions did not align significantly and their orientation angle (Fig. 51) did not differ greatly from 45 degrees, the angle expected for cells orienting randomly on a surface which lacks orienting influences.  2.  R Grooves  Figures 48 and 49 illustrate the appearance of cells spread on grooved surfaces in the presence of C B (48f-j) or C B / C (49f-j), respectively. (Compare to Fig. 3b in Section I which illustrates the appearance of cells spread on grooved surfaces under control conditions). In contrast to cells spread on smooth surfaces, cells spread on grooves had more elliptical shapes (Fig. 50) and were aligned to the direction of the grooves (Fig. 51).  a. C B On grooves, the effects of CB upon cell ruffling, motility and the development of arborized processes (Fig. 52) were similar to those described for cells on smooth surfaces. However, in contrast to apolar, round cells on smooth surfaces, cells spread on grooves were aligned with the grooves. On grooves, cells predominantly extended processes in directions parallel to the grooves. However, some processes which were not aligned with the grooves at the point of process origin or emergence from a central site, became aligned with the grooves from the point at which they contacted the groove. The probability of alignment behaviour was related to the length of the processes as short processes, which never encountered grooves did not demonstrate alignment.  b. C B / C In contrast to the polygonal apolar cells observed on smooth surfaces, cells spread on grooves in the presence of CB/C (CB<1.0 Ltg/ml) aligned with the direction of the grooves and polarized and travelled along the grooves. However, CB/C-treated cells required significantly more time to  222 accomplish these behaviours than cells under control conditions. The proportion of cells displaying these behaviours decreased as the concentration of CB/C increased from 0.3 to 1.0 p.g/ml C B , but, for any concentration of CB (<1.0 p,g/ml), a greater number of cells exhibited ruffling and directional motility along the grooves in the presence of CB/C than did cells in the presence of the same concentration of CB alone. At 1.0 pg/ml C B in the presence of colcemid, a few aligned monopolar cells still travelled along the grooves, although the majority of the cells had ceased ruffling and locomotion and remained stationery as aligned, bipolar cells. At higher CB/C (CB>1.5 p,g/ml) cells were round and apolar.  On grooves, cells in the presence of CB or CB/C were less spindle-shaped and less aligned than cells under control conditions, but were significantly (P<0.05) more elliptical and more ahgned than cells on smooth surfaces under the same treatment conditions (CB<1.5 pg/ml)(Figs. 50, 51). At C B concentrations above 2.0 pg/ml C B , alone or in combination with colcemid, cells on both smooth and grooved surfaces were round.  On grooves at 2 hours after plating, cells under control or C B conditions (CB<2.0 pg/ml) were comparable in shape and alignment and differences in shape and alignment between cells spread on grooved and smooth surfaces were evident at this time. In contrast, at 2 hours after plating in the presence of C B / C , cells were round and not aligned with the grooves, and 24 hours were required before cell shapes were comparable to controls on grooves and for the majority of cells to be aligned (CB<1.5 p.g/ml)(Fig. 52). Thus, ahgnment and shape changes on grooved surfaces under control, C B and CB/C conditions were time dependent processes and considerably more time was required by CB/C-treated cells to accomplish similar changes in shape and ahgnment.  B.  Cytoskeleton and Associated Elements  The patterns and distribution of M T in CB-treated cells were investigated because it was hypothesized that if MTs were the primary or principal cytoskeletal determinant of cell orientation  223 on grooved substrata, then MTs in control cells spread on R grooves should behave similarly to MTs in CB-treated cells lacking M F bundles, spread on R grooves. The distribution of IFs and kinesin were investigated because these components codistributed with MTs in control cells and would be expected to do so in CB-treated cells.  The patterns and distributions of MTs, IFs and kinesin were investigated in CB/C-treated cells to confirm that their distributions were similar to the distributions observed in colcemid-treated cells (see Section HI. above). That is, disruption of M T filaments and collapse of IFs to the perinuclear regions were expected as a result of the action of colcemid; the combined presence of CB and colcemid was not expected to alter the effect of colcemid on MTs or IFs.  The patterns and distributions of actin in CB- and CB/C-treated cells were investigated to determine which concentrations of CB or CB/C resulted in the disappearance of actin M F bundles or stress fibers in human gingival fibroblasts. It was important to confirm the disappearance of M F bundles because the purpose of the experiment was to observe the orientation responses of gingival fibroblasts and their MTs in the absence of M F bundles.  1.  CB-treated  a. M T . Vimentin and Kinesin At all concentrations of C B , the distributions of MTs and IFs were similar and kinesin staining was distributed with these elements (Fig. 53). MTs and IFs were present along the lengths of the processes of the arborized cells and these filaments conformed very closely with cell shape. In general, as the concentration of CB increased, the degree of lateral packing or bundling of M T and IF filaments increased, and the overall length of filaments and bundles decreased. Bundles of MTs radiating from a common central site or MTOC were observed.  On smooth surfaces, in media containing CB, M T and IF bundles radiated from a central site with  224 no preferred orientation (Figs. 53,54). On grooves in media containing C B , MT/IF bundles also radiated from a central site and several grooves and ridges were spanned (Figs. 55, 56).  Based on observations of optical sections obtained by confocal microscopy (eg. Fig. 55), the responses of MTs to the topography were consistent at all concentrations of C B examined (Fig. 56) and appeared to be related to four factors:  1. The vertical dimension of the origin or initial emergence of the MTs, relative to the height of the ridges and depth of the grooves (3 Ltm). If MTs originated above the level of the ridge, M T orientation did not appear to be affected by the topography until the MTs contacted either a ridge or groove. That is, if the MTs did not encounter either a ridge or groove, then M T orientation was not affected.  2. The initial direction of M T origin from the MTOC. Some MTs originated at either the ridge or groove level in a direction parallel (P) to the direction of the grooves; these MTs are described as parallel (P) MT. P-MTs were prominent along the ridge/groove edges and along the length and depth of the walls of the grooves (Figs. 55, 56).  3. The length of the MTs, relative to the width the of grooves (15 Ltm) and ridges (15 fim). Those MTs that did not originate in parallel directions to the grooves are described as non-parallel (NP) MTs. If NP-MTs originated at the ridge or groove level, and if MTs were of sufficient length to encounter either a ridge edge or wall of a groove, then some NP-MTs appeared to become diverted from their NP orientation into a tangential course to run along either the ridge edge or wall. Thus, alignment of NP-MTs appeared to occur at and continue from the point of apparent contact with either the wall of the groove or the ridge/groove edge as NP-MTs appeared to bend and change direction to align along the wall or edge of the ridge (Fig. 55,56). These M T distributions implied certain M T behaviours (see also Chapter 4, Section I.B.2.b.iii. and Chapter 4, Section U.B.):  225 (i) . Barrier Behaviour Some N P - M T displayed what I termed "barrier behaviour" in which the alignment of NP-MT occurred predominantly along the length and depth of the walls of the grooves.  (ii) . Ridge-edge Behaviour Some N P - M T displayed what I termed ridge-edge behaviour" in which ahgnment of NP-MT occurred along the ridge/groove edges.  Exceptions to barrier and ridge-edge behaviours occurred because some NP-MTs that encountered the wall of a groove appeared to stop or end at the wall barrier. As well, when some NP-MTs encountered either a ridge edge or wall, they continued across that topographic feature to the other side of the feature. NP-MTs that arrived at the ridge edge or wall gave the impression that their behaviour may be to be related to a fourth factor,  4. the angle of contact between the NP-MTs and the wall of the groove or edge of the ridge. It appeared that the smaller the angle was, under which the NP-MTs approached the wall or ridge edge, the more likely the NP-MTs were to be deflected in a tangential direction along that feature.  (iii) . Wall-hugging Behaviour Particularly at larger angles of contact, some NP-MTs were not diverted by either the ridge edge or wall; instead, these NP-MTs retained their NP-orientation. For example, some NP-MTs crossed over the ridge edge, bent down into the grooves, descending along the wall of the groove onto and across the floor of the groove and even up along the far wall of the groove, onto another ridge (Fig. 55). I termed this pattern of M T orientation "wall-hugging" behaviour (see also Chapter 4, Section II.B.). Wall-hugging behaviour was also evident in CB-treated cells, spread at or near the intersection of two sets of R grooves (Figs. 57, 58), in which MTs were closely adapted to the walls of the grooves and intersection and followed the walls around the corners of the intersection.  226 b. Actin The pattern of actin-filament staining changed with the concentration of C B . As the concentration of C B increased, the proportion of cells with discernible actin M F bundles decreased and these proportions were comparable in cells on smooth and grooved surfaces (Fig. 52).  In cells spread on smooth surfaces at low concentrations of C B , short, needle-like actin MFs were interspersed on a diffuse background of small brightly-staining actin clumps. Actin edge-bundles (AEB) (Zand and Albrecht-Buehler, 1989) were prominent along concave edges of non-arborized, polygonal cells. At higher concentrations of CB, arborized processes developed and A E B and small actin MFs disappeared. Instead, actin aggregated in star-like clumps along the processes and accumulated in the distal ends of some processes.  In cells spread on grooves, the actin distribution followed a similar pattern but shorter actin MFs and A E B were also observed along the walls of the grooves and along ridge/groove edges (Fig. 59). In arborized cells spread on grooves, actin clumps distributed along the processes which were predominantly aligned with the grooves (Fig. 59).  2. CB/C-treated a. M T . Vimentin and Kinesin The distribution of tubulin, vimentin and kinesin was similar to that observed in the presence of colcemid alone (see Chapter 4, Section III.A.2.). Tubulin staining was diffuse without evidence of a fibrillar network (Fig. 60); vimentin staining appeared as a dense stain without specific structure or as coils of filaments near the nucleus. Kinesin staining codistributed most consistently with vimentin.  b. Actin The pattern of actin-positive staining changed as the concentration of C B in combination with  227 colcemid increased (Fig. 52). In comparison to cells in the presence of the same concentration of C B (<1.5 pg/ml) alone, the proportion of cells with actin M F bundles was greater in the presence of C B / C than in the presence of the same concentration of CB alone (Fig. 52). A E B were not prominent in cells spread on either smooth or grooved surfaces in the presence of C B / C . In cells spread on smooth surfaces, bands of circumferential fibers along round cell perimeters were common, along with diffuse clumps of actin-positive material.  Those cells spread on grooves (CB<1.0 pg/ml) that were aligned either as mono or bipolar cells generally contained aligned actin M F bundles in addition to actin clumps. However, some aligned and distinctly monopolar cells exhibited no actin M F bundles (Fig. 60). At >2.0 pg/ml C B in the presence of colcemid, actin M F bundles were no longer detected.  C.  Summary  In the presence of different and increased concentrations of CB, the mot