Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Ecm-cell adhesion-dependent control of cancer progression genes Christian, Sonja 2016

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

Item Metadata

Download

Media
24-ubc_2016_september_christian_sonja.pdf [ 16.16MB ]
Metadata
JSON: 24-1.0306937.json
JSON-LD: 24-1.0306937-ld.json
RDF/XML (Pretty): 24-1.0306937-rdf.xml
RDF/JSON: 24-1.0306937-rdf.json
Turtle: 24-1.0306937-turtle.txt
N-Triples: 24-1.0306937-rdf-ntriples.txt
Original Record: 24-1.0306937-source.json
Full Text
24-1.0306937-fulltext.txt
Citation
24-1.0306937.ris

Full Text

ECM-CELL ADHESION-DEPENDENT CONTROL OF CANCER  PROGRESSION GENES   by  Sonja Christian  Dipl. Bioinformatic, Johann Wolfgang Goethe University, 2006  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Microbiology & Immunology)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  July 2016  © Sonja Christian, 2016   ii  Abstract  Interactions with the extracellular matrix (ECM) are critical for tumor cell survival and dissemination. Cell-ECM interactions are mediated primarily by integrins, cell surface receptors that nucleate the formation of adhesomes. Adhesome complexes contain signaling proteins as well as proteins that link integrins to the cytoskeleton, thereby transducing extracellular forces into the cell. Both altered ECM composition and altered adhesome signaling can contribute to cancer progression. In this thesis I tested the hypothesis that ECM-integrin interactions drive cancer progression and that this depends on the adhesome proteins FAK and talin, and on mechanobiological tension. In chapter 3, I show that in B16F1 melanoma cells, the expression of three cancer signature genes, Cyr61, MUC18 and TRPM1, is strongly regulated by cell-ECM adhesion, independently of cytoskeletal tension. In chapter 4, I used global transcriptome profiling to compare gene expression changes caused by increased ECM ligand density versus cytoskeletal tension. These two perturbations regulated genes that belong to distinct pathways. Increasing ECM ligand density upregulated genes that are associated with adhesion, migration and ECM remodelling pathways, as well as the Hippo pathway, whereas applying mechanical stretch to the cells upregulated genes associated with metabolic pathways and the HIF-1 pathway.  In chapter 5, I show that the regulation of the cancer signature genes is dependent on the adhesome proteins talin and FAK. Consistent with a role for talin in adhesome signaling, loss of talin had the same effect on MUC18 and TRPM1 mRNA levels as forcing B16F1 cells into suspension. Knocking down FAK however, regulated Cyr61 and MUC18 differently than knocking down talin.   iii  In chapter 6, I show that the two isoforms of talin, talin 1 and talin 2, differentially regulate expression of the three cancer signature genes and have different effects on B16F1 spreading, migration and in vivo tumor growth.  Together, my findings illustrate the complexity of how changes in ECM-cell interactions, and subsequent adhesome signaling, influence processes that are critical for cancer progression.    iv  Preface I planned, performed and analyzed all experiments with the exceptions listed below.   May Dang-Lawson performed - 2 repeats for Figure 12B - All repeats of Figure 18A - Cloning egfp-talin 1 and egfp-talin 2 into an expression vector with the β-actin promoter, which were used in section 6.2.6  Caroline Chu provided technical help with: - Generating and screening the B16F1 KO clones that were used in Chapter 6. Under my guidance Caroline Chu also performed and analyzed  - All repeats for Figure 32B, including data analysis  - All repeats for Figure 39, including data analysis - Figure 42 right; I performed the analysis.  Kate Choi assisted with tumor cell injection and the monitoring of mice in the experiments shown in Figure 43. She also performed some of the repeats for the experiments in Chapter 6.2.6.    Animal studies were conducted in the Modified Barrier Facility at the University of British Columbia. All animal work was performed under strict accordance with the recommendations of the Canadian Council for Animal Care. Protocols were approved by the Animal Care Committee   v  (ACC) (A11-0199) of the University of British Columbia. Cell sorting was done by the UBC Flow Cytometry Facility.   vi  Table of Contents Abstract ........................................................................................................................................... ii	Preface ............................................................................................................................................ iv	Table of Contents ........................................................................................................................... vi	List of Tables .................................................................................................................................. x	List of Figures ............................................................................................................................... xii	List of Abbreviations .................................................................................................................... xv	Acknowledgements .................................................................................................................... xviii	Chapter 1: Introduction ................................................................................................................... 1	1.1	 Cancer as a multifactorial disease ............................................................................... 3	1.2	 The hallmarks of cancer .............................................................................................. 5	1.2.1	 Self-sufficiency in growth signals ......................................................................... 5	1.2.2	 Insensitivity to anti-growth signals ....................................................................... 6	1.2.3	 Unlimited replicative potential .............................................................................. 6	1.2.4	 Evasion of apoptosis.............................................................................................. 7	1.2.5	 Sustained angiogenesis .......................................................................................... 8	1.2.6	 Tumor invasion and metastasis ............................................................................. 9	1.2.7	 Enabling characteristics and emerging hallmarks ............................................... 12	1.3	 Mechanobiological forces ......................................................................................... 18	1.4	 The ECM plays an important role during cancer progression .................................. 20	1.4.1	 ECM and cancer cell metastasis .......................................................................... 25	1.4.2	 ECM and tumor cell migration ............................................................................ 25	1.4.3	 ECM stiffness and adhesion-ligand densities regulate tumor cell behavior ....... 28	1.5	 Integrin-containing adhesion complexes couple the ECM to cellular functions ...... 29	1.6	 The role of integrin signalling in cancer ................................................................... 33	1.7	 Adhesome proteins and cancer ................................................................................. 34	1.8	 The role of integrins and the adhesome in resistance to chemotherapy ................... 35	1.9	 Outside-in integrin signalling activates signalling pathways that regulate cell behavior and gene expression ............................................................................................... 35	1.9.1	 Integrin signalling pathways control cell migration, invasion and adhesion ...... 35	1.9.2	 Regulation of Rap1 during integrin signalling .................................................... 37	1.9.3	 Integrin signalling as a regulator of transcription ............................................... 38	1.9.4	 Actin dynamics link integrin function to changes in gene expression ................ 40	  vii  1.9.5	 Mechanosensitive control of transcription regulation ......................................... 42	1.10	 The role of talin and FAK in adhesome function and integrin signalling ................ 44	1.10.1	 Talin is a scaffolding protein with binding domains for adhesome proteins ... 44	1.10.2	 Talin as an integrin activator ........................................................................... 46	1.10.3	 Regulation of talin-induced integrin activation via autoinhibition .................. 46	1.10.4	 Talin as a mechanical link between integrins and the cytoskeleton ................ 48	1.10.5	 The role of talin in FA dynamics and turnover ............................................... 49	1.10.6	 The role of talin in cell proliferation ............................................................... 50	1.10.7	 Talin 1 and talin 2 ............................................................................................ 50	1.10.8	 Talin and cancer ............................................................................................... 52	1.10.9	 FAK is both a kinase and scaffolding protein ................................................. 54	1.10.10	 Integrin-dependent FAK activation ................................................................. 55	1.10.11	 The role of FAK in cancer ............................................................................... 56	1.10.12	 FAK functions in the nucleus .......................................................................... 58	1.11	 Rationale and thesis aim ........................................................................................... 59	Chapter 2: Material and methods .................................................................................................. 62	2.1	 Materials ................................................................................................................... 62	2.2	 Cell culture procedures ............................................................................................. 70	2.2.1	 Culture and storage of B16 cells ......................................................................... 70	2.2.2	 Transfection of B16 cells with plasmid DNA or siRNA .................................... 70	2.2.3	 Cell sorting and clonal expansion ....................................................................... 71	2.2.4	 Cell spreading assay and image quantification ................................................... 71	2.2.5	 2D bead-clearing motility assay .......................................................................... 72	2.2.6	 Motility assay with real-time imaging ................................................................ 73	2.2.7	 Preparation of polyHEMA-coated plates for cell culture .................................... 73	2.2.8	 Cell growth and viability assay ........................................................................... 74	2.2.9	 3D collagen I/FN for cell embedding .................................................................. 74	2.2.10	 Staining of cells and RNA isolation from cells in 3D collagen/FN gels ......... 75	2.2.11	 Cell stretching .................................................................................................. 75	2.3	 Molecular biology techniques ................................................................................... 76	2.3.1	 RNA techniques .................................................................................................. 76	2.3.2	 Bacterial transformation ...................................................................................... 79	2.3.3	 Plasmid preparation ............................................................................................. 79	2.3.4	 Design and cloning of CRISPR/CAS9 plasmids ................................................. 79	  viii  2.4	 Biochemical methods ................................................................................................ 80	2.4.1	 Preparation of cell lysates ................................................................................... 80	2.4.2	 SDS-PAGE and western blotting ........................................................................ 80	2.4.3	 Rap activation assay ............................................................................................ 81	2.5	 Subcutaneous growth of B16F1 tumors in C57BL/6 mice ....................................... 81	Chapter 3: Cancer signature genes are regulated by integrin ligand density but not by cellular tension ........................................................................................................................................... 83	3.1	 Introduction ............................................................................................................... 83	3.2	 Results ....................................................................................................................... 85	3.2.1	 Identification of genes regulated by cell adhesion to increasing FN density ...... 85	3.2.2	 Identification of cancer signature genes .............................................................. 91	3.2.3	 Adhesion stimulates FA signalling in B16F1 cells ............................................. 92	3.2.4	 Adhesion is a strong regulator of the cancer signature gene expression ............. 94	3.2.5	 Mechanobiological force is a weak regulator of the cancer signature genes ...... 96	3.3	 Summary and discussion ......................................................................................... 110	Chapter 4: FN density and cell stretching result in overlapping but distinct patterns of gene regulation .................................................................................................................................... 119	4.1	 Increasing FN density and applying stretch regulate different set of genes ........... 121	4.2	 The majority of the top 10% genes regulated by FN density show a dose-dependent regulation pattern ................................................................................................................ 138	4.3	 FN density-regulated and stretch-regulated genes have limited overlap ................ 140	4.4	 Increasing FN density and applying stretch regulate genes that cluster in different pathways ............................................................................................................................. 141	4.5	 FN density and stretch regulate distinct pathways and distinct members of shared pathways ............................................................................................................................. 144	4.6	 FN density and mechanical stretch may activate distinct sets of TFs ..................... 145	4.7	 Summary and perspectives ..................................................................................... 148	Chapter 5: The adhesome proteins talin and FAK regulate expression of the cancer signature genes ........................................................................................................................................... 156	5.1	 Introduction ............................................................................................................. 156	5.2	 Talin is a regulator of MUC18 and TRPM1 gene expression ................................ 157	5.2.1	 Talin is important for cell spreading in B16F1 cells ......................................... 157	5.2.2	 Talin regulates MUC18 and TRPM1 expression in B16F1 cells ...................... 159	5.2.3	 Talin 1 is important for the regulation of MUC18 and TRPM1 ....................... 160	5.2.4	 Talin 1 and talin 2 both regulate B16F1 cell spreading and cell motility ......... 162	5.3	 FAK regulates expression of the cancer signature genes ........................................ 164	  ix  5.3.1	 Knocking down FAK increases cell spreading ................................................. 165	5.3.2	 Knocking down FAK causes distinct changes in cell morphology changes including loss of cell polarity .......................................................................................... 166	5.3.3	 Knocking down FAK regulates signature genes ............................................... 168	5.4	 Summary and discussion ......................................................................................... 169	Chapter 6: Talin 1 and talin 2 differentially regulate B16F1 cell spreading, migration and tumor growth ......................................................................................................................................... 179	6.1	 Introduction ............................................................................................................. 179	6.2	 Results ..................................................................................................................... 180	6.2.1	 Creation of talin 1 and talin 2 single and double KO B16F1 cell lines ............. 180	6.2.2	 Talin 1 and talin 2 regulate cell spreading and cell morphology ...................... 182	6.2.3	 Talin 1 and talin 2 regulate cell motility and velocity ....................................... 185	6.2.4	 Talin 1 and talin 2 may determine the mode of cell motility in B16F1 cells .... 188	6.2.5	 Talin regulates B16F1 tumor growth in vivo .................................................... 191	6.2.6	 Re-expressing talin in talin KO clones partially rescues normal cell behavior 194	6.3	 Summary and discussion ......................................................................................... 201	Chapter 7: Overall conclusion and future directions .................................................................. 209	References ................................................................................................................................... 219	Appendix ..................................................................................................................................... 284	  x  List of Tables Table 1: Materials ......................................................................................................................... 62	Table 2: Antibodies and staining reagents for immunofluorescence ............................................ 63	Table 3: Commercially available solutions and chemicals ........................................................... 64	Table 4: Reagent compositions ..................................................................................................... 65	Table 5: Kits .................................................................................................................................. 66	Table 6: Biological materials ........................................................................................................ 67	Table 7: Cell lines ......................................................................................................................... 67	Table 8: Equipment ....................................................................................................................... 67	Table 9: Primers for qPCR ............................................................................................................ 68	Table 10: CRISPR primers with encoded gRNA site ................................................................... 69	Table 11: siRNAs .......................................................................................................................... 69	Table 12: Software ........................................................................................................................ 69	Table 13: PCR Program for qRT-PCR ......................................................................................... 78	Table 14: Top 20% upregulated genes from low to medium FN ................................................. 87	Table 15: Top 20% upregulated genes from low to high FN ....................................................... 88	Table 16: Top 20% downregulated genes from low to medium FN ............................................. 89	Table 17: Top 20% downregulated genes from low to high FN ................................................... 90	Table 18: Top 10% upregulated genes from low to medium FN ............................................... 125	Table 19: Top 10% upregulated genes from low to high FN ..................................................... 127	Table 20: Top 10% upregulated genes from no stretch to stretch .............................................. 129	Table 21: Top 10% downregulated genes from low to medium FN ........................................... 131	Table 22: Top 10% downregulated genes from low to high FN ................................................. 134	  xi  Table 23: Top 10% downregulated genes from no stretch to stretch ......................................... 137	Supplementary Table 24: Top 10% upregulated genes from low to medium and low to high FN and ratio between both groups for expression pattern A in Figure 23 ........................................ 287	Supplementary Table 25: Top 10% upregulated genes from low to medium and low to high FN and ratio between both groups for expression pattern B in Figure 23 ........................................ 293	Supplementary Table 26: Top 10% upregulated genes from low to medium and low to high FN and ratio between both groups for expression pattern C in Figure 23 ........................................ 293	Supplementary Table 27: Top 10% upregulated genes from low to medium and low to high FN and ratio between both groups for expression pattern D in Figure 23 ........................................ 295	Supplementary Table 28: Top 10% upregulated genes from low to medium and low to high FN and ratio between both groups for expression pattern E in Figure 23 ........................................ 295	Supplementary Table 29: Enriched SuperPaths for genes upregulated in response to increased FN density ......................................................................................................................................... 296	Supplementary Table 30: Enriched SuperPaths for genes in upregulated in response stretch ... 298	   xii  List of Figures Figure 1: The hallmarks of cancer. ................................................................................................. 5	Figure 2: Loss of tissue architecture and ECM remodeling during cancer progression. .............. 22	Figure 3: A comparison of mesenchymal and amoeboid modes of single cell migration. ........... 26	Figure 4: Model of integrin activation and the integrin adhesome. .............................................. 31	Figure 5: Integrin ‘outside-in’ signalling governs cellular processes ........................................... 37	Figure 6: Connections from the ECM to the genome. .................................................................. 43	Figure 7: Domain organization of talin 1. ..................................................................................... 45	Figure 8: Talin exposes vinculin-binding sites in response to force-induced conformational change. .......................................................................................................................................... 47	Figure 9: Schematic representation of FAK. ................................................................................ 55	Figure 10: Cell adhesion to ECM regulates cell morphology. ...................................................... 86	Figure 11: Cell adhesion to ECM regulates gene expression of Cyr61, MUC18 and TRPM1. ... 92	Figure 12: Adhesion to TC plastic results in adhesome signalling. .............................................. 94	Figure 13: Adhesion to TC plastic regulates cancer signature gene expression. .......................... 95	Figure 14: Stretch does not regulate cancer signature gene expression. ....................................... 99	Figure 15: Mechanical stretch in 3D is a weak regulator of the signature genes. ...................... 101	Figure 16: Blebbistatin alters cell morphology, as well as FA size and spatial distribution in B16F1 cells. ................................................................................................................................ 103	Figure 17: Myosin II activity weakly regulates the expression of Cyr61 and MUC18. ............. 104	Figure 18: Blocking Rap activation weakly affects cancer signature gene expression. ............. 106	Figure 19: Blocking Rap activity is a weak regulator of the signature gene expression. ........... 109	  xiii  Figure 20: Cell adhesion and adhesion to increasing FN density strongly regulate the signature genes. .......................................................................................................................................... 112	Figure 21: Changes in mRNA levels upon FN density or cellular stretch. ................................. 123	Figure 22: Validation of fold changes in signature genes by qRT-PCR. .................................... 124	Figure 23: Comparison of gene expression changes between B16F1 on low to medium and low to high FN density. ...................................................................................................................... 139	Figure 24: FN density and stretch regulate distinct sets of genes. .............................................. 141	Figure 25: FN density and mechanical stretch regulate distinct pathways. ................................ 143	Figure 26: Pathways or GO biological processes with different sets of genes regulated by FN density, stretch or both stimuli. ................................................................................................... 145	Figure 27: Increasing FN density and mechanical stretch upregulate gene clusters that are enriched for distinct TFBS motifs............................................................................................... 147	Figure 28: Transcriptome analysis of B16F1 cells plated on increasing FN density or subjected to mechanical stretch. ...................................................................................................................... 149	Figure 29: Knocking down talin reduces cell spreading and FAK phosphorylation. ................. 158	Figure 30: Talin expression regulates MUC18 and TRPM1 expression. ................................... 160	Figure 31: Talin 1 is a stronger regulator of MUC18 and TRPM1 expression than talin 2. ...... 161	Figure 32: Talin 1 and talin 2 regulate cell spreading and motility. ........................................... 163	Figure 33: Knocking down FAK increases B16F1 cell spreading. ............................................ 165	Figure 34: Loss of cell polarity and distinct changes in cytoskeletal organization are caused by FAK KD. ..................................................................................................................................... 167	Figure 35: FAK KD differentially regulate mRNA levels of the signature genes. ..................... 168	Figure 36: Changes in adhesome composition regulate cancer signature gene expression. ....... 170	  xiv  Figure 37: Model for how talin KD could affect adhesomes. ..................................................... 172	Figure 38 CRISPR/Cas strategy to target talin 1 and talin 2 in B16F1 cells. ............................. 182	Figure 39: Talin 1 and talin 2 have different effects on cell spreading and cell shape. .............. 185	Figure 40: The loss of either or both talins results in reduced 2D migration distance and velocity on FN. ......................................................................................................................................... 187	Figure 41: Talin KO and control B16F1 cells have different morphology during 2D migration...................................................................................................................................................... 190	Figure 42: The loss of both talin isoforms reduces B16F1 cell viability when cell are forced into suspension. .................................................................................................................................. 192	Figure 43: Talin 2 KO impairs subcutaneous tumor growth ...................................................... 194	Figure 44: Reconstitution of double and single talin KO cells with egfp-tagged talin proteins. 196	Figure 45: Reconstitution of double talin KO cells with either egfp-talin 1 or egfp-talin 2 partially restores cell spreading on FN. ...................................................................................... 198	Figure 46: Reconstitution of talin 1 KO cells with egfp-talin 1 restores cell spreading on FN. . 199	Figure 47: Reconstitution of talin 2 KO cells with egfp-talin 2 does not affect cell spreading on FN. .............................................................................................................................................. 200	Figure 48: Knocking out talin 1, talin 2, or both talins has different effects on processes related to cancer progression. ..................................................................................................................... 201	Figure 49: ECM-cell dependent control of cancer progression. ................................................. 210	Figure 50: An unbiased systems biology approach to identify novel therapeutic targets for cancer progression associated with alterations in cell-ECM communication. ....................................... 217	    xv  List of Abbreviations 2D 3D Two-dimensional Three-dimensional AJCC American Joint Committee on Cancer AMT Amoeboid–mesenchymal transition ARHGEF28 Rho guanine nucleotide exchange factor 28 Arp2/3 Actin-related protein-2/3 ATP Adenosine triphosphate CAF Cancer-associated fibroblasts CDK5 Cell division protein kinase 5 CIN Chromosome instability CRISPR Clustered regulatory interspaced short palindromic repeat crRNA CRISPR RNAs  CTC Circulating tumor cells Cyr61 Cysteine-rich angiogenic inducer 61 DSB Double-strand break  ECM Extracellular matrix EGFR Epidermal growth factor receptor EGFP Enhanced Green Fluorescent Protein EMT Epithelial-Mesenchymal Transition ERK Extracellular signal-regulated kinase FAK Focal adhesion kinase FAT Focal adhesion targeting    xvi  FERM 4.1 protein, ezrin, radixin and moesin FN Fibronectin GRB2 Growth factor receptor-bound protein 2 HER2 Human epidermal growth factor receptor 2 HIF Hypoxia-Inducible Factor IL Interleukin i.v. Intravenous  kDa Kilodalton KD  Knockdown KO Knockout LOF Loss-of-function MAT Mesenchymal–amoeboid transition MEFs Mouse embryonic fibroblasts MMPs Matrix metalloproteinases MUC18/MCAM Melanoma cell adhesion molecule  NES Nuclear export sequence  NHEJ Non-homologous end joining  NLS Nuclear localization signal  N-WASP Neuronal Wiskott–Aldrich Syndrome protein OMIM Online Mendelian Inheritance in Man pTyr Phospho-Tyrosine p1µCas Crk-associated substrate of 130 kDa p53 Tumor protein 53   xvii  PDGFR Platelet-derived growth factor receptors PI3K Phosphoinositide 3-kinase PTEN Phosphatase and tensin homolog RIAM Rap1-GTP-interacting adaptor molecule  RNA-seq RNA sequencing ROCK Rho-associated protein kinase s.c. Subcutaneous  SD Standard deviation SEM Standard error of the mean SH2 Src homology 2 TAF Tumor-Associated Fibroblasts TAM Tumor-associated macrophage TC Tissue culture  TF Transcription Factor TFBS Transcription Factor Binding Site TGF Transforming growth factor TNFα Tumor necrosis factor α TP53 Tumor protein p53 gene TRPM1 Transient receptor potential cation channel subfamily M member 1 VCAM Vascular cell adhesion molecule YAP Yes-associated protein   xviii  Acknowledgements Firstly, I would like to express my gratitude to my advisor Dr. Michael Gold for the opportunity and continuous support of my Ph.D. study. His scientific guidance, his patience, and his encouragement to form and test my own ideas, have helped me develop personally and scientifically, and enabled me to acquire most valuable skills for future challenges. I would like to thank Dr. Calvin Roskelley, for his contributions to my projects and his generosity in providing materials, sharing equipment and expertise. My sincere thanks also go to my committee members, Dr. Kenneth Harder, Dr. Calvin Roskelley, Dr. Georgia Perona-Wright, and Dr. Gerry Weeks, who provided me with scientific guidance and advice on my projects, and generously allowed me to use their equipment. I would like to thank Dr. Linda Matsuuchi for helpful suggestions and advice. I thank many past and present lab members of the Gold, Matsuuchi and Roskelley labs, many of which have become friends over the past years. I am thankful for their stimulating discussions, support, encouragement, hands on help and for being great colleagues. Past members I would like to thank are Dr. Marcia Graves, Dr. Spencer Freeman, Dr. Raibatak Das, Dr. Sarah McLeod, Dr. Caylib Durand, Dr. Kevin Lin, Dr. Kathy Tse, Victor Lei, Zinaida Tebaykina, Aileen Xiaolin Wang-Liu, Dr. Steve Machtaler, Caren Grande, Letitia Falk, Stephanie Mancini, Megan Gilmour, and Jane Cipollone. Current lab members I would like to thank are Madison Bolger-Munro, Dr. Libin Abraham, Jia Wang, Josh Scurll, Henry Lu, Timothy Jou, Farnaz Pournia, Tak Poon, Erin Bell and Pamela Dean. I would like to particularly thank May Dang-Lawson, Kate Choi, and Caroline Chu, who not only lent many helping hands during experiments, but also contributed to my personal and professional development. I also would like to express my gratitude to Darlene Birkenhead for her continuous support and advice. I would like to extend my gratitude to many members of the department for   xix  being great collogues. Great thanks go to family and friends, in particular Dawn and Reini, who have supported me from near and far over the past years. I am very grateful for my family’s financial and moral support. Importantly, special thanks go to Adam Plumb.     1  Chapter 1: Introduction Despite all efforts to find better treatments and cures for cancer it remains one of the leading causes of death. The World Health Organization estimates that there were 14.1 million cancer diagnoses and 8.2 million cancer deaths in 20121. For 2015, the Canadian Cancer Society predicted that 196,900 Canadians would be diagnosed with cancer and 78,000 would die2. Hence we need to better understand the underlying mechanisms of cancer to prevent its occurrence and improve outcomes.  Cancer is typically thought of as a multi-step disease3-6. Normal cells start proliferating in an uncontrolled manner, progress to infiltrating the surrounding healthy tissue and lymph nodes, and in the most advanced stage, metastasize to new organs, where they establish secondary tumors5, 7, 8. Based on clinical presentation, which can vary between different types of cancers, clinicians have modeled staging systems that rate cancers according to their state of progression. For solid tumors, in stage 0, tumor cells are found in situ, but have neither infiltrated surrounding tissue nor spread to other organs. In stage I, the tumor is larger in size and can be surgically removed. In stages II and III, the tumor cells have started to infiltrate locally and have reached lymph nodes. Exact staging and therapy is dependent on the tumor type. Therapies for these stages include chemotherapy, radiation therapy or surgery. In stage IV, the most advanced stage, the cancer has metastasized and spread to other organs9. The relevance of this staging is demonstrated by the case of malignant melanoma, a highly metastatic cancer. Although it accounts for < 2% of all skin cancer cases, it causes 75% of all skin cancer-related deaths10, 11. In malignant melanoma, patient survival is inversely correlated with progression stages, in that patients who are diagnosed in stage I have a 97% 5-year survival rate whereas stage IV patients have a 16% 5-year survival rate12.    2  Melanoma is a complex disease with genetic (e.g. gene mutations) and environmental (e.g. intense sun exposure and toxin exposure) aspects. Development of melanoma progresses through a multi-step process. Initially, normal-appearing melanocytes start proliferating and grow into precursor lesions, e.g. dysplastic nevi. These lesions can then enter the horizontal/radial growth phase in which the melanoma is flat and grows slowly. Subsequently, upon entering the vertical growth phase, the melanoma thickens, raises, and starts invading the underlying tissue. The depth of tissue penetration is a strong predictor for outcome, as thicker and deeper melanomas are more likely to metastasize13. Ultimately, the invading cells can enter the lymph nodes or blood vessels and produce distant metastases. Prior and during the melanoma stages, the cells change in morphology, acquire gene mutations and change their protein expression pattern. Some of these changes are well characterized. For example, many melanomas carry activating mutations in BRAF as well as loss of functional p16INK4a and p14ARF 14. These mutations facilitate uncontrolled cell cycle progression and proliferation, and some patients benefit from BRAF inhibitors15. Other mutations and gene expression changes, such as downregulation of E-cadherin and c-KIT, loss of TRPM1, and gain-of-function mutations in BRN2, are associated with the later stages of melanoma. Recent progress has been made in identifying distinct gene expression changes that correlate with progression through the stages of melanoma. However, we still need to identify the underlying pathways that lead to melanoma progression, in particular how transcriptional changes that drive melanoma progression are regulated. Achieving this would allow us to develop molecular diagnostic tools as well as targeted therapies for melanoma.   Until 2011 there were few treatment options for metastatic melanoma, and these were mostly nonspecific cytotoxic agents. Later stage disease (stages III and IV) was considered to be   3  incurable16. In 2011, breakthroughs in research led to the approval of several immunotherapies that boost the immune system by targeting the PD-1/PD-L1 pathway (pembrolizuma and nivolumab) or CTLA-4 (ipilimumab). These therapies extend progression-free survival, or in some cases, cure the disease17-19. Although these new treatments shifted an untreatable disease towards becoming a manageable disease, it quickly became clear that the new therapies had limitations. Specifically, these novel approaches only benefit a small subset of patients and it is not possible to predict who will respond. Toxicity and resistance further limit broad application and success. For example, patients who initially respond well to MAP kinase inhibitors often relapse (within a year) due to the development of resistance15, 20, 21. Although current clinical studies are exploring combinations of existing immunotherapies and targeted therapies in order to overcome and/or avoid drug resistance, there is a need to improve treatment options by identifying new targets and developing new targeted drugs. New treatment approaches also have the potential to improve the treatment of other metastatic cancers (e.g. lung, colon, breast, pancreatic and prostate) as well as blood-borne malignancies with poor survival rates.   1.1 Cancer as a multifactorial disease  In contrast to diseases that can be traced to a single underlying cause, such as viral or bacterial infections, cancer is a large class of very diverse diseases that ultimately lead to uncontrolled cell growth, destruction of functional tissues and organs, and the spread of cancer cells throughout the body. Many researchers have studied different aspects of this disease and contributed to the understanding of its complexity. However in 2000, Douglas Hanahan and Robert Weinberg published a ground-breaking paper ‘Hallmarks of Cancer’ in Cell5, which changed how we research and approach cancer treatment today. The authors shared their idea of   4  how cancer research would develop into a conceptual science, where instead of adding more layers of complexity to the growing literature, we would use current and future data to describe a finite number of underlying principles (‘traits’) that are common to all cancers.  Based on observations from human cancers and animal models, Hanahan and Weinberg defined the hallmarks that are common to all cancers as: self-sufficiency in terms of growth signals, loss of sensitivity to anti-growth signals, unlimited replicative potential, sustained angiogenesis, evasion of apoptosis, tissue invasion and metastases. In order to become a cancer cell, a cell would have to acquire mutations that would result in each of these hallmarks. Today, with some additions and refinements, these proposed ‘traits’ or hallmarks are widely accepted and serve as a framework for development of new concepts in cancer research. A recent update to this article added several additional characteristics of cancer, including immune evasion, tumor-promoting inflammation, and the role of stromal cells and extracellular matrix (ECM) in supporting tumor growth6. Figure 1 provides an overview of these hallmarks of cancer, each of which will be described in the following sections.    5   Figure 1: The hallmarks of cancer.  This illustrates the six hallmarks of cancer as proposed by Hanahan and Weinberg in 2000 as well as the emerging hallmarks and enabling characteristics added by the same authors in 2011.Based on Hanahan & Weinberg5, 6.   1.2 The hallmarks of cancer 1.2.1 Self-sufficiency in growth signals Normal cells mostly enter the cell cycle in response to growth factor stimulation. In contrast, cancer cells have acquired the ability to enter the cell cycle and proliferate independently of external cues. In this way they evade tissue homeostasis. There are several ways in which cancer cells can acquire the ability to undergo sustained proliferation: 1) By producing and releasing their own growth factors (autocrine growth) or by inducing other cells to produce growth factors that bind to receptors on the cancer cells (paracrine growth); 2)   6  Upregulating the expression of receptors so they become hyperresponsive to limited amounts of growth factors; and 3) Constitutive activation of growth factor signalling pathways that allow the cell to enter the cell cycle independently of ligand-induced signals.   1.2.2 Insensitivity to anti-growth signals Besides the ability to promote their own proliferation, cancer cells must also become resistant to signals that otherwise inhibit cell proliferation. This usually involves the loss of proteins that are encoded by tumor suppressor genes, which occurs in many cancers22, 23. Many tumor suppressors function as checkpoints that determine whether a particular cell can enter the cell cycle and proliferate, as opposed to undergo senescence. The best example is the tumor protein 53 (p53), which is able to halt the cell cycle and induce apoptosis in response to damage to the genome that cannot be repaired. Mutations in this gene can lead to proliferation of cells with chromosomal aberrations that activate oncogenes, often resulting in malignant neoplasms.   1.2.3 Unlimited replicative potential  Most cells go through a limited number of cell cycles before they enter senescence, a non-proliferative state, and finally undergo cell death. In contrast, cancer cells become ‘immortalized’, which means that they can potentially undergo unlimited growth-and-division24. The ability to overcome the cell division limit has been attributed to enzymes termed telomerases, which prevent telomere shorting25. Telomerases are not expressed in non-dividing somatic cells but are highly expressed in many progressive cancers24, 26. Telomeres are repetitive DNA sequences at the ends of linear chromosomes that form a protective capped structure. During the S phase of each cell cycle, telomeres become shorter. This triggers senescence and   7  apoptosis if the telomeres become shortened beyond a certain point. By reactivating telomerases, cancer cells can maintain telomere length and evade senescence and cell death27, 28.   1.2.4 Evasion of apoptosis  Apoptosis, the controlled dying of a cell, is a highly regulated fate that most cell types undergo. Once triggered, the cell body is broken down until it is engulfed by surrounding phagocytic cells29. In contrast to necrosis, where the cell contents are released into the surrounding space, potentially disturbing the function of neighboring cells and triggering an inflammatory response30, apoptosis is a physiological program that ensures tissue homeostasis by eliminating aberrant cells that could otherwise give rise to cancer31. Hence, apoptosis is a natural barrier against cancer development and progression31-33. The apoptotic machinery can be divided into two parts: sensors and effectors. Sensors monitor the extracellular environment, for example, Fas ligand binding to Fas receptors or tumor necrosis factor (TNF) α binding to TNF-Receptor 134. There are also intracellular triggers of apoptosis such as DNA damage, signalling imbalance, hypoxia and the lack of survival factors33. Once these sensors are activated, they trigger effectors of the apoptotic cascade. Key effectors of apoptosis are members of the Bcl-2 family34. Pro-survival members of this family (e.g. Bcl-2, Bcl-x) inhibit pro-apoptotic Bcl-2 family members, such as Bax and Bak. In the absence of Bcl-2 or Bcl-x, Bax and Bak triggers the activation of a caspase cascade, which promotes apoptosis by cleaving multiple substrates, including cytokeratins, the poly ADP ribose polymerase (PARP), the plasma membrane cytoskeletal protein α-fodrin, and the nuclear protein NuMA35-37. Cancer cells exhibit a diverse range of antiapoptotic strategies, such as the loss of critical danger signal molecules (e.g.p53)   8  that would lead to crisis and cell death, the upregulation of anti-apoptotic regulators, or the downregulation of pro-apoptotic factors.   1.2.5 Sustained angiogenesis Almost all cells rely on a steady supply of nutrients and oxygen that is ensured by their proximity to capillary blood vessels. Initially, blood vessels are formed though vasculogenesis during embryonic development, which involves the differentiation of endothelial cells from angioblasts38. New blood vessels are then formed by angiogenic sprouting from existing vessels. This is a highly regulated mechanism that is active primarily during embryonic development and then transiently switched on during wound healing39, the female reproductive cycle, and pregnancy40. During tumor progression, cancers usually transition from an avascularized hyperplasia to a vascularized tumor that can progress into a malignant tumor. During this process, cancer cells develop a connection to the vascular system. This is referred to as the ‘angiogenic switch’41. In order to satisfy the tumor’s need for oxygen and nutrients42, cancer cells usurp the same angiogenic signalling pathways that allow vascular remodeling by producing angiogenic growth factors such as vascular endothelial growth factor43, 44 or angiopoietins45. Macroscopic cancerous lesions usually exhibit increased angiogenesis, but the degree of vascularization can vary depending on the tumor type. Some highly aggressive tumors show little vascularization and even seem to actively limit angiogenesis46. Other tumors are hypervascularized47, 48, often with chronically activated angiogenesis that is characterized by excessive vessel branching and by enlarged vessels that are prone to microhemorrhaging and leakiness49. This makes the angiogenic pathway an attractive target for anti-cancer drugs.      9  1.2.6 Tumor invasion and metastasis During disease progression, tumor cells often invade and destroy surrounding tissue and then proceed to colonize distant organs to form new secondary tumors, a process called metastasis. Tumor metastases are often difficult to treat50 and are responsible for more than 90% of all cancer deaths51-53. In 1889 Stephen Paget suggested that tumor cells follow a ‘seed and soil’ principal54 where organ-specific metastasis depends on the primary tumor cells (the seed) being able to grow in the microenvironment of a distant organ (the metastatic niche55). The anatomy of vascular and lymphatic drainage from the site of the primary tumor were also recognized early on as additional important factors56.  It is now known that metastasis is a multistep and multifactorial event that is often referred to as the ‘invasion-metastatic cascade’57, 58. Briefly, tumor cells invade the surrounding local tissue, intravasate into the nearby blood stream or lymphatic vessels and then travel through these systems until they extravasate out of these vessels into the parenchyma. Once in the tissue at a secondary site, the tumor cells establish micrometastatic lesions. This is followed by the development of macroscopic tumors, a process referred to as ‘colonization’. This cycle can be reiterated and metastatic cells can produce additional metastases59. Metastasis is a highly inefficient process and most circulating tumor cell never succeed in forming secondary growths60. Multiple studies61 have shown that cell proliferation at the secondary site is a major limiting step in the metastatic cascade and only cells that exhibit all of the hallmarks of cancer are able to form secondary tumors. Acquiring these hallmarks recapitulates a developmental program that coordinately controls cell migration, proliferation and angiogenesis during embryonic morphogenesis. Cancer cells can transiently reactivate this program, which is referred to as the ‘epithelial-mesenchymal transition’ (EMT), a process that is   10  especially important for cancer invasion and metastasis62-65. EMT is driven by a set of developmental transcription factors (TFs) including Snail, Slug, Twist and Zeb1/2. In many types of cancer, various combinations of TFs promote (or are required for) tumor progression66, 67. However, how these TFs are regulated, and how their targets regulate other pathways that are crucial for tumor progression, is not fully understood.  During EMT, cancerous epithelial cells lose adhesions to the cells in the tissue. Both tight junctions and adherence junctions are disassembled. Epithelial cadherin (E-cadherin) is a key factor in the maintenance of normal cell-cell junctions68. E-cadherins form homophilic adhesive complexes between contacting cells69, 70. Its cytoplasmic tail associates with intracellular proteins, such as β-catenin, p120-Catenin and vinculin, which connect to actin and actin-binding proteins71, 72, thereby linking cell-cell adhesions to the underlying actin cortex and allowing actin filament reorganization in response to changes in cell-cell contact73 74, 75. Together with vinculin and α-actinin, E-cadherins play an important role in mechanically coupling the actomyosin networks of adjacent cells, thereby acting as a mechanosensor and distributing forces across the tissue76. In addition, signals originating from E-cadherin-containing cell-cell junctions can be transduced to the nucleus to regulate gene expression77. The TF β-catenin, which is associated with the E-cadherin cytoplasmic tail, can translocate to the nucleus, for example in response to applied mechanical forces72. This leads to the transcription of the oncogenes Myc and Twist178. In this way, the loss of E-cadherin-containing cell junctions can promote cancer progression and the loss of E-cadherin is a common feature in many cancers. In polarized epithelium, such as the breast lumen79, E-cadherin expression is associated with apical-basal polarity80, a functional asymmetry of membrane composition, cytoskeletal organization, organelle distribution, and protein distribution that regulates key aspects of   11  epithelial function such as regulated transport across tissue barriers in the body81. This asymmetry is determined in large part by three major groups of polarity determinants, the Crumbs/Stardust/Discs lost complex (Crb/Sdt/Dlt)82-84, the Par3/Par6/aPKC complex85-87, and the Scribble/Discs large/Lethal giant larvae complex (Scrib/Dlg/Lgl)88, most of which were originally identified as tissue organization and polarity genes in Drosophila 89. There is recent evidence that E-cadherin influences epithelial polarity through interaction with several of these polarity factors90-92. Apical-basal polarity in epithelial cells is thought of as a major gatekeeper against cancer93 and the loss of polarity that results in tissue disorganization and excessive cell growth is an indicator of cancer progression.  In many cancers, where cell polarity is lost, E-cadherin is downregulated or mutated resulting in reduced cell-cell adherence. This highlights the role of E-cadherin as a potential suppressor of invasion and metastasis94, 95. In several cancer types, including melanoma, the loss of E-cadherin function is accompanied by the de novo expression of N-cadherin (and other mesenchymal cadherins)96, a member of the cadherin family that is normally expressed in migrating neurons and mesenchymal cells during organogenesis97, 98. This observation has led to the theory of a ‘cadherin switch’, which occurs normally during embryogenic development, when cell types segregate from each other. This switch also occurs during the transition from benign to malignant tumors99, 100. This inappropriate expression of N-cadherin has been shown to promote cancer cell motility and invasion101-103. Although it is not entirely clear how the cadherin switch is regulated during development and in cancer cells, it is known that EMT-associated TFs regulate cadherin expression104, 105. For example, in several melanoma types, the phosphoinositide 3-kinase (PI3K)/phosphatase and tensin homolog (PTEN) pathway promotes   12  ‘the cadherin switch’ by upregulating the TFs Snail and Twist, which in turn transcriptionally repress E-cadherin expression and induce expression of N-cadherin106.   1.2.7 Enabling characteristics and emerging hallmarks In 2011, Hanahan and Weinberg updated their ‘hallmarks of cancer’ to include potential new hallmarks and enabling characteristics, and to acknowledge the role of the tumor microenvironment6. The two emerging hallmarks, ‘reprogramming energy metabolism’ and ‘evading immune destruction’ are crucial factors in many cancers and may qualify as additional hallmarks107, 108. In addition, the two enabling characteristics, ‘genomic instability and mutation’ and ‘tumor-promoting inflammation’ are seen as important supporting factors for cancer progression.   1.2.7.1 Emerging hallmarks 1.2.7.1.1 Reprogramming energy metabolism In the 1930s Otto Warburg noticed that cancer cells differ in their energy metabolism from normal differentiated cells109-112. Instead of predominantly producing energy through an aerobic process in mitochondria, even when oxygen is available, malignant cells use ‘aerobic glycolysis’, a process that occurs in the cytosol and is characterized by a high rate of glycolysis and lactic acid fermentation. This increased glycolysis rate is accompanied by the modulation of other metabolic proteins, for example upregulation of glucose transporters and enzymes of the glycolytic pathway113, 114. This shift in energy metabolism, which is now termed ‘the Warburg effect’, is also seen in highly proliferative cells such as rapidly dividing embryonic tissues. This supports the idea that aerobic glycolysis may be beneficial for fast growing cells, even though it   13  generates less adenosine triphosphate (ATP) than oxidative phosphorylation115. One possible explanation is that rapidly growing cells also need to generate other precursors for cellular macromolecules such as nucleosides and amino acids. Increased glycolysis allows for glycolytic intermediates to be diverted into these pathways, enabling the tumor cells to increase their biomass116. Interestingly, many tumors are comprised of different type of cells living a symbiotic lifestyle, with some cells undergoing aerobic glycolysis, producing and secreting high levels of lactate, and other cells using this ‘wasted’ lactate as their main energy source117-119.  Recent analysis of this metabolic switch in cancer cells has revealed mutations and changes in the expression levels of tumor-associated genes that control metabolic pathways114, 120-122. Cancer cells are dependent on metabolic reprogramming not only for increased proliferation but also other cancer progression traits such as evasion of growth-inhibitory signals, oncogenic signalling123, attenuation of apoptosis, tumor dissemination, and the establishment of metastases 124-126. However, the metabolic contribution to the mechanisms of disease progression is not well understood. Recent studies have highlighted the role of energy metabolism in tumor invasion and metastasis, leading to a model in which mitochondrial Hsp90 proteins, which are often highly expressed in tumor cells127, prevent autophagy and cell death and instead promote cell motility128. In this study the authors show that Hsp90 is able to maintain mitochondrial function and ATP production under bioenergetics stress, which usually would trigger compensatory autophagy via activation of energy sensors. In addition to opposing the induction of autophagy, Hsp90 proteins also activate the focal adhesion kinase (FAK), a key regulator of cell motility. Thus, upregulation of Hsp90 enhances tumor cell invasion under nutrient starvation conditions. This links a regulator of bioenergetic pathways to tumor metastasis.     14  1.2.7.1.2 Evading immune destruction  Tumor cells are not only influenced by neighboring cells in situ but also by immunological monitoring by the organism’s immune system. The idea that the immune system is able to recognize and protect against cancer was introduced in the early 20th century129. Almost 70 years later, T-cells were identified as the first cell type involved in immunosurveillance. We now know that many types of innate and adaptive immune cells can have both tumor-enabling and tumor-controlling functions, with the latter referred to as ‘the cancer immunoediting theory’ 130, 131. However, it wasn’t until the development of appropriate mouse models that it was shown that mice lacking specific types of immune cells, such as T-cells, natural killer (NK) cells and natural killer T-cells (NKT)132, exhibit increased rates of tumor formation and are more susceptible to carcinogens than immunocompetent mice133-135.  Cancer immunoediting consists of three phases: elimination, equilibrium and escape131. During the elimination phase the immune system is able to successfully detect and destroy cancerous cells and ensure the integrity of healthy tissue. The effective elimination of cancer cells relies on CD8+ cytotoxic T cells, CD4+ Th1 cells, dendritic cells that act as antigen-presenting cells, and NK cells136. Some tumor cells are rapidly proliferating, highly immunogenic cells that elicit a strong immune response. However, many cancer cells are weakly immunogenic and instead of being eliminated, their growth is only limited by the immune system. This is called the equilibrium phase. During the escape phase, further acquired mutations in these tumor cells result in the evasion of tumor cell recognition and killing137 as well as the development of the ability to suppress immune functions138, 139. This phase is also characterized by chronic inflammation (see below) that is driven by pro-inflammatory cytokines such as TNF and interleukin 6 (IL-6)140. All three phases of immunoediting are dependent on the cancer cell   15  communicating with their cellular and non-cellular environment via direct contact, cytokines and chemokines in an autocrine and paracrine fashion to create a complex network that fosters a tumor-permissive microenvironment (see 1.2.7.2.2).   1.2.7.2 Enabling Characteristics 1.2.7.2.1 Genome Instability and Mutation Cancer cells acquire the hallmarks of cancer in multiple ways, with the most common being alterations to their genomes (e.g. mutations), epigenomes (e.g. histone, DNA modifications) or transcriptomes (e.g. gene expression changes). Observers at the turn of the 20th century detected abnormal chromosomal structures and numbers in cancer cells141, 142; these are now known to be part of the acquired characteristic of chromosomal instability (CIN)143-145. Recently, high-throughput sequencing studies of human cancers have identified thousands of mutations in primary tumors and cancer cell lines146-151. Often, these mutations are found in genes involved in the DNA repair machinery152, 153, such as the tumor suppressor BRCA1154. In hereditary cancers, for example, mutations in many of the so called caretaker genes have been identified155-158. Caretaker genes are a type of tumor suppressor gene (e.g. TP53, BRCA1, mlh1 and msh2) with the ability to prevent mutations in other genes152, 159. When a caretaker gene is inactivated by mutations or epigenetically repressed, the absence of caretaker gene function allows for an increased spontaneous mutation rate, which can result in a tumor progression genotype. This process is known as the ‘mutator hypothesis’144, 160, 161.  In sporadic (non-hereditary) cancers, the molecular basis of genomic instability is less clear and mutations in the DNA caretaker genes seem to make only a minor contribution to CIN and tumor development162, 163. Mutation patterns in sporadic human cancers suggest that only a   16  few genes are mutated, deleted or amplified, although these genes frequently lie within growth signalling pathways or are cell cycle checkpoint genes149. A second model to explain CIN in sporadic cancer, termed the ‘oncogene-induced DNA replication stress model’, is based on recent studies which have identified thousands of mutations in primary cancers. This model partly attributes CIN to oncogene-induced DNA replication stress, a process that impairs DNA replication (e.g. stalls replication fork progression) and induces DNA damage (e.g. promotes generation of reactive oxygen species (ROS))164, 165. For example, mutations that activate oncogenes, such as in members of the ras family (K-ras, H-ras, and N-ras), induce replicative stress, generating chromosome mis-segregation and promoting aneuploidy166. Defects in genome maintenance and repair machinery are found in the majority of human cancer cells, which leads to the conclusion that mutations causing these defects provide a selective advantage for tumor cells by accelerating the rate at which premalignant cells accumulate mutations that drive a progressive phenotype.    1.2.7.2.2 Tumor-promoting inflammation  Cancer development is influenced not only by cell-intrinsic factors, but also by cell-extrinsic factors such as obesity, dietary factors, inhaled pollutants, tobacco use and autoimmunity167. A common denominator for these factors is the presence of chronic inflammation168, which can be tumor promoting and even inducing169, 170. The initial clue that inflammation could be important for the development of neoplasms came in 1881, when Rudolf Virchow observed the infiltration of immune cells into tumor masses171. Initially this was thought of as having only anti-tumor effects and serving as a surveillance program to detect and eliminate aberrant cells, consistent with the current concept of immunoediting172-176. However,   17  we know now that immune cells, with their repertoire of pro- and anti-inflammatory cytokines and signalling molecules, can have both tumor-suppressive and tumor-promoting effects177, 178.  Unresolved inflammation has been linked to tumor-promotion and progression for several cancer types179. These include colorectal cancer, which has been linked to inflammatory bowel diseases such as ulcerative colitis and Crohn's disease180. Similarly, chronic pancreatitis can be a precursor to pancreatic cancer181. As well, there is an emerging role for chronic inflammation during the development of inflammatory breast cancer, an aggressive type of breast cancer182, 183. Inflammation can alter the tumor microenvironment or niche, modulate the expression of growth factors, survival factors and proangiogenic factors, and facilitate EMT, invasion and metastasis177, 178, 184, 185. Many of these inflammation-associated effects depend on alterations to the ECM that favor tumor progression186. For example, increased deposition of fibronectin (FN) at the metastatic niche recruits bone marrow-derived cells that favor metastasis187, 188.  Tumour-associated macrophages (TAMs) are significant components of the inflammatory environment in tumors. Although TAMs can kill cancerous cells when activated by IL-2, IL-12 or γ-interferon, they can also promote the spread of tumors189, 190. For example, TAMs produce angiogenic and lymphangiogenic growth factors, pro-survival cytokines and proteases191. These proteases can remodel the ECM to facilitate tumor growth and invasion192, and can also inhibit the anti-tumor response of cytotoxic T-cells.  In melanoma and other cancers, inflammation is important at several stages of tumorigenesis193. Inflammation-associated production of ROS can result in DNA damage, possibly contributing to tumor initiation. Immune and inflammatory cells produce cytokines, such as members of the IL-10 family, growth factors such as EGF, and pro-angiogenic molecules (VEGF, bFGF, CXCL12), which promote cancer cell survival, proliferation and angiogenesis   18  during tumor progression. For metastatic cancer, cytokines such as transforming growth factor (TGF)-β, IL-1, TNF-α and IL-6 can trigger EMT and facilitate migration and invasiveness194, 195. Similarly, inflammatory signals facilitate survival, recruitment, colonization and growth at secondary tumor sites196, 197, indicating a potential role for inflammatory processes in the formation of pre-metastatic niches, which precede the formation of secondary metastases in the ‘seed and soil’ hypothesis198.  1.3 Mechanobiological forces  Mechanobiology describes how biological systems sense and respond to mechanical signals, and involves both the extracellular environment and intracellular signalling199. Mechanobiological effects are best illustrated in bones and blood vessels. Bones change their shape, density and stiffness when their mechanical loads are altered200, 201. Blood vessels undergo remodeling when exposed to altered pressure or shear stress202, 203. In addition, there is growing evidence that biomechanical stimuli such as strain and shear stress, as well as substrate rigidity and topography, have profound impacts on stem cell phenotypes204-208. Neural stem cells and progenitor cell differentiate into oligodendrocytes when subjected to stretch, demonstrating a direct role for tensile strain in dictating the lineage choices of stem cells209, 210.   On a cellular level, mechanotransduction refers to mechanisms by which tissues or individual cells convert mechanical stimuli from their environment into intracellular signals, which then regulate processes such as adhesion, migration, and survival211, 212. Both, extra- and intracellular factors (mechanical loads) impact mechanotransduction. Cells react to mechanical loads that are created by the rigidity of the ECM, fluid flow, shear stress, or tissue growth. They sense these forces through adhesome proteins that function as mechanosensors213, 214. Upon   19  physical force, the mechanosensors, e.g. integrins, talin and p130Cas, undergo conformational changes that initiate downstream signalling215-218. This is then translated into changes in cell behavior such as cell reorientation and migration. It also can cause cells to secrete molecules such as TGF-β and MMPs that promote ECM remodeling219-222. Cells and tissues use cytoskeletal networks, actin filaments, microtubules and intermediate filaments, as well as cytoskeletal regulatory proteins such as non-muscle myosin II to create cellular tension and to equilibrate the intra- and extracellular forces acting on them223, 224. This tensional balance stabilizes cell shape225 and mechanics, and plays a fundamental role in tissue homeostasis.  Alterations in this balance generated by excessive or abnormal ECM, as well as changes in cellular components that regulate internal tensile forces, lead to pathologies226. For example, many mutations have been identified in patients with cardiomyopathy that affect components of the cytoskeletal apparatus such as actin, titin and the β-myosin heavy chain227, 228. Defects in mechanotransduction are involved in several diseases including deafness229, arteriosclerosis230, 231, cancer and are associated with metastasis226, 232. Increased external mechanical loads, caused by compression233, increased interstitial fluid pressure/flow234-236, and ECM stiffening199, 226, 237 are generally thought of as metastasis-promoting factors. Cellular changes, such as genetic mutations or changes in gene expression that affect mechanotransduction pathways can also promote invasion and metastasis. One important mechanism of intracellular force generation is RhoA-dependent actomyosin contractility, which affects adhesion signalling and motility238-240. Increased Rho expression is associated with cancer including colon, breast and lung cancers241 and disruption of Rho signalling reduces cytoskeletal tension and can reduce tumor cell proliferation in in vitro systems226, 242.    20  During cancer progression, imbalances between extracellular and intracellular forces can lead to changes in gene expressions that promote invasion and metastasis. For example, increased ECM stiffness is associated with gene expression changes that promote EMT, with epithelial genes being downregulated and mesenchymal genes being upregulated243, 244. Recent findings suggest that mechanotransduction also regulates mitotic checkpoint genes and cell cycle regulators. In particular, the TFs YAP and TAZ, which translocate to the nucleus in response to tension, drive gene expression changes that promote growth, proliferation and differentiation245. Matrix rigidity also activates ERK in an integrin-dependent manner, which in turn drives cell cycle progression242, 246.   1.4 The ECM plays an important role during cancer progression Parallels have been drawn between cancer development and embryogenesis, which have led to a new view of the importance of the microenvironment, or niches, in regulating cell behavior247-249. The microenvironment is composed of cellular components, such as stromal cells and infiltrating immune cells, as well as non-cellular components, such as the ECM. Together, both components have a strong impact on tumor progression. Until recently the focus of research has been on how cellular components affect cancer progression, but recent progress has highlighted the importance of the ECM250. The biophysical and biochemical cues from tumor-associated ECM can influence each of the cancer hallmarks. Moreover, abnormally regulated ECM structure and dynamics is a hallmark of cancer. The ECM not only promotes cellular transformation and metastasis but also alters stromal cell function to create a tumorigenic microenvironment186, 251.    21   The ECM is composed of a collection of multiple proteins and polysaccharides, which assemble into three dimensional supramolecular structures that can regulate cell growth, survival, motility and differentiation252, 253. The ECM is dynamically remodeled and its composition and biophysical properties, such as stiffness and compliance, are tailored to the particular physiological state of each tissue254, 255. For example, chondromodulin-I, an ECM component that is found only in cartilage, is a strong inhibitor of angiogenesis and is responsible for the avascular nature of cartilage256, 257. During bone mineralization, another ECM protein, the glycoprotein osteonectin (secreted protein acidic and rich in cysteine (SPARC)), initiates mineralization by binding collagen and Ca2+ 258. Therefore, bone morphogenesis is dependent on the ECM; this has also been shown for other tissues, including intestine, lungs, and mammary glands259.  In healthy epithelial and endothelial tissue, cells are usually separated from the underlying ECM, which is also called the stroma or interstitial matrix, by a specialized, compact 50- to 100-nm thick ECM layer called the basement membrane (Figure 2, left). The basement membrane and stroma are both types of ECMs, but differ in their biochemical and biophysical properties260. The basement membrane, whose major components are type IV collagen, laminin, FN, nidogen/entactin, and perlecan, provides structural support in addition to shielding epithelia, endothelia, peripheral nerves, muscle cells, and fat cells from signalling ligands, and serving as an anchorage site for adjacent cells. The composition of the basement membrane and its physical properties varies between organs and tissue types261, 262. The interstitial stroma is produced by mesenchymal fibroblastic cells (stromal cells). It contains high levels of fibrillar glycoproteins such as collagen I and II as well as FN263. These two structures, the basement membrane and the stroma maintain the integrity of the tissue, enforce apicobasal cell polarity. They are also critical   22  for the function of the parenchyma of organs and play important roles during wound healing and embryonic development264.    Figure 2: Loss of tissue architecture and ECM remodeling during cancer progression.  A feed forward loop of cancer cell intrinsic factors, the ECM and cancer-associated fibroblasts (CAFs) promote cancer progression. Cancer cells in contact with interstitial ECM prime CAFs to remodel the ECM into a tumor-promoting environment. The resulting altered biochemical and biomechanical properties of the ECM further increase EMT and cancer progression.   Around tumors, the dynamic regulation of ECM is often altered and displays desmoplasia. Desmoplasia is a fibrotic state characterized by altered ECM composition, increased deposition of ECM components, and enhanced post-translational modifications of ECM-proteins186. Typically, desmoplasia is only associated with malignant neoplasms and correlates with tumor progression and poor disease prognosis265-267. In these pathological stages,   23  the organization of the ECM is disrupted, the basement membrane is often reduced in thickness or degraded242, and epithelial cells come in direct contact with the ECM of the interstitial stroma (Figure 2, right). At this point, transformed epithelial cells may undergo EMT243. Tumor cells that have undergone this transition are recognizable by their morphology (e.g. loss of polarity, spindle-like, elongated shape) and by the expression of EMT markers (e.g. vimentin, α-smooth muscle actin, Snail)66, 67. This ‘breach of barrier’ and the loss of homeostatic equilibrium has consequences for the cancer cells and for the stromal cells (now referred to as myofibroblastic cancer-associated fibroblasts (CAFs)), or tumor-associated fibroblasts (TAFs)). It alters the biochemical and biomechanical properties of the interstitial stromal ECM242, leading to a state referred to as a ‘primed’ or ‘activated’ stroma. Cancer cells, CAFs and ECM can then together act in a feed-forward loop that promotes the formation and expansion of the desmoplastic ECM, which in turn further promotes cancer progression (Figure 2, circle).  CAFs are the main producer of ECM components in the tumor microenvironment and, when activated, produce increased levels and altered forms of ECM structural proteins that are characteristic of desmoplasia. This includes ECM-remodeling proteins such as matrix metalloproteases (MMPs), lysyl oxidase, and fibroblast activating protein268-275. In addition, cancer cells themselves can secrete some of these factors and further activate CAFs, thereby contributing to matrix remodeling. A prominent sign of a progressing cancer is the transition from normal ‘curly and isotropic’ collagen fibers to more straightened and parallel fibers around the tumors. In the invasive state, these ‘thickened and aligned’ fibers provide tracks for cell migration (Figure 2)276-278. As the fibers align, ECM stiffness increases, which exacerbates tumorigenic changes in cancer cells, promotes tumor growth and migration, and triggers CAFs to further promote desmoplastic ECM remodeling. A key component in the ECM remodeling–  24  cancer progression feed-forward loop is increased integrin-mediated adhesion and signalling. These processes transduce mechanical forces into the cell and activate the Rho-associated protein kinase (ROCK), FAK, and the Yes-associated protein (YAP) signalling pathways that promote EMT and activate CAFs.   The changes in the ECM that create a tumorigenic microenvironment affect most of the hallmarks of cancer: self-sufficiency in growth, insensitivity to growth inhibitors and evasion of apoptosis, limitless replicative potential, sustained angiogenesis, and tissue invasion and metastases251. For example, in response to an aligned, stiffened matrix, integrin signalling promotes increased FAK phosphorylation and activates the extracellular signal-regulated kinase (ERK), PI3K and Rac, resulting in increased cell proliferation279-282. This results in the expression of genes that positively correlate with a proliferative gene signature279, 283. At the same time, stiff ECM reduces the expression of cell cycle inhibitors284-286, as well as the tumor suppressor PTEN. Decreased PTEN expression results in increased PI3K/Akt-dependent cell survival and growth287. Furthermore, ECM-dependent integrin signalling induces the expression of several of the anti-apoptotic Bcl-2 family members288-290, promotes anchorage-independent growth via activation of NF-кB, and suppresses p53-induced apoptosis in response to DNA damage291-294. In addition to these direct effects on cancer cell growth, the ECM also promotes tumor growth by stimulating angiogenesis. Growing tumors are in constant need of oxygen and nutrients295. Stiffer tumor-associated ECMs promote neo-vascularization by increasing endothelial cell migration, growth, survival and vessel formation296-300. Taken together, these findings indicate that tumor cells interacting with a stiff ECM have distinct growth advantages and that the ECM plays an important role in tumor progression.     25  1.4.1 ECM and cancer cell metastasis One cancer hallmark in which the ECM plays a major role is tissue invasion and metastasis. As previously mentioned, CAFs, the ECM, and cancer cells engage in a feed forward loop that promotes tissue invasiveness301. Once normal tissue structures are compromised, tumor cells can migrate towards the vasculature and intravasate into blood or lymphatic vessels where they can persist as circulating tumor cells (CTCs)302-306, in some cases for years307, 308. The CTCs then can extravasate from the circulation into a secondary tissue and establish a secondary tumor, completing the metastatic cascade309. Tumor cell extravasation is highly dependent on cell adhesion. In melanoma cells the expression levels of specific integrins correlate with tumor progression. For example, α4β1 integrins, expressed on CTCs, interact with the vascular cell adhesion molecule (VCAM) on endothelial cells, potentially allowing the arrest of the CTCs and extravasation310-314. However, the nature of the ECM at the premetastatic and metastatic niches is important for establishing tumor metastases250. As well, ECM stiffening and increased deposition of ECM components such as FN can induce the angiogenic switch and promote tumor cell growth as it does at the primary tumor site, and lead to the expansion of tumors from micrometastases to macrometastases250.  1.4.2 ECM and tumor cell migration A key aspect of the metastatic cascade is the ability of tumor cells to migrate. At some point in their development, all nucleated cells have the ability to migrate. For most cells, migration is confined to morphogenesis and ends upon terminal differentiation of the cell. However, cells such as immune cells migrate throughout their lifespan. Cancer cells can adopt several modes of migration. This is true for cells that migrate as singlets (amoeboid or   26  mesenchymal motility) and as cohorts (cell chain or collective sheet motility)315, 316. Cells migrating in a mesenchymal mode typically exhibit a spread morphology, higher levels of cell-matrix attachments involving focal adhesions (FAs), prominent actin stress fibers, and leading edge structures such as lamellipodia and filopodia317. These cells rely on cell protrusions, such as pseudopodia, filopodia and lamellipodia to form cell-ECM adhesions. Traction forces, generated by the Rho/Rock pathway and by actomoyosin contractility mediate forward movement and retraction of the trailing edge318 (Figure 3, left).     Figure 3: A comparison of mesenchymal and amoeboid modes of single cell migration.  Single cells can migrate in mesenchymal (left) or amoeboid (right) mode. Some cells can switch modes by undergoing either mesenchymal to amoeboid transition (MAT) or amoeboid to mesenchymal transition (AMT). ECM fibers are represented in green. Based on Friedl and Wolf, and Pankova et al.316, 319.   Amoeboid migration refers to the movement of round or ellipsoid cells, and is characterized by weaker cell-matrix adhesions, a lack of mature FA, and lack of stress fibers320, 321. Although less defined than mesenchymal migration, amoeboid migration still relies on a   27  balance of protrusion, contractility and adhesion320, as well as actin nucleation and myosin contractility322. Different sub-types of amoeboid migration are distinguished by their leading edge structures, which form either bleb-like (blebby type) protrusions or poorly defined actin-rich filopodia-like protrusions (pseudopodia-like type)316 (Figure 3, right).  Cells have the ability to switch migration modes in response to changes in intrinsic factors, intracellular processes, or the biophysical and biochemical properties of the ECM they encounter323-328. For example, whether single cells migrate in mesenchymal mode after undergoing EMT, or undergo so-called mesenchymal-to-amoeboid transition (MAT) and migrate in amoeboid mode, is dependent on the type of cell morphology, the mechanism of force generation, cytoskeleton organization, physical confinement, and characteristics of the cell-substrate interactions329. When ECM-cell adhesion strength is weak, for example due to increased elasticity of the ECM or low integrin-ligand density, reduced expression of adhesion molecules, or interference with the integrin-talin axis330, most cells move via amoeboid migration. In contrast, when these factors are strong (increased ECM stiffness, high integrin-ligand density, increased adhesion), cells migrate via mesenchymal mode. The mechanisms underlying the plasticity between motility modes, which is determined by how cells integrate and adapt to stimuli from their environment (e.g. adhesion strength, physical confinement)329, 331, is only starting to be understood. It has been proposed that the ability to switch between modes of migration in response to different ECM geometries enables tumor cells to escape their site of origin and metastasize to distant organs.     28  1.4.3 ECM stiffness and adhesion-ligand densities regulate tumor cell behavior ECM composition, density and biophysical properties such as stiffness and fibril orientation control many cellular processes including cell motility332 and tumor progression242, 333. To study the effects of ECM stiffness, ECM density, material porosity, and ligand-substrate coupling (also referred to as ligand tethering) on cellular behavior is challenging in in vitro systems because modifying ECM stiffness, composition and architecture independently is technically difficult. For example, increasing material stiffness can produce secondary effects in the other variables, such as changes in the ECM ligand density or variation in ECM porosity, each of which can affect cell behavior on its own334. Recent studies have developed tissue mimetic systems that address these issues and allow each property to be manipulated independently. By controlling the porosity and stiffness of polyacrylamide gel substrates, Wen et al.335 were able to dissect the contributions of ECM stiffness and adhesive-ligand tethering (the latter affected by porosity and ligand-substrate coupling) on stem cell differentiation. They concluded that stiffness, rather than ligand tethering, regulates cell behavior. However, if material stiffness is kept constant, differences in the concentrations of specific adhesive ligands, such as increasing amounts of collagen or FN in hydrogels, can affect cell behavior336. Similarly, Chaudhri et al.337 were able to decouple ECM stiffness from ligand density and examine the effects of these parameters on mammary epithelial morphogenesis that results in the assembly of acini. By combining synthetic (polysaccharide alginate) and natural (reconstituted basement membrane) ECM materials to form hybrid ECMs in which stiffness can be controlled independently of ligand density, they showed that increasing matrix stiffness was sufficient to drive a malignant phenotype characterized by increased proliferation, reduced apicobasal polarity and matrix invasion337. In contrast, increasing the concentration of laminin, a major   29  component of the basement membrane, blocked the ability of increased stiffness to induce a malignant phenotype337, 338. These findings illustrate that ECM stiffness and adhesion ligands coordinate cell behavior as well as cancer progression, and that changes in the ECM can drive tumor progression via interdependent biophysical and biochemical properties.   1.5 Integrin-containing adhesion complexes couple the ECM to cellular functions Most cells have interactions with neighboring cells or form adhesive interactions with the ECM. Cell adhesion to the ECM is often mediated by integrins, a class of transmembrane receptors that bind components of the ECM such as FN, vitronectin, osteopontin and collagen. In migrating cells there are different sub-types of adhesions, including focal complexes, nascent, growing and mature focal adhesions (FAs), fibrillary adhesions, and podosomes339. These cell-ECM adhesion complexes vary in their location along the cell-ECM interface, their size, shape, protein compositions, dynamics, and turnover rate340, 341. Many, but not all, types of adhesions may be present within the same cell at the same time. Sub-types of adhesions differ between cells and tissue type and can dynamically transform into each other, depending on the physiological state of the cell (e.g. cell migration).  These adhesion complexes are most often nucleated at the site of integrin-ECM contact342, 343. Integrins are transmembrane proteins that span the cell membrane and connect the ECM to the cytoskeleton of the cell344. In mammals, integrins are α-β heterodimers composed of different combinations of 18 α subunits and 8 β subunits to produce a total of 24 different integrin molecules345. Integrin function and ligand specificity can be regulated on multiple levels, including different α-β subunit pairing, expression levels on the cell surface, and by conversion to an activated state that can bind ligands. Together, this influences what type of ECM proteins a   30  cell can bind to as well as the strength of adhesion346-350. For example, there are 8 α-β heterodimers that can bind FN, which have specific and redundant functions351-359.  All integrins share a conformational mechanism that regulates their activation state, i.e. their ability to bind ligands360. The general structure of an integrin can be described as a head structure that protrudes from the cell’s plasma membrane, a transmembrane domain and two legs that extend into the cytoplasm (Figure 4A, left). Crystal structures of integrins have revealed three conformational states representing different degrees of activation and different affinities for ligands. In the inactive state, integrins have low affinity for the ligand, the headpiece is bent over and the cytoplasmic legs are in close proximity to each other. The binding of adaptor proteins to the cytoplasmic domain of the integrin β subunit can convert the integrin to a ‘primed’ or activated state, in which the cytoplasmic domains of the two subunits separate and the ligand-binding extracellular domain is converted to an upright conformation361. In this state, the integrin has a high affinity for its ligands. A third stage, with the ligand bound, is characterized by an open headpiece, the separation of the cytoplasmic domains being more stabilized, and a high affinity for the ligand345, 362-364 (Figure 4A). This model is still being refined, especially with regards to whether integrin activation occurs after binding to the ligand (the deadbolt model365-367) or if ligand binding only occurs after the integrin has transitioned into the ‘primed’ state with the extended conformation (the switchblade model368 369-371). Electron microscopy studies and x-ray structures, although not entirely consistent, tend to favor the switchblade model, in which integrins must be primed before they can bind ligand 360, 372.     31   Figure 4: Model of integrin activation and the integrin adhesome.  (A) Schematic representation of the inactive, primed, and activated ligand-bound state of an integrin molecule. (B) Inside-out and outside-in integrin signalling. Panels A and B are based on Bouvard et al.373. (C) Clustered integrins serve as anchor points for adhesome proteins, many of which are phosphorylated. Recruited adhesome proteins together with actomyosin contractility mediate mechanotransduction, mechanosensing and signalling, which regulate physiological states of the cell.    32  A noteworthy feature of integrins is their ability to signal bidirectionally374, 375. The interplay of ‘inside-out’ and ‘outside-in’ signalling coordinates integrin affinity with integrin signalling capabilities 376. During ‘inside-out’ signalling, scaffolding proteins such as talins and kindlins 377 bind to the integrin cytoplasmic tails, inducing a conformational change that causes the integrin to transition to a state that has a higher affinity for its ligand345, 371, 378, 379 (Figure 4B). ‘Outside-in’ signalling requires ligand binding to a primed integrin dimer (Figure 4B). This ligand-bound and activated integrin state leads to integrin clustering and the formation of integrin adhesions, which contain signalling enzymes. Integrin adhesions that recruit signalling proteins and structural proteins, are collectively referred to as the integrin adhesome339. Many of these recruited adhesion proteins become phosphorylated, a modification that promotes protein-protein interactions and activates enzymes such as FAK and Src380. Figure 4C highlights some of the adhesome proteins that link integrins to signalling pathways and to the cytoskeleton. Most integrin adhesions share some common proteins, such as talin, FAK, Src, Crk-associated substrate of 130 kDa (p130Cas), vinculin and paxillin, although their abundance and order of recruitment differ among adhesion types, and for different integrins381-383.  The binding of integrins to the ECM generates forces that are transduced across the plasma membrane to the adhesome. Adaptor proteins within the adhesome such as talin and p130Cas then change their conformation to expose cryptic protein-binding sites, allowing other adhesome proteins to bind215, 217. This enables cells to transfer ‘tension’ information from their surroundings to the inside of the cell. The subsequent recruitment of proteins such as paxillin, zyxin and vinculin to the adhesome, a process referred to as adhesion maturation, is further enhanced by internal forces that are generated by actomyosin contractility239, 240. Thus, integrins   33  and adhesomes transduce information from the extracellular environment, specifically the ECM, to signalling pathways. These signalling pathways control complex cellular behaviors such as cell migration, cell survival, cell cycle progression, cell fate, maturation and differentiation384, 385 (Figure 4C).  1.6 The role of integrin signalling in cancer Dysregulation of integrin adhesion-mediated signalling pathways can contribute to cancer progression. For example, a chronically-elevated outside-in/inside-out FAK-Rho signalling loop creates and maintains an invasive phenotype in mammary epithelial cells, activates proliferation signature genes, and correlates with poor disease outcome386. Interestingly, these effects are strongly influenced by mechanical properties of the matrix such as rigidity and tensile strength, illustrating how the activation of integrin-dependent pathways is tightly linked to the translation of external forces to cytoskeletal forces that activate signalling pathways386. Integrin adhesion-mediated changes in gene expression in the context of both development and cancer are associated primarily with a proliferative phenotype. However, other adhesion-induced changes promote proliferating tumor cells to become more aggressive and metastatic. For example, in human melanoma cells, the activation of some integrins has been associated with altered expression of MMPs387, 388, enzymes that remodel the ECM at the invasive front of migrating cancer cells. Thus, integrin-mediated adhesion signalling can promote tumor progression by regulating genes related to proliferation, tumor cell survival, and metastasis-related processes such as cell migration and invasion.      34  1.7 Adhesome proteins and cancer  By comparing different types of integrin adhesomes across multiple cell types, Geiger et al.339 defined a ‘canonical adhesome’ that consists of adhesion-associated molecules and includes their interactions with each other (www.adhesome.org). As currently defined, the canonical adhesome is comprised of 156 components and 690 interactions. Most of the components are proteins but the list also includes lipids and second messengers such as Ca2+ ions. The members of the adhesome network include kinases and phosphatases, proteases, GTPases and their regulators, adhesion receptors, cytoskeletal regulators, and lipids, such as PIP2 and the PI3K product PIP3339.  Modulating the expression or activation of adhesome components has provided insights into their individual functions as well as the pathological consequences of altered adhesomes389-392. Using bioinformatic approaches, Winograd-Katz et al.393 linked specific adhesome components to molecular functions in diseases so that this information could be used to identify disease models and potential therapeutic targets. A prominent finding from their studies was that disease-related genes were overrepresented in the adhesome and that cancer was the top disease category in terms of being associated with the greatest number of adhesome components. This was true regardless of whether the cancer had a monogenetic cause (Online Mendelian Inheritance in Man (OMIM) database394) or a polygenetic cause (Genetic Association Database (GAD)395). This suggests that dysregulated integrin adhesion signalling drives cancer progression. For example, many tumors exhibit increased activity of the adhesome protein FAK, which promotes tumor cell growth and anchorage-independent growth, and correlates with poor outcome396.    35  1.8 The role of integrins and the adhesome in resistance to chemotherapy The success of many cancer therapies is limited by the development of drug resistance. Recently, signalling resulting from integrin/ECM interactions has been shown to contribute to drug resistance397. Specifically, integrins can drive resistance to radiotherapy and chemotherapeutics in head and neck cancers, breast cancers, and lung cancers398-400. For example, the resistance to the epidermal growth factor receptor (EGFR) inhibitor, erlotinib, a chemotherapeutic used to treat lung cancer correlates with increased expression of β1, α2, and α5 integrin as well as with enhanced cell adhesion. In addition, knocking down β1 integrin in an erlotinib-resistant human non-small cell lung cancer cell line restores erlotinib resistance400. This points to the importance of signalling network crosstalk in cancer treatment, as integrin signalling and growth factor signalling share overlapping signalling molecules (e.g. Src and Akt), and can cross-modulate responses to receptor activation (section 1.6)401. Taken together, this suggests that multiple integrin signalling pathways contribute to cancer drug resistance and that understanding these pathways and their differences between cancer types could lead to better therapeutic strategies.   1.9 Outside-in integrin signalling activates signalling pathways that regulate cell behavior and gene expression   1.9.1 Integrin signalling pathways control cell migration, invasion and adhesion Integrin-mediated adhesion activates downstream pathways that regulate immediate, short-term cell behavior, such as changes of cell shape and the initiation of cell migration. It also regulates long-term adhesion effects including changes in gene expression associated with cell   36  survival, proliferation, and differentiation. Integrin-mediated adhesion signalling leads to activation of downstream signalling enzymes such as FAK, Src and the PI3-kinase, and GTPases such as Ras, Rac and Cdc42.  The FAK and Src tyrosine kinases are central signalling molecules that are downstream of integrins and in turn activate several other pathways (Figure 5). For example, following integrin activation, Src/FAK phosphorylates p130Cas and paxillin. This induces a FAK/Grb-2/Sos complex that leads to the activation of Ras and Erk402. Phosphorylated p130Cas creates a docking site for binding partners including Crk, which results in activation of JNK. In addition, through activation of Crk/Dock180 or PIX/GIT pathways, and the recruitment of the RhoGEF p190, the Scr/FAK complex regulates the activation of GTPases, which regulate many adhesion- and migration-related processes such as FA dynamics, actin dynamics, and actin-dependent formation of lamellipodia and filopodia. Several small GTPases and their regulators (p190, Dock180, C3G, Pix/GIT) have been found in FAs, including Rac, Cdc42, Rap1, and RhoA403 (Figure 5). Activated Rac promotes the formation of lamellipodia, whereas activated Cdc42 regulates cell polarity and induces the formation of filopodia. Both Rac and Cdc42 activate the Arp2/3 complex, which controls the assembly of a branched actin filament network through its actin nucleation function404. Together these pathways coordinate cell migration, invasion, adhesion and polarity, processes that are essential for tumor progression.    37   Figure 5: Integrin ‘outside-in’ signalling governs cellular processes  Modified from Legate et al.403, see text for description.  1.9.2 Regulation of Rap1 during integrin signalling The Rap1 GTPase plays a central role in cell adhesion and migration405-407 by regulating integrin activation408-412, actin cytoskeletal dynamics413, membrane protrusion414, and cell polarity406. Rap1 is activated by many receptors including integrins. Activated integrins recruit the adaptor protein Crk, which constitutively associates with the Rap1 GEF C3G. The Crk/C3G complex binds to the adhesome protein p130Cas, which is recruited, often via FAK, to the cytoplasmic domains of activated integrins. In response to cytoskeletal tension that is generated   38  by the forces that arise from the link between the ligand-binding integrin and the cytoskeleton, p130Cas reveals cryptic tyrosine phosphorylation sites. These sites are subsequently phosphorylated by Src, creating binding sites for Crk216. Rap1 is therefore activated by mechanical transduced signals and may also serve as a mechanosensor that acts downstream of adhesion-dependent integrin activation. Once activated, Rap1-GTP can bind multiple effector proteins and with that further activates integrins through RIAM/talin (see section 1.10.2 and 1.10.3) and promotes actin polymerization. Activated Rap1 also promotes the activation of Cdc42 and Rac, which in turn induce the formation of lamellipodia and filopodia.  Loss- and gain-of function experiments have shown that Rap1 is important for integrin function. Overexpression of constitutively active Rap1 increases integrin activation, and results in increased adhesion. Conversely, preventing Rap1 activation reduces integrin activation415. Importantly, Rap1 is a sensor and regulator of FA responses to tensional changes. Upon external physical stretch or shear forces, Rap1 is activated and regulates the association of vinculin with FAs. This makes Rap1 an important molecular mediator of adhesome changes in response to mechanical loads416.   1.9.3 Integrin signalling as a regulator of transcription Integrin-mediated signalling also regulates cell proliferation and differentiation by mediating the response to growth factors417. For example, in human intestinal smooth muscle cells, αvβ3 integrin fine-tunes the response to the activation of growth factor receptors such as the insulin-like growth factor-1 receptor (IGF1R) and epidermal growth factor receptor (EGF1R)418. The basis for the crosstalk between integrin signalling and growth factor signalling   39  is that both can activate the same pathways including the PI3K/Akt, Mek/ERK, Rho and Rac pathways.  Integrin activation promotes cell growth by activating the Akt and ERK pathways. Downstream of the Src-FAK complex, ERK is phosphorylated through activation of the Ras-Raf module. Akt activation is a consequence of the production of PIP3 by PI3K. This leads to phosphorylation of the Akt pro-survival kinase and its activation loop by PDK1, which is recruited to the plasma membrane following integrin activation419. Further activation of Akt requires the phosphorylation of another key residue in Akt by the mammalian target of rapamycin (mTOR) and ILK kinases420, 421 (Figure 5). Both activated ERK and Akt contribute to cell cycle progression by controlling the activation of the Elk TF and inducing the expression of cyclin D1, which allows for transition through G1/S phase422. Activated Akt also supresses the cell cycle inhibitor FOXO3a, further promoting cell cycle progression423.  In addition to ERK-induced proliferation, integrin-dependent cell-matrix adhesion also promotes cell proliferation by stimulating the expression of immediate-early genes (c-myc, c-jun, c-fos)424 425, 426 in a c-jun-N-terminal kinase (JNK) dependent manner427 (Figure 5). Expression of immediate-early genes after mitogen stimulation does not require de novo protein synthesis and is mediated by pre-existing transcription factors. Following adhesion to ECM, JNK activates members of the ternary complex factor (TCF) family, which act as transcriptional cofactors for the TF serum response factor (SRF). This stimulates SRF-targeted immediate-early gene transcription, and subsequently cell proliferation427.  Integrin-mediated activation of GTPases also contributes to cell growth by regulating adhesion-dependent cell cycle progression428, 429. For example, Rac1 promotes cell cycle progression via several pathways including the recruitment and activation of ERK (through   40  Pak1) and JNK430, 431, as well as by suppressing cell cycle inhibitory pathways (e.g. Skp2-p27)432.  Integrin-adhesome signalling also regulates the expression of cytoskeletal regulators. Many of the genes that can be induced by SRF, in addition to the immediate-early genes that drive cell proliferation, encode for proteins that are part of the adhesome (e.g. β1 integrin, vinculin, talin), components of the actin cytoskeleton (e.g. several isoforms of actin and myosin), or regulators of actin dynamics (e.g. cofilin 1, gelsolin)433, 434. Thus, SRF and other transcriptional regulators induced by integrin signalling pathways have a central role in regulating cell morphology and cell motility. Other genes that are induced downstream of integrin signalling encode for proteins that are crucial for tissue formation and remodelling. In fibroblasts for example, integrin-induced signalling via ERK or JNK is transmitted into the nucleus where they activate TF AP-1, which induces type I collagen and osteopontin expression435-437. Integrin signalling-induced transcriptional changes include regulation of genes that regulate cell proliferation, cytoskeletal regulation, FA dynamics, and ECM remodelling, all of which can contribute to tumor development and progression.  1.9.4 Actin dynamics link integrin function to changes in gene expression Cytoskeletal dynamics play a key role in transducing adhesion-dependent signals into changes in gene expression. For example, the ratio between filamentous actin (F-actin) and globular actin (G-actin) can regulate the shuttling of TFs and their regulators into the nucleus and thereby regulate transcription. Because integrins are potent regulators of actin dynamics, integrin adhesomes are important regulators of actin-dependent transcriptional regulation. Changes in   41  actin dynamics in response to integrin activation and ligand binding is mediated by the active forms of the Rap1, RhoA, Cdc42 and Rac1 GTPases. Effectors of these GTPases, including ROCKs, formins, WASP/WAVEs and the Arp2/3 complex orchestrate actin polymerization by promoting the addition of G-actin to actin filaments438.  The balance between G-actin and F-actin regulates the activity of the F-actin binding proteins (F-ABPs), G-actin binding proteins (G-ABPs), and F-actin binding complex-associated proteins (F-ACAPs), such as FA-complexes. G-ABPs only bind G-actin and are retained in the cytoplasm when G-actin levels are high. However, they translocate into the nucleus upon the formation of F-actin microfilaments and depletion of cytosolic G-actin. Architectural rearrangement of F-actin fibers then lead to the release of F-ABPs and F-ACAPs, allowing them to translocate into the nucleus to act as co-regulators of transcription (Figure 6). For example, the incorporation of G-actin into F-actin liberates members of myocardin-related transcriptions factors (MRTFs), which translocate to the nucleus and promote the expression of specific subsets of SRF-dependent genes439. Other examples of G-ABPs include profilin, Arp2/3, and Neuronal Wiskott–Aldrich Syndrome protein (N-WASP). F-ABPs include LIM domain proteins such as cofilin, gelsolin, filamin, paxillin and α-actinin. Examples for F-ACAPs are the LIM kinase domain proteins LIMS1 and FHL2. Together, integrins and other plasma membrane receptors, regulate actin and actin binding proteins (ABPs) to create a cytoskeleton-to-genome relay system by which cells communicate information from the ECM to the genome via dynamic changes of cytoskeletal structures. This allows for the regulating of cellular functions in response to changes in the ECM environment.    42  1.9.5 Mechanosensitive control of transcription regulation Many TF pathways downstream of ECM-cell contact are strongly regulated by mechanical forces. When cells experience mechanical strains, forces are transmitted through integrins (or other cell-ECM and cell-cell receptors) to intracellular signalling pathways, resulting in the generation of second messengers (e.g. Ca2+) and dynamic changes in the cytoskeleton. This network of mechanochemical signalling controls the activation of transcriptional regulators such as MRTF-A, YAP/TAZ, KLF2, GATA2/4, TFII-I, AP-1, STAT5, MKL1/2, NF-кB, SRF, and CREB440. For example, ECM-bound integrins induce Ca2+ influx through mechanosensitive transient receptor potential cation channel subfamily V members (TRPM). Cyclic mechanical stretch applied through ECM-bound integrins receptors induces Ca2+ influx within 5 mseconds after application of the mechanical force441. The resulting influx of cations stimulates PI3K signalling, which promotes cytoskeletal remodeling, the activation of additional integrins and driving transcriptional changes via ERK1/2 activation. In neurons, this activates CREB TF, which regulates the expression of genes that are essential for neuronal survival and for memory formation442.  Direct connections between the ECM and the nucleus are created by proteins that are called linkers of nucleoskeleton and cytoskeleton (LINCs). LINCs (e.g. SUN proteins) are complexes that are embedded within the nuclear envelope via lamins on the inside on the nucleus and link the nucleoskeleton to the actin cytoskeleton (Figure 6). Sun proteins bind to nesprins on the cytosolic side, which create anchor to F-actin. This provides a direct link from the integrin adhesome to nuclear components and allows for rapid propagation of external forces to the nucleus443, 444. This link places nuclear envelope proteins at the crossroads between extracellular   43  signalling and transcriptional responses. However, how TFs are controlled by mechanical force transmission via nuclear envelope proteins is poorly understood.   Figure 6: Connections from the ECM to the genome. Schematic illustration highlighting important routes by which integrin-dependent signalling relays changes in the ECM to the nucleus, and regulates gene expression. For description see text. Integrin-dependent signalling triggers pathways that cause signalling molecules and transcriptional regulators (e.g. ERK1/2, JNK) to enter the nucleus via the nuclear pore complex (NPC). ABPs, once dissociated from the cytoskeleton after integrin-induced changes in actin dynamics, also translocate into the nucleus. Members of the LINC family (e.g. SUN1/2) bind F-actin via nesprins, linking the nucleus to the integrin adhesome, allowing for the transmission of mechanical forces from the ECM to the nucleus. This may affect gene expression by promoting chromatin remodelling via mechanisms that are not fully understood. Several members of the LINC family, as well as nesprin isoforms also connect to microtubules and intermediate filaments, but are not shown here.     44  1.10 The role of talin and FAK in adhesome function and integrin signalling  Adhesome proteins act in concert to regulate adhesion dynamics and adhesion-dependent responses, such as changes in cell shape, motility and cell cycle progression. Adhesome components regulate the assembly, maturation and disassembly of adhesion complexes as well as their signalling functions and connection to the cytoskeleton. Two key proteins that are found in most adhesions are talin (of which there are two isoforms in vertebrates, talin 1 and talin 2, which are described in more detail in chapter 6), and FAK. Both talin and FAK are recruited to adhesions and regulate integrin signalling, cell morphology and motility. Talin is a multi-domain scaffolding protein, that can bind to the β-tail of integrins, and is a key regulator of integrin activation and clustering445, 446. FAK is a kinase as well as a scaffolding protein and its functions have been studied extensively in the context of integrin signalling208, 447, 448. Both proteins, as well as their binding to each other at integrin adhesions are required for adhesion turnover, cell motility and integrin-stimulated cell cycle progression449, 450.   1.10.1 Talin is a scaffolding protein with binding domains for adhesome proteins  Talin is conserved throughout metazoans and is essential for developmental processes in organisms ranging from the multicellular slime mold Dictyostelium discoideum to vertebrates451-457. Talin knockout mice die at the gastrulation stage456. In vertebrates, talin plays a key role during integrin activation by regulating the affinity of integrins for their ligands and promoting integrin clustering. From a structural point of view, talin is composed of two multi-protein binding regions that are connected by an unstructured linker region458. The talin head domain is an atypical FERM (F for 4.1 protein, E for ezrin, R for radixin and M for moesin) domain with binding sites for actin, β-integrin, the Cdc42-activating GTPase exchange factor TIAM1459,   45  FAK460, the cytoplasmic tail of the hyaluronan receptor layilin, and type 1 PIP-kinase γ-isoform (PIPKγ )461-464. PIPKγ plays a major role in the synthesis of PIP2, which is required to orientate talin such that it can activate integrins465. The talin head domain is connected via a linker466 to the talin rod, which is organized into thirteen 4- or 5-helix bundles458. The head and rod domains can be dissociated by calpain 2-mediated cleavage at a site in the linker467. The rod domain contains multiple vinculin binding sites (Figure 7, indicated in red)468, several binding sites for the Rap1A effector Rap1-GTP-interacting adaptor molecule (RIAM)458, the RhoGAP DLC1469, and the intermediate filament binding protein synemin470. In addition, the talin rod domain harbors several additional actin binding sites and integrin binding sites471, 472. At the C-terminus of the rod domain is a single helical domain that allows for talin dimerization and can which be cleaved off by calpain 2.    Figure 7: Domain organization of talin 1. The talin head contains an atypical FERM domain consisting of the F0, F1, F2 and F3 sub-domains. It is joined to the rod domain by an unstructured linker. The rod is composed of thirteen 4- or 5-helix bundles (numbered blue boxes), followed by a single helical dimerization domain (DD). The positions of the calpain 2 cleavage sites are indicated with purple stars. Protein interaction sites for talin-binding proteins and lipids are indicated. FAK, focal adhesion kinase; DLC1, deleted in liver cancer 1; PIP2, phosphatidylinsotitol-4,5-bisphosphate; PIPγ; TIAM1, T lymphoma invasion and metastasis-inducing protein 1. Graphic based on Calderwood et al.473.    46  1.10.2 Talin as an integrin activator  The role of talin in regulating integrin activity has been studied in cell culture systems474-477, transgenic mice478, 479 and purified protein systems480. Molecular details of talin-induced integrin activation have emerged from several studies and have led to a model for talin-mediated integrin activation481-483. Upon activation, the F3 sub-domain of the talin head domain binds a NPXY motif in the β-integrin cytoplasmic tail. The interaction between the F3 sub-domain and the β-integrin tail is thought to weaken the inhibitory conformation of the integrin heterodimer by altering the positioning of the β-integrin transmembrane domain484 and the disruption of the contacts between the α-chain and β-chain in both the transmembrane domain and cytoplasm485. Interactions between talin and the membrane are also necessary for integrin activation. This is mediated by the talin head domain’s ability to bind to acidic plasma membrane phospholipids such as PIP2, which orients the talin head domain such that it increases the affinity of talin for integrins and allows talin to induce integrin activation465, 486, 487.   1.10.3 Regulation of talin-induced integrin activation via autoinhibition  Talin can exist in a globular and an extended conformation. The globular conformation, where the F3 domain binds to the R9 domain in the rod, is auto-inhibited, since the F3 domain is prevented from binding to membrane-proximal region of β-integrin. The result is that autoinhibited talin is primarily located in the cytosol488-490. Talin may also form an inhibitory dimer in which the two rod domains surround the head domains in a donut-shaped structure491. There are several models for how talin is activated and translocates from the cytosol to the plasma membrane. In one model, talin is recruited to nascent adhesions by binding to FAK. The binding of talin to FAK may induce a conformational change that exposes the talin head domain   47  so that it can induce integrin activation. Alternatively, activated Rap1A and its effector RIAM can recruit talin to the cytoplasmic domain of integrins361, 475, 476 (Figure 8A). ECM-dependent integrin activation results in increased Rap activation, and Rap1-GTP binds to the RA domain of RIAM. Through its N-terminal region, RIAM binds to talin at the same time, therefore bridging activated Rap1 to talin. RIAM also contains pleckstrin homology (PH) domains, suggesting that the translocation of RIAM-talin complexes to the plasma membrane occurs primarily in PIP2-rich microdomains492. This could contribute to the spatial regulation of integrin activation.   Figure 8: Talin exposes vinculin-binding sites in response to force-induced conformational change.  (A) Autoinhibited talin is recruited to the plasma membrane in a Rap1-dependent manner. The vinculin binding sites within talin are not accessible. (B) Through interaction with the membrane phospholipids the autoinhibition of talin is relieved. The talin head can bind to integrin and the talin rod is free to capture the retrograde flow of actin. (C) When talin bridges integrin to actin, forces exerted on talin cause conformational changes within the R3 domain that weaken the biding to RIAM and expose vinculin binding sites. Vinculin can now bind to talin and actin strengthening the adhesion. (D) Greater forces expose additional vinculin binding sites, which allows for more crosslinking of talin to actin by vinculin. Modified with permission from Yao et al.217     48  1.10.4 Talin as a mechanical link between integrins and the cytoskeleton   A major function of the talin rod is to connect integrins to the actin cytoskeleton, allowing for ECM-dependent forces to be transduced to the interior of the cell493. Structural analysis of the talin rod domain showed that the conformation of the rod changes in response to forces exerted on the integrin-talin-actin axis and that this affects the binding of the integrins to their ligands as well as the binding of other adhesome proteins to talin. For example, some of the vinculin binding sites within the helical bundles of talin (Figure 7, red in blue boxes) are normally buried within talin494 and are only exposed upon mechanical force495. Vinculin binding to these cryptic sites then inhibits talin refolding after force is released, allowing for sustained integrin activation217. Recently Yao et al.217 proposed a model in which the Rap1-GTP-RIAM complex recruits inactive talin to the plasma membrane (Figure 8A). Via its head domain, talin can then bind to the integrin β-tail and activate integrin. The rod domain of talin then binds to actin, allowing actomyosin driven forces to expose more vinculin binding sites. When talin binds to the β-integrin tail via its N-terminal head domain but does not bind to F-actin with its C-terminal actin-binding site, the forces experienced on talin are very low and no vinculin binding sites are exposed (Figure 8B). When integrin-bound talin also captures actin fibers that undergo retrograde flow (e.g. at the leading edge), pulling forces (tension) are exerted on the talin molecule, that cause the R3 rod domain to unfold, which reduces RIAM binding and exposes two vinculin binding sites. Vinculin can then bind to talin and F-actin, crosslinking the two structures (Figure 8C). This leads to an initial talin/actin linkage as present in nascent adhesions. Newly linked actin fibers may result in increased forces (e.g. by actin-myosin contractility) on talin and exposes more vinculin binding sites promoting the formation of larger and more stable focal adhesions with a stronger linkage to the actin cytoskeleton (Figure 8D). If forces exerted on   49  single talin molecules exceed a limit, vinculin is displaced from the talin rod and the vinculin binding sites refolded to a random coil. Taken together, this suggests that talin is a mechanosensitive protein, which in response to force changes its affinity for RIAM, vinculin, F-actin and potentially other adhesome proteins. By directly and indirectly influencing the recruitment of proteins to the adhesome talin regulates adhesion assembly and maturation.  1.10.5 The role of talin in FA dynamics and turnover  In addition to its roles in integrin activation and mechanotransduction, talin regulates adhesion dynamics and turnover. Turnover of FA is essential for mesenchymal cell migration. The disassembly of adhesions correlates with integrin clusters coming apart and integrins being converted to their inactive state. Recent reports suggest that calpain 2-mediated cleavage of talin 1 between the head and the rod domain (Figure 7) promotes FA turnover. Once cleaved, the talin 1 head is either targeted for degradation by SMURF1-mediated ubiquitination, which promotes FA turnover, or stabilized through cell division protein kinase 5 (CDK5)-dependent phosphorylation, which promotes FA stabilization496. FAK seems to be a regulator of calpain 2-mediate talin cleavage since cleavage of talin between the head and tail domain does not occur in cells expressing a mutant form of FAK that cannot bind talin460. Recently a second calpain 2 cleavage site in talin has been identified, which is located N-terminal of the dimerization domain of talin 1 (Figure 7). A point mutation at this cleavage site prevents calpain 2-mediated cleavage, inhibits FA turnover, and decreases the persistence of cell protrusions466. Taken together this suggests that the interaction between talin and FAK is important for FA dynamics.     50  1.10.6 The role of talin in cell proliferation  There is increasing data supporting the idea that talin head and rod, once cleaved, have distinct roles in the cell. The head domain promotes integrin activation, FA turnover466 and is essential for the early phases of cell spreading including integrin activation and Src activation (but not FAK signalling or FA assembly)493. Besides mediating recruitment to FAs, the rod domain may also regulate proliferation in adherent cells. In mammary epithelial cells, knocking down talin 1 leads to cell cycle arrest that is attributable to defective FAK signalling. This phenotype is rescued by re-introducing a talin 1 rod fragment493, suggesting that a talin rod/FAK module is essential for integrin-initiated cell proliferation. Moreover, recent evidence links talin 1 to resistance to apoptotic cell death due to insufficient cell-ECM interaction (anoikis), a critical component of tumor progression and metastasis497. In prostate cancer cells, overexpression of talin 1 enhances AKT signalling, resulting in anoikis resistance and promoting cancer cell invasion. Consistent with this, high talin 1 expression in human prostate cancer samples correlates with disease progression498. Taken together, this suggests that talin 1 contributes to cancer progression by regulating tumor cell survival.  1.10.7 Talin 1 and talin 2   The two talin genes in vertebrates have conserved intron-exon boundaries and encode proteins that have 74% overall amino acid sequence499, 500. Talin 2 is believed to be the ancestral gene, having undergone duplication in chordates prior to the emergence of vertebrates, which express talin 1500. Interestingly, Dictyostelium discoideum also has two talins, TalA and TalB, which have distinct functions. TalA is required for cell-substrate adhesion, phagocytosis and   51  cytokinesis, whereas TalB regulates force transmission to support morphogenetic movements during differentiation501.  Studies in mammalian cells have largely focused on the ubiquitously expressed talin 1, which was discovered over 30 years ago. Talin 1 is the most abundant isoform expressed in lymphoid cells, where it has been studied extensively in its role as an integrin activator. Much less is known about the functions of talin 2. Initial data suggested that in mice, talin 2 expression is restricted to specific tissues, with high levels found in heart, brain and skeletal muscles 502. However, more recent and more sensitive detection methods indicate that talin 2 is more widely expressed than originally believed503, 504.  The tln1 gene spans ~30 kb with 56 coding exons, while tln2 spans >400 kb, with larger exons, several promotors and multiple splice variants505. In mice, disruption of both tln1 alleles results in embryonic lethality at E8.5-E.9.5 due to gastrulation defects506. In contrast, mice with complete deletion of the tln2 coding sequences are viable and fertile, with a mildly dystrophic phenotype. However, they are difficult to maintain as a colony, with variable numbers of pups surviving to adulthood for unknown reasons507, suggesting more subtle developmental defects.   Studies suggest that talin 1 and talin 2 have both overlapping and unique functions. In human endothelial cells and in MEFs talin 2 can compensate for loss of talin 1, supporting cell spreading and FA assembly493, 508. However, in other tissues, such as the heart, distinct patterns of talin 1 versus talin 2 expression suggest non-overlapping roles. During embryogenesis, both talin 1 and talin 2 are highly expressed in cardiac myocytes, but in fully mature cardiac myocytes, talin 1 expression is downregulated and talin 2 becomes the predominant isoform. In healthy adult hearts, talin 2 expression is restricted to cardiac myocytes, whereas talin 1 is the dominant isoform in neighboring endothelial cells. However, when the heart is subjected to   52  pressure overload, talin 1 expression selectively increases in cardiac myocytes and causes hypertrophy, fibrosis, and reduced cardiac function509.  Recently Praekelt et al.510 developed new isoform-specific monoclonal antibodies against talin 1 and talin 2. With these tools, they characterized the subcellular locations of talin 1 and talin 2 in human macrophages, NIH3T3 cells, smooth muscle cells, and MEFs, as well as in tissues such as kidney, muscle, brain, heart and lung. Talin 1 and talin 2 have different subcellular localizations in muscle but co-localize in kidney cells, where talin 1 but not talin 2 is critical for normal function. In NIH3T3 cells, rat aortic smooth muscle cells, and MEFs, talin 1 is primarily localized to FAs, while talin 2 localizes to FAs and to elongated structures along the cell body, which are most likely stress fibers and fibrillar adhesions. Talin 2 staining overlaps with FN and α-integrin staining. Although knocking down talin 2 does not affect FAs, stress fiber formation or cell spreading in NIH3T3 cells, it does reduce the number of cells with FN fibrils. Further data suggests that talin 2 is important for macrophage function. Although it is talin 1 that localizes to podosomes in macrophages, knocking down talin 2 significantly reduces the ability of the cell to degrade ECM despite unaltered expression of matrix metalloproteinases. This suggests a role for talin 2 in the secretion or activation of MMPs. However, the relative roles of the two talin isoforms in cancer cells have not been addressed.   1.10.8 Talin and cancer   As an important regulator of integrin mediated ECM-cell interaction, talin has recently been linked to tumor progression and metastasis511. Talin 1 overexpression is correlated with a metastatic phenotype in oral squamous cell carcinoma512 and in prostate cancer, talin 1 overexpression promotes adhesion, migration and survival, resulting in resistance to anoikis498   53  and bone metastases513. Conversely, knocking down talin 1 in glioma cells reduces cell spreading and migration and impairs the ability of these cells to adapt to changes in ECM stiffness514. Clinically, serum levels of talin 1 have been suggested as a potential screening test for colon cancer515. Similarly, proteomic analyses show that highly metastatic cells express significantly higher levels of talin compared to poorly metastatic cells516. However, depending on the cancer type and stage, talin levels may be either tumor-promoting or reducing. For example, in certain human liver cancer cell lines, high talin 1 expression correlates with reduced invasion and migration, as well as decreased malignancy517. In contrast, talin 1 upregulation is associated with a decreased time to recurrence after resection of hepatocellular carcinomas518. In the human breast carcinoma cell line MDA-MB-231, talin 1 stabilizes invadopodia and knocking down talin 1 reduces cell invasion, intravasation and lung metastases, but causes larger primary tumors519. This points to diverse roles for talin in cancer progression and the metastatic cascade.   Because most studies to date have not distinguished between talin 1 and talin 2, often because talin 1 is more highly expressed in some cell types and there are better reagents for talin 1, the role of talin 2 in cancer progression is significantly understudied. Correlative studies suggest that talin 2 may also be important for tumor cell invasion and migration in some cancers such as liver and breast cancer517, 520. Recent reports show that knocking down talin 2 in human epidermal growth factor receptor 2 (HER2)-positive breast cancer cells reduces migration and invasion. Interestingly, talin 2 expression is directly reduced by miR-194, a microRNA that is upregulated following HER2-targeted trastuzumab treatment520. Taken together, this suggests that targeting talins may hold potential value for cancer treatment.    54  1.10.9 FAK is both a kinase and scaffolding protein FAK, a key component of the integrin adhesome, is a non-receptor kinase that is a core component of FAs521, 522 (Figure 9). FAK is a 125-kilodalton (kDa), multi-domain protein consisting of a central kinase domain that is flanked by proline-rich domains on each side. The N-terminus contains a FERM domain whereas the C-terminus contains a focal adhesion-targeting (FAT) domain. These domains act as platforms for protein-protein interactions that regulate FAK kinase activity (e.g. RET, NANOG, platelet-derived growth factor receptors (PDGFR), EGFR, MET523 , paxillin524) and link FAK to proteins that regulate cell motility (Arp2/3, N-WASP, TM4SF5, cortactin, p130Cas, talin, paxillin, growth factor receptor-bound protein 2 (GRB2), ARHGEF28 and Src) and cell survival (vascular endothelial cadherin (VEC), p53, MDM2, the p85 regulatory subunit of PI3K, endophilin A2, Grb2, ARHGEF28, Src)523 (Figure 9). Both, the kinase and scaffolding functions of FAK are important. FAK-null mice die at E8-E9 due to malformation of organs and blood vessels525-528. Knock-in mice in which FAK has been replaced by a catalytically-inactive version die a few days later (E13.5) and do not exhibit identical defects to the FAK-null mice529, 530.This suggests that FAK’s kinase activity and protein-protein interactions have different functions.     55   Figure 9: Schematic representation of FAK. The central kinase domain is flanked by the FERM and FAT domain. Interacting proteins are shown above and key phosphorylation sites (P) below the schematic. Nuclear localization signal (NLS), nuclear export signal (NES), proline-rich regions (PRRs). Adapted from Sulzmaier et al. 523.   1.10.10 Integrin-dependent FAK activation  FAK plays an important role in the response to integrin-mediated cell adhesion and contributes to the sensing of ECM rigidity208, 531-533. FAK activation has been studied extensively in the context of the integrin-adhesome, which led to the identification of several phosphorylation sites that are crucial for FAK activation. Briefly, ECM-induced integrin clustering leads to FAK autophosphorylation on Y397, which creates a binding site for the SH2 domain of the Src tyrosine kinase. This promotes Src activation and Src then phosphorylates Y567 and Y577 in the activation loop of the FAK kinase domain534-536. Although this is a well-defined mechanism of FAK activation, alternative modes of FAK regulation include Y397 phosphorylation in trans by FAK524, direct phosphorylation of the activation loop by the RET receptor tyrosine kinase537, and activation of FAK by intracellular pH changes, G-protein-coupled receptors, and cytokine receptors523. The mechanism by which FAK is recruited to integrin adhesions and its kinase activity is regulated by adhesome interactions may vary between cell types and types of adhesions. For example, recent studies suggest that the   56  mechanism by which FAK is recruited to nascent adhesion is different than the mechanism by which FAK is recruited to more mature adhesions. This has led to a model in which FAK is recruited to nascent adhesions independently of talin. In fact, FAK may recruit talin to nascent adhesions. In contrast, in mature adhesions, talin seems to be required for the recruitment of FAK. As part of its role in the integrin-adhesome, FAK also regulates the dynamic rearrangement of the actin cytoskeleton. FAK-associated proteins, such as cortactin, talin, Arp2/3 and N-WASP, link FAK to actin dynamics538 and F-actin polymerization539, 540.   1.10.11 The role of FAK in cancer Since the first report of substantially increased expression of FAK in human cancers541, several studies have shown that FAK is involved in many aspects of cancer: angiogenesis, migration, invasion, metastasis, EMT, promoting tumor survival, and maintenance of cancer stem cells542-549. Apart from neuronal tissue, FAK expression is relatively low in most adult tissues550, 551, but it is often upregulated at the mRNA and protein levels in primary and metastatic tumors, a change that is correlated with poor patient survival523, 544, 547, 551, 552. FAK promotes EMT-like phenotypes553, including decreased expression of E-cadherin in breast cancer cells554. FAK also induces transcriptional changes that promote tumor cell invasion and metastasis including upregulating MMP9 expression in spontaneous breast carcinoma555. Many studies have placed FAK as a central signalling molecule of cancer cell growth and regulation. For example, the loss of FAK suppresses tumor formation in mouse models556-559 and also results in the downregulation of several signalling pathways linked to proliferation such as the ERK, PI3K, and Rho/ROCK pathways557-559. FAK also interacts with tumor suppressors such as p53560, 561 and neurofibromin-1 (NF-1)562, as well as oncogenes such as human epidermal growth factor   57  receptor 2 (HER-2)563, MET (the hepatocyte growth factor receptor, encoded by c-Met)564, and Src565.  Recent data suggests that FAK has opposite effects on cell migration and invasion, two processes that are important for tumor metastasis. In B16F10 melanoma cells, FAK directs the location of its binding partner Src to either FAs or invadopodia566. FAK deletion in these cells leads to the absence of active Src in FAs but an increase in Src at invadopodia. This is in contrast to control cells in which activated Src is found at FAs and invadopodia. Functionally, this results in increased numbers of invadopodia, decreased migration, and increased invasion into matrigel. This points to FAK as being an important regulator of multiple steps during cancer progression and makes FAK an interesting target when investigating tumor behavior and therapeutic strategies.  FAK promotes cancer progression via kinase-dependent and –independent functions. The kinase-dependent functions of FAK promote cell motility538, invasion and survival567, as well as transcriptional responses that lead to EMT553, 568. Indeed, high levels of FAK phosphorylation at Y397 and Y576, which are markers of FAK kinase activity, correlate with increased migration and invasion in late-stage melanoma569. The kinase-independent functions of FAK support cell survival and proliferation, which is presumably regulated by FAK’s multiple protein-protein interactions294, 570, 571.  Based on studies in mouse models, in which FAK inhibition prevents tumour growth, metastasis, vascular permeability and angiogenesis, Phase I and II clinical trials with pharmacological FAK inhibitors have been started523. Studies with these inhibitors have revealed that the therapeutic effects occur via direct effects on the tumor cells including decreased cell motility, invasion, survival, and proliferation as well as effects on alteration on the tumor   58  microenvironment that limit tumor-promoting inflammation572 and prevent tumor metastasis by reducing vascular permeability573. Understanding the relative roles of the kinase-dependent and kinase-independent functions of FAK, and identifying the downstream pathways that each is connected to, will help us to develop chemotherapeutic drugs that target selective functions of FAK. 1.10.12 FAK functions in the nucleus Increasingly, attention has been given to the role of FAK in the nucleus and its effect on gene expression. Besides its role in adhesome signalling at the plasma membrane, FAK can also translocate into the nucleus, where it affects the transcription of genes that promote G1 and G2/M progression574, as well as developmental and inflammatory genes such as VCAM-1575. However, only a few nuclear interaction partners of FAK (p53, MDM2, MBD2, GATA4, CHIP) have been documented576. Recent studies on FAK-/- mice have shown that FAK has a kinase-independent scaffolding function in the nucleus, where FAK’s FERM domain acts to suppress the activation of p53294, 577, and therefore promotes cancer progression. It is only recently that the nuclear localization signal (NLS) and nuclear export signal (NES) sequences have been identified in FAK294, 578. Only a few physiological stimuli have been shown to cause FAK to shuttle from the cytoplasm to the nucleus576. Various stresses, such as oxidative stress579, chemical-induced apoptosis, and de-adhesion from the substratum294, and shear stress,580 cause FAK to translocate to the nucleus. It is likely that during adhesion FAK is anchored at adhesion sites (or to the cytoskeleton) to keep it out of the nucleus. Once these adhesion sites disassemble during cell detachment, or are altered by shear stress, the pool of ‘free’ FAK increases and FAK can translocate into the nucleus294. Interestingly, both reduction of FAK levels and pharmacological inhibition of the FAK kinase activity promote nuclear   59  localization of FAK575, 581, although the mechanism is not known. Nevertheless, nuclear FAK may play an important role in both normal and cancerous cells by regulating gene expression.  1.11 Rationale and thesis aim Changes in integrin expression and function, whether cell intrinsic or caused by ECM, may reprogram cancer cells to adapt a more aggressive phenotype that is resistant to existing therapies582. Because many tumors arise from epithelial cells, integrins expressed on epithelial cells are generally also present on tumor cells. In fact, studies that correlate integrin expression levels with metastasis or patient survival have established a direct correlation between the expression level of specific integrins and cancer progression583. For example, the αvβ3, α5β1 and αvβ6 integrins are usually expressed at low or undetectable levels in healthy epithelial cells but are highly expressed in some tumors583. In melanoma, progression from the radial growth phase to the more invasive, vertical growth phase is associated with increased expression of α5β1 and αvβ3 integrins584, 585.  During the metastatic cascade, cancer cells encounter ECMs of different composition and densities. Specifically, tumor cells come into contact with altered ECM after they have broken through the local basement membrane and invaded into the surrounding stroma. In addition, tumor microenvironments are characterized by increased ECM production by stromal cells. Many of these interactions between the tumor cells and their new environment are mediated by integrins, which are important for adhesome formation and outside-in integrin signalling. Thus, altered ECM-integrin interactions can drive tumor progression in multiple ways. Cancer progression is driven partly by cell-intrinsic gene expression, e.g. altered expression of oncogenes and tumor suppressors, as well as by microenvironmental effects on   60  gene expression. For example, in B-cell malignancies, adhesive interactions appear to regulate transcriptional differences in individual ARH-77 human malignant B-cells586. ARH-77 cells that were able to adhere to FN expressed higher levels of oncogenes (e.g. v-maf, v-fos and v-myc) and were also highly tumorigenic. A subpopulation of the same cells that was not able to adhere, were less tumorigenic and expressed genes associated with terminal B-cell differentiation. This supports the hypothesis that cell adhesion can drive gene expression changes that lead to cancer progression. In that context, it is of interest not only to identify cancer promoting genes that are regulated by integrin adhesion signalling but also to elucidate pathways that mediate these gene expression changes. This may facilitate the identification of targets and the development of new therapies to treat or prevent cancer progression.  In this thesis I first identified genes whose expression is regulated by ECM ligand density or tension. From these genes, I identified 3 genes that are associated with cancer progression in melanoma and other cancers, namely Cyr61, MUC18, and TRPM1. I then used these genes as ‘signature genes’ to investigate how cell adhesion signaling promotes cancer progression. Because the microenvironment of cancers is composed of highly diverse, cellular and non-cellular elements, I focused my analysis on sustained (i.e. 24-48 h) interactions of B16F1 mouse melanoma cells with FN, an ECM component that is associated with melanoma cells switching to an invasive growth phase587 and higher invasive capacity588. Because this adhesion-dependent interaction activates not only integrin signalling pathways, but also transduces mechanobiological tension into the cells, I assessed the contribution of mechanobiological forces (section 3.2.5) and two canonical adhesome proteins, talin and FAK (chapter 5), in the regulation of these cancer signature genes.   61  Although talin has been implicated in cancer progression, it is not known whether talin 1 and talin 2 have different roles in tumor progression. Therefore in chapter 6, I investigated the relative roles of talin 1 and talin 2 on cellular functions related to cancer progression such as cell spreading, motility, viability and tumor growth.   Hypothesis: ECM-integrin interaction alters the expression of cancer signature genes in a manner that depends on FAK, talin and mechanobiological tension.   Specific aims: Chapter 3  Identify genes that are regulated by ECM ligand density and/or tension.   Test the hypothesis that ECM ligand density, and/or tension regulate the expression of the chosen cancer signature genes (Cyr61, MUC18 and TRPM1). Chapter 4  Test the hypothesis that adhesion to increasing FN densities and the induction of mechanobiological tension regulates the expression levels of distinct sets of genes. Chapter 5  Test the hypothesis that the expression of these cancer signature genes is regulated by the adhesome proteins talin and FAK.  Chapter 6   Test the hypothesis that talin 1 and talin 2 have distinct effects on cell spreading, cell migration and cell growth in vitro as well as tumor growth in vivo.    62  Chapter 2: Material and methods 2.1 Materials Table 1: Materials  Item Company Catalogue number Polystyrene tissue culture dish, 10 cm BD Falcon 353003 Falcon 50 ml centrifuge tubes  VWR/Calbiochem/Axygen CA21008-940 Falcon 15 mL centrifuge tubes  VWR/Calbiochem/Axygen CA21008-935A 96 well tissue culture plate VWR/Calbiochem/Axygen CA62406-081 Cryovials 2 ml round bottom Fisher Scientific N368632 Falcon 6 well tissue culture plate VWR/Calbiochem/Axygen 087721B Circular coverslips, 12 mm diameter, for confocal microscopy Fisher Scientific 12-545-80 Cover glass, 18 mm circular Fisher Scientific 12-545-100 Parafilm VWR P1150-2 Glass microscope slides, 3’ x 1’ Fisher 12-550-123 10 µl filter tips DiaMed Lab Supplies DTECFT10GLR-480 20 µl filter tips DiaMed Lab Supplies DTECFT20GLR-480 200 µl filter tips DiaMed Lab Supplies DTECFT2000GLR-480 1000 µl filter tips DiaMed Lab Supplies DLSPT1000-500S Stainless steel beads Qiagen 69989   Multiplate™ Low-Profile 96-Well unskirted PCR plates Bio-Rad MLL-9601 Sealing tape (optically clear) for 96 well PCR plates Sarstedt 95.1994 Tissue Train® culture plates Flexcell International Corporation TT-4001U-cs Flexible silicone-bottom BioFlex® culture plate Flexcell International Corporation BF-3001U Nitrocellulose membrane Bio-Rad 162-0115 Classic blue autoradiography film BX Mandel Scientific EBA45   63  Item Company Catalogue number Glutathione-Sepharose 4B beads GE Healthcare 17-C756-01   Table 2: Antibodies and staining reagents for immunofluorescence Item Company Catalogue number Working dilutionAnti-vinculin (mouse)*, monoclonal, concentration not specified Sigma V4505 IF 1:100 WB 1:1000 Rhodamine-phalloidin Life Technologies R415 1:200 ProLong Gold anti-fade Molecular Probes P36935 Drop to cover surface DAPI Molecular Probes D1306 300 nM Anti-talin clone 8D4 (mouse), concentration not specified Sigma  T3287-.2mL WB 1:1000 Anti-talin clone C20 (goat) Santa Cruz sc7534 WB 1:500 Anti-talin 1 (rabbit), polyclonal, 0.4 mg/ml Abcam Ab71333 WB 1:500 Anti-talin 2 (mouse) clone 68E7, 1 mg/ml Abcam Ab105458 WB 1:500 Anti- MCAM/CD146 (sheep), polyclonal, 0.2 mg/ml Cedarlane AF6106 WB 1:500 Anti-FAK (mouse), polyclonal, concentration not specified  Molecular Probes AMO 0672 IF 1:500 WB 1:1000 Anti-FAK pY397 (rabbit), polyclonal, concentration not specified Molecular Probes 44624G WB 1:1000 (milk!) Anti-p130Cas (mouse), clone 21, 250 µg/ml BD Biosciences BD610271 WB 1:100 Anti-p130Cas pY165 (rabbit), polyclonal, concentration not specified Cell Signalling Technology 4015S WB 1:100 Anti-paxillin (mouse), clone 349, 1.0 mg/ml BD Biosciences 610051 WB 1:1000   64  Item Company Catalogue number Working dilutionAnti-paxillin pY118 (rabbit), polyclonal, concentration not specified Molecular Probes 44-722G WB 1:1000  Anti-paxillin pY31 (rabbit), polyclonal, concentration not specified Abcam ab4832 IF 1:500 WB 1:1000  Anti-Rap1A/Rap1B clone 26B4 (mouse), concentration not specified Cell Signalling Technologies 2399S WB 1:1000 Anti-Rap2A/Rap2B (mouse), clone EPR12825(B), concentration not specified Abcam ab173296 WB 1:1000 * Species of antibody, IF = immunofluorescence, WB = western blot Table 3: Commercially available solutions and chemicals  Item Company Catalogue number AlamarBlue® Cell Viability Reagent Life Technologies DAL1100 Ampicillin Sigma A9518 Bacto-agar BD Biosciences 214010 Bacto-tryptone BD Biosciences 211705 Bacto-yeast extract BD Biosciences 212750 DEPC-treated water Invitrogen/Gibco 750024 Dimethylsulfoxide (DMSO) MP Biomedicals, Solon 191418 Dulbecco’s Modified Eagle Medium (DMEM) Gibco, Invitrogen 11960 Fluorescent blue FluoSpheres Molecular Probes/Invitrogen F8815 Kanamycin sulfate Sigma Aldrich Chemical 60615 L-Glutamine Sigma C8540 Lipofectamine RNAiMax Invitrogen/Gibco 13778075 Lipofectamine™ 2000 transfection reagent Invitrogen/Gibco 11668019 Methanol Fisher  A412-4 NaCl Fisher BP358-212   65  Item Company Catalogue number Amersham enhanced chemiluminescence (ECL) western blotting detection reagent GE Healthcare RPN2106VV1/2 Paraformaldehyde (PFA) Cedar Lane Labs  15710 Penicillin streptomycin Invitrogen 15140122 Phosphate-buffered saline (PBS) Gibco 10010-023 Poly (2-hydroxyethylmethacrylate), PolyHEMA Sigma P3932-25G ProLong Gold anti-fade reagent supplemented with DAPI Molecular Probes/Invitrogen P36935 rhodamine-phalloidin Invitrogen/Gibco R415 RNASE AWAY MBP 7003 Sodium pyruvate Sigma P5280 Triton-X 100 Sigma Aldrich Chemical T8787 Trypsin Gibco, Invitrogen 25200072 IGEPAL Sigma CA-630 CellMask™ Orange plasma membrane stain Molecular Probes/Invitrogen C10045  Table 4: Reagent compositions Solution Composition RIPA lysis buffer 30 mM Tris-HCl pH 7.4 150 mM NaCl 1 % IGEPAL 0.5 % deoxycholate 0.1 % SDS  2 mM EDTA 10 µg/ml leupeptin 1 µg/ml pepstatin A 1 mM Na3VO4 1 mM PMSF 1 µg/ml aprotinin Rap lysis buffer (2X) 10 % glycerol 1 % IGEPAL 50 % mM Tris-HCl (pH 7.5) 200 mM NaCl 2 mM MgCl2 LB agar plates LB media 20 g/l bacto-agar   66  Solution Composition Running buffer 50 mM Tris 0.45% glycine 0.1 % SDS Transfer buffer 20 mM Tris-HCl (pH 8.3–9.2) 50 mM glycine 20% methanol SDS-PAGE sample buffer (5X) 312.5 mM Tris base (pH 2.8) 10% glycerol 11.5% SDS 500 mM DTT 0.1% bromophenol blue Tris-buffered saline (TBS) 2.5 g/l Tris-HCl, pH 8 8.8 g/l NaCl Cell freezing medium (2x) 50 % FBS, 20 % DMSO in DMEM 4.8% SDS-PAGE stacking gel 3.7 ml H2O 600 µl 1M Tris (pH 6.8) 600 µl 40 % bis acrylamide 50 µl 10 % SDS 50 µl 10 % ammonium persulfate  6 µl TEMED SDS-PAGE separating gel  8% 10%  12% H2O (ml)  4.275 3.875 3.475 1.5 M Tris (pH 8.8) (ml) 2 2 2 40 % bis-acrylamide (ml) 1.6 2 2.4 10% SDS (µl) 80 80 80 10 % ammonium persulfate (µl) 40 40 40 TEMED (µl) 5 5 5   Table 5: Kits Item Company Catalogue number RNeasy Mini RNA isolation Kit with Qiashredder columns Qiagen 74104 RNase-Free DNase set Qiagen 79254 SuperScript III reverse transcriptase Invitrogen/Gibco 18080-044   67  Item Company Catalogue number SsoFast EvaGreen Supermix Bio-Rad 172-5200 GenElute HP plasmid miniprep kit Sigma NA0160 PureLink™ HiPure plasmid filter maxiprep Invitrogen/Gibco K210017 GeneArt® CRISPR nuclease vector with OFP reporter Kit Invitrogen/Gibco A21174 Bicinchoninic acid (BCA) Kit Pierce Biotechnologies 23225  Table 6: Biological materials Item Company Ordering number Fetal bovine serum (FBS) Bio-Rad 161-0404 Fibronectin (bovine plasma) Sigma Aldrich  F4759-5MG BSA, Fraction V Fisher Scientific BP1600-100 Normal goat serum (NGS) Cedarlane Labs 005-000-121 Random primers  Invitrogen/Gibco 48190011 dNTP mix  Fisher Scientific FERR1121 Subcloning efficiency DH5α bacteria Invitrogen/Gibco 18265017 Rat tail collagen I (5 mg/ml)  BD Biosciences 354236   Table 7: Cell lines Cell line Source Reference B16F1 cells ATCC® CRL-6323™ B16F0 cells ATCC® CRL-6322™ B16F10 cells ATCC® CRL-6475™ Table 8: Equipment Item Company Ordering number Tissue culture incubators, Forma 3326 Fisher 15-465-117 Vortexer Fisher S8223-1 IEC Centra-CL3R Thermo Electron 32704   68  Item Company Ordering number centrifuge Corporation Olympus Fluoview 1000 confocal microscope  Contact company Zeiss 200M Axiovert microscope Zeiss Contact company Leica DM1400B modular system Leica Microsystems Contact company Life Cell adapter Chamlide TC-L-Z013 5415C Microcentrifuge Eppendorf discontinued Nanodrop 1000 spectrometer Thermo Scientific ND-2000 TissueLyser LT Qiagen 85600 Bio-Rad Mini Trans-Blot Bio-Rad 170-3930 Bio-Rad CFX96 Real-Time PCR System Bio-Rad Contact company Mini-gel apparatus with water-cooling system CBC Scientific DCX-700 Kodak X-OMAT 1000 processor MedTec Marketing Group Contact company SPECTRAmax® GEMINI-XS spectrophotometer Molecular Devices,  Contact company   Table 9: Primers for qPCR Primer name Sequence Target SC_202_f TGGGAAAGGTCTGATCAAGG TRPM1 (Exons 4-5) SC_203_r CTCTGGACTTGGAGGAGTGG TRPM1 SC_204_f AGCCAGATCTTCGTCTTTGG TRPM1 (Exons 10-11) SC_205_r TCCTTCTCTGTGGCTTTGGT TRPM1 SC_135_r TTAGCGCAGACCTTACAGCA Cyr61 (Exons 1-2) SC_136_f CCCTTCTCCACTTGACCAGA Cyr61 SC_138_r AGGGTCTGCCTTCTGACTGA Cyr61 (Exons 2-3) SC_139_f TGCTGTAAGGTCTGCGCTAA Cyr61 SC_152_f GGCTACCCCATTCCTCAAGT MUC18 (Exons 5-6) SC_153_r CTAGGCGTGCACTCAGAACA MUC18 SC_154_f TTCCTGGCTTGAATCGTA CC MUC18 (Exons 10-11) SC_155_r TCAGCACTGCATTCTCTTGC MUC18 SC_3f CGTCCCCTCATATCGGTGTA RPL4 (Exons 1-2) SC_4r CATGATCGGTTCCACTTGGT RPL4   69  Primer name Sequence Target Actin-f GGCTGTATTCCCCTCCATCG Beta-actin Actin-r CCAGTTGGTAACAATGCCATGT Beta-actin GAPDH_f CATTTGCAGTGGCAAAGTGGAG GAPDH GAPDH_r GTCTCGCTCCTGGAAGATGGTG GAPDH   Table 10: CRISPR primers with encoded gRNA site Primer name Sequence Target SC_285f GCTCACGAATCATGCGGCGTTTT Talin 1 (Exon 2) SC_286r GCCGCATGATTCGTGAGCCGGTG  SC_289f ATGACGCTTGTCGAGTCATTGTTTT Talin 2 (Exon 3) SC_290r AATGACTCGACAAGCGTCCGGTG    Table 11: siRNAs Target Company Catalogue number Talin 1 Invitrogen/Gibco s75202 Talin 2 Invitrogen/Gibco s88928 Negative control for siRNA Invitrogen/Gibco 4390843 FAK (PTK2) Invitrogen/Gibco s65839   Table 12: Software Name Source Institute/Location Image Pro Plus 6.2 analysis software  Media Cybernetics Rockville, Maryland ImageJ Reference 589 Open source Slidebook v6  3i Intelligent Imaging Innovations Denver, CO, USA     70  2.2 Cell culture procedures  2.2.1 Culture and storage of B16 cells  B16 cells were maintained in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% heat-inactivated Fetal bovine serum (FBS), 2 mM L-glutamine, 1 mM sodium pyruvate and 50 µg/ml penicillin-streptomycin, which I will be referring to as DMEM-complete. At 70% confluence, the cells were washed with 10 ml of phosphate buffered saline (PBS) and incubated for 1-5 min with 1 ml of trypsin, until the cells detached. The cells were then centrifuged at 1,500 rpm for 5 min and resuspended in DMEM-complete to the desired density. The cells were plated in 10 cm polystyrene tissue culture dishes and grown at 37°C in 5% CO2. For long-term storage in LN2, cells were resuspended to 2-5 x 106 cells per ml in 1 ml of 50% FBS, 10% dimethyl sulfoxide (DMSO) in DMEM-complete, and aliquoted into cryovials.   2.2.2 Transfection of B16 cells with plasmid DNA or siRNA  For each transfection, 2-3 x 105 cells were resuspended in 2 ml of DMEM-complete and seeded in one well of a 6-well plate. After 3-4 h the cells were washed with DMEM and transfected according to the instruction manual for Lipofectamine™ 2000 Transfection Reagent (0.5-2 µg plasmid DNA) or Lipofectamine RNAiMax Transfection Reagent (siRNAs, see Table 11). The siRNAs were used at final concentration of 40 nM. Single siRNAs were used for each target. After 24-48 h, the cells were tested for expression of the relevant proteins.     71  2.2.3 Cell sorting and clonal expansion  When cells were transfected with plasmids encoding EGFP or RFP, fluorescence-activated cell sorting (FACS) was used to enrich for EGFP- or RFP-positive (+) cells. The positive cell population was then plated at 1 cell/well in 50 µl DMEM-complete in 96-well plates. Clones derived from single cells were recovered and expanded for 2-3 weeks. One clone for each KO and control was successfully established.  2.2.4 Cell spreading assay and image quantification  Glass coverslips were sterilized with 100% methanol, air dried, and coated with 1-2 µg/cm2 FN for 1 h at 37°C. The coverslips were then washed with PBS before adding 2-4 x 104 cells in 0.5-1 ml DMEM-complete. After incubating the cells at 37°C, the medium was carefully removed and 4% paraformaldehyde (PFA) in PBS was added for 20 min at room temperature. To permeabilize cells, the coverslips were washed with PBS and incubated with 0.25% Triton-X 100 in PBS for 10 min at room temperature. The coverslips were then washed with PBS and blocked with 2% BSA, 10% normal goat serum (NGS) in PBS for 30 min. Primary antibodies (30-60 µl, see Table 2) were diluted in PBS containing 2% BSA. The coverslips were placed on parafilm with the cells and antibody solution facing down, and then incubated overnight at 4°C. The coverslips were then transferred into tissue culture dishes and washed 3X with PBS with gentle agitation for 10 min. Alexa fluorophore-conjugated secondary antibodies (see Table 2) were generally diluted 1:200 in PBS with 2% BSA. Rhodamine-phalloidin was added to the antibody solution and 30-60 µl was applied to the coverslip. After 1 h at room temperature in the dark, coverslips were transferred into tissue culture dishes and washed 3X with PBS with gentle agitation, protected from light for 10 min. The coverslips were mounted on glass microscope   72  slides using ProLong Gold anti-fade reagent supplemented with DAPI, allowed to dry overnight, then sealed with clear nail polish, and stored at -20°C until imaged. Imaging was performed using an Olympus Fluoview 1000 confocal microscope (UPlan 60X/1.35 Oil and UPlanApo 20X/0.70 objectives, based on the inverted Olympus IX81 microscope) or a Zeiss 200M Axiovert spinning disk microscope (63X/1.4 Oil DIC Plan Apo and 20X/0.5 EC Plan Neofluar objectives, QuantEM 512SC Photometrics camera). Analysis for cell spreading area and circularity was performed using ImagePro Plus 6.2 analysis software (Media Cybernetics), ImageJ, or Slidebook v6.5 software. Images taken at the cell-substrate interface of phalloidin-stained cells were used to determine the cell spreading area and circularity. Specifically, triangle thresholding was used as a cut-off criterion to generate binarized cellular shape masks for each cell in order to measure cell spreading area and circularity. Circularity was quantified using the formula circularity = 4π(area/perimeter2)  A circularity value of 1.0 indicates a perfect circle. As the value approaches 0.0, it indicates an increasingly elongated polygon.  Cell area in Figure 18 was determined by phase-contrast microscopy using a Leica DM1400B. Cell area was masked and measured with ImageJ.     2.2.5 2D bead-clearing motility assay  Bead-clearing motility assays were performed as described by Freeman et al.415. Coverslips (18 mm) were coated with FN as in section 2.2.4. For each coverslip, 10 µl of fluorescent blue FluoSpheres were pelleted at 2000 rpm (326 x g), washed in 1 ml of PBS, then resuspended in 50 µl PBS and spread on the FN-coated coverslip. FluoSpheres were allowed to   73  attach overnight at room temperature, before washing the coverslips with PBS. Cells (2x103) were then plated on each coverslip. After 17h at 37°C the cells were fixed, stained with rhodamine-phalloidin and imaged by using an Olympus Fluoview 1000 confocal microscope (UPlan 60X/1.35 Oil and UPlanApo 20X/0.70 objectives, based on the inverted Olympus IX81 microscope) Migrating cells clear the FluoSpheres off the area they have moved over. The cleared area per cell was quantified using ImageJ.  2.2.6 Motility assay with real-time imaging B16F1 cells were stained with CellMask™ Orange Plasma membrane stain (1:4000) in DMEM, for 30 min at 37°C in the dark. Cells were washed once with and then resuspended to 1x104 cells/ml in DMEM plus 1% FBS. Cells (100 µl) were mixed with 300 µl of DMEM plus 1% FBS and applied to the observation field of a coverglass chamber that had been coated with 2 µg/cm2 FN. The cells were allowed to attach for 1 h before performing live cell imaging using a Leica DM1400B modular system with the Live Cell adapter. Analysis of tracks was performed manually using the ImageJ chemotaxis and migration macro (http://rsb.info.nih.gov/ij/plugins/track/track.html).  2.2.7 Preparation of polyHEMA-coated plates for cell culture Six-well plates were coated twice with 3 ml of 10 mg/ml polyHEMA in 100% ethanol and dried at 37°C until all liquid was evaporated. The procedure was repeated. The wells were washed 6 times with PBS and the plates were stored at 4°C until use.    74  2.2.8 Cell growth and viability assay  Cells (1x103) in 100 µl DMEM-complete were seeded in triplicate in 96 well plates. Four hours before the end of the culture period, 10 µl Alamar Blue was added to each well. The plates were spun down and incubated at 37°C, before measuring fluorescence using the 560EX nm/590EM nm filter settings on SPECTRAmax® GEMINI-XS spectrophotometer. Each condition was done in triplicate (values obtained from 3 different wells averaged) to take in account the technical variability due to the efficiency of the Alamar Blue assay, sensitivity of the plate reader or sample preparation.  2.2.9 3D collagen I/FN for cell embedding   To create three-dimensional collagen/FN gels, 1.3 ml collagen I (5 mg/ml stock solution), 130 µl 10x DMEM and 41.6 µl 1N NaOH  (to adjust pH to 7.4) were mixed in this order on ice. Cells were trypsinized, resuspended to 1.75 x106 cells/ml, and 400 µl was added to 400 µl collagen I. On ice, 40 µg FN (in 40 µl) was added to the collagen I/cell mix. The cell/collagen/FN mixture (100 µl) was pipetted into 12 well Transwell dishes and incubated for 2 h at 37°C to ensure gel assembly, before adding 400 µl DMEM-complete. Collagen I monomers self-assemble into fibrillar structures given the right temperature and  pH. Since both temperature and pH can influence self-assembly, assembly conditions (time on ice, volume and time allowed for self-assembly at 37°C) were kept the same between samples and repeats.  To create collagen I/FN/cell matrices for the 3D stretching experiments, Tissue Train® culture plates were loaded into tissue train® linear trough loaders under 15% vacuum to create a linear trough. Collagen I/FN/cell mixtures (120 µl, prepared as above) were loaded into the troughs and left in the incubator for 2 h at 37°C to polymerize. The ECM/cell mixture only   75  attached to anchor meshes on the two opposing sides of the membrane (see Figure 21). This creates a free-floating ECM/cell gel strip that is held in place on either end. After gel polymerization, the vacuum was released and 3 ml DMEM-complete was added to each well. The ECM/cell gels were incubated overnight at 37°C prior stretching.  2.2.10 Staining of cells and RNA isolation from cells in 3D collagen/FN gels While still in the Transwells, the gels were fixed with 4% PFA in PBS overnight, permeabilized with 0.5% Triton X-100 in PBS for 10 min, and blocked with 10% NGS in PBS with 2% BSA for 1 h. Rhodamine-phalloidin diluted in PBS with 2% BSA was then added for 2 h. After washing with PBS containing DAPI (1:1000), the gels were mounted onto coverslips using ProLong Gold anti-fade reagent. Cells were visualized using an Olympus Fluoview 1000 confocal microscope. Alternatively, the gels were washed with PBS, lysed in 750 µl of RNA lysis buffer and subjected to RNA isolation as in section 2.3.1.1.   2.2.11 Cell stretching  Cell stretching in 2D was carried out according to Wang et al. 590. B16F1 cells (5x104) were plated in each well of 6 well flexible silicone rubber-bottom BioFlex® culture plate, that had been coated with 5 µg/cm2 of FN. The cells were allowed to adhere for 3 h at 37°C. The 35 mm diameter rubber-bottom plates were then placed on a Flexcell baseplate with 25 mm diameter cylindrical posts that fit underneath each rubber-bottom well. When a downward vacuum is applied the silicone rubber is pulled downwards the side of the posts as the posts are smaller than the silicone rubber-plates on top or them. This stretches the substratum on top of the posts in all directions. Vacuum pressure was applied to generate a 10% elongation of the silicone   76  rubber membrane. The stretching of the silicone rubber membrane was either applied constantly for 24 h or as cycles of 2 min stretch and 2 min release for 24 h. Immediately after stretching, the cells were washed with PBS and lysed in 750 µl of RNA lysis buffer before isolating total RNA as in section 2.3.1.1.  Cell stretching in 3D was carried out according to Garvin et al.591. To stretch cells that were embedded in collagen I/FN matrices, Tissue Train® culture plates containing the strip of cells embedded in ECM gels (for assembly see 2.2.9) were placed into Arctangle® Loading Posts. A 46.55 kPa pressure was applied via the vacuum pulling down the flexible-membrane where the mesh was connected to the gel. This created a uniaxial 10% elongation on the two opposing ends of the gel strip. Immediately after stretching, the ECM gels were washed with PBS, lyzed in 750 µl of RNA lysis buffer and subjected to RNA isolation as in (2.3.1.1).    2.3 Molecular biology techniques  2.3.1 RNA techniques RNA work was performed under RNAase-free conditions using RNASE AWAY to clean surfaces and equipment. Filter tips were used for all pipetting.  2.3.1.1 Total mRNA isolation Cells were washed 1X with PBS. After completely removing all liquid, total mRNA was isolated using the RNeasy Mini RNA Isolation Kit with Qiashredder Columns according to the manufacturer’s instructions. To ensure complete lysis of the cells, cell lysates were shredded using the TissueLyser LT (Qiagen) with stainless steel beads (Qiagen). On-column DNase   77  digestion was carried out using the RNase-Free DNase set. Isolated total mRNA was resuspended in DEPC-treated H2O. The concentration and purity was assessed using a Nanodrop 1000 spectrometer (Thermo Scientific) and RNA samples were stored at -80°C.  2.3.1.2 Transcriptome analysis by RNA-seq  Total RNA from B16F1 cells was isolated as described in section 2.3.1.1. Transcriptome analysis was performed from 2 independent biological experiments. RNA-seq libraries were generated from 1 µg total RNA. Samples were run in two lanes at PE75s (paired-end 75 bp) on a HISeq2500 in high output mode. For each library ~100 million reads were generated and ~75 million aligned within gene boundaries. mm10 software was used to generate Reads Per Kilobase of transcript per Million (RPKM) values for comparative analysis.   To identify mRNAs that were differently expressed to a significant extent, the data were pre-processed by adding 0.1 to each RPKM value, in order to limit false discovery rates due to noise in low-read samples. The data were then subjected to log2 transformation. Multiple t-tests, with a correction for p-value based on the Sidak-Holme method592, 593 were used to identify genes with significantly different expression between low FN and medium FN, between low FN and high FN, and between no stretch and stretch conditions. Calculations were carried out using GraphPad Prism 6.   Pathway and GO cluster analyses were performed using GeneAnalytics© software from the GeneCardsSuite©. Putative TF binding sites (TFBS) were identified using the ToppCluster software594. The resulting network was visualized using Cytoscape595 version 3.2.1.    78  2.3.1.3 Reverse transcription  Total RNA (0.4-2 mg) was used for converting total mRNA into cDNA. cDNA libraries were generated using SuperScript III Reverse Transcriptase using random primers and a dNTP mix according to the manufacture’s instruction. Samples were stored at -20°C.   2.3.1.4 Real-time quantitative PCR (qPCR)  To assess the expression of selected genes, qPCR was performed using SsoFast EvaGreen Supermix with the Bio-Rad CFX96 Real-Time PCR System. Each well of a Multiplate™ Low-Profile 96-well Unskirted PCR plate contained 5 µl of SsoFast EvaGreen Supermix, 400 nM of each forward and reverse primer, and 20 ng cDNA template in a final volume of 10 µl. Each sample was run in triplicate as technical repeats. Samples were run with the program shown in Table 13. Gene expression was normalized to beta-actin, GAPDH and RPL4. Primer details are listed in Table 9. Data were analyzed using the Bio-Rad CFX manager.  Table 13: PCR Program for qRT-PCR Cycling step Temperature ( °C) Time (sec) Numbers of cycles Enzyme activation 95 30 1 Denaturation 95 5 45  Annealing/extension 60 5 Melting curve 65 to 95 in 0.5 steps 5 1     79  2.3.2 Bacterial transformation  Plasmid DNA was transformed into E. coli DH5α competent bacterial cells according to the manufacturer’s instructions and plated on LB agar plates containing either 100 µg/ml ampicillin or kanamycin, as appropriate. Single colonies were isolated and grown up in LB media with antibiotics for plasmid isolation.  2.3.3 Plasmid preparation  Plasmid preparation was performed using either the GenElute HP Plasmid miniprep kit or the PureLink™ HiPure Plasmid Filter maxiprep kit according to the manufacturer’s instruction. Plasmids were eluted in DEPC-treated water and the concentration and purity were assessed using a Nanodrop 1000 spectrometer. Plasmids were stored at -20°C or 4°C.  2.3.4 Design and cloning of CRISPR/CAS9 plasmids    To create knockout cell lines the CRISPR/Cas9 technique was used596. Using the online CRISPR design software (http://crispr.mit.edu/), targeting sequences for talin 1 and talin 2 were designed. Forward and reverse oligonucleotides (see Table 10) were annealed and cloned into the GeneArt® CRISPR nuclease vector with orange fluorescence protein (OFP) according to manufacturer’s instructions. B16F1 cells were transfected with 2 µg of CRISPR nuclease vector targeting either talin 1 or talin 2. The cells were FACS-sorted for OFP expression 36 h post-transfection. Cells were plated in 96-well plates using limiting dilution technique and clones arising from single cells were grown up. Details of CRISPR/Cas9 mechanism and the KO design are explained in sections 1.10.8 and 6.2.1.    80  2.4 Biochemical methods  2.4.1 Preparation of cell lysates  Cells were washed once with ice-cold PBS and lysed on the plate using RIPA lysis buffer containing protease and phosphatase inhibitors (Table 4 and Table 3, respectively). For protein phosphorylation assays, phosphatase inhibitor concentrations were doubled. Lysates were incubated on ice for 20 min, and scraped into pre-chilled 1.7 ml Eppendorf tubes and then kept on ice for 20 min with periodic vortexing. To remove insoluble material, lysates were centrifuged at 15,000g for 15 min at 4°C. The supernatant was removed and protein concentrations were determined using the BCA protein assay. One fifth volume of 5X SDS-PAGE-sample buffer (Table 4) was added to lysates. The samples were boiled for 5 min and stored at -80°C prior to SDS-PAGE and western blotting (section 2.4.2).   2.4.2 SDS-PAGE and western blotting  Cell lysates (20-30 µg) were loaded onto 8%, 10% or 12% SDS-PAGE gels and separated at 200 V for approx. 1.5 h. Separated proteins were then transferred onto nitrocellulose membranes at 100 V for 1 h at 20 V overnight using a Bio-Rad Mini Trans-Blot apparatus. Membranes were then rinsed with Tris-buffered saline (TBS) and blocked in TBS 0.05% Tween-20 (TBST) containing 5% skim milk powder or 0.05 mg/ml BSA for 30 min. Primary antibodies were diluted in TBST containing 5% skim milk powder or 0.05 mg/ml BSA (Table 2) and incubated at 4°C overnight with gentle agitation. This was followed by three 15-min washes in TBST at room temperature with shaking. Secondary antibodies were diluted in TBST with 5% skim milk powder and incubated with the filter for 1 h at room temperature with shaking. ECL   81  detection was performed. Molecular weights and band intensities were quantified using ImageJ. Phosphorylation of FAK on Y397 was calculated as follows: the densitometry signal for phospho-FAK Y397 was divided to that of total FAK. The percent inhibition of FAK was calculated using the following formula:   %	݄ܾ݅݊݅݅ݐ݅݋݊ ൌ 100 െ ሺ௡௢௥௠௔௟௜௭௘ௗ	௣௛௢௦௣௛௢௥௬௟௔௧௜௢௡	௢௙	ி஺௄	௢௙	௉ிିଶଶ଼	௧௥௘௔௧௘ௗ	௖௘௟௟௦∗ଵ଴଴௡௢௥௠௔௟௜௭௘ௗ	௣௛௢௦௣௛௢௥௬௟௔௧௜௢௡	௢௙	ி஺௄	௢௙	௖௢௡௧௥௢௟	௖௘௟௟௦ሻ ሻ  2.4.3 Rap activation assay  Rap activation assays were performed as described previously597. All steps were performed on ice. Briefly, cells were washed with PBS and lysed in Rap lysis buffer (Table 4) with protease inhibitors. Lysates were kept on ice for 30 min, with vortexing 3X during that time. A GST-fusion protein containing the Rap1 binding domain of Ral-GDS, which only binds to active Rap, was used to pull down activated Rap. Cell lysates were mixed with 30 µl of glutathione-sepharose 4B beads which had been coated with 30 µl of bacterial lysate containing GST-Ral-GDS for 1h at 4°C. After overnight incubation at 4°C, the beads were washed 2X with PBS and boiled in reducing SDS-PAGE sample buffer. The proteins eluted from the beads, as well as total cell lysates were separated by SDS-PAGE and analyzed by western blotting with anti-Rap antibodies (section 2.4.2).   2.5 Subcutaneous growth of B16F1 tumors in C57BL/6 mice B16F1 cells were resuspended to 106 cells/ml in DMEM, and 100 µl was injected into the left and right flanks of C57BL/6 mice. Each mouse received the same number of talin KO cells in one flank and control cells in the other flank. Mice were sacrificed when tumors were palpable   82  or at experimental endpoints between 13 days and 27 days post-injection. Tumors were weighed as an indicator of size. Differences between knockout and control cells were evaluated using a paired t-test where the pair was defined as tumors coming from the same mouse. Procedures were carried out and mice were monitored in accordance with the University of British Columbia animal license A11-0199.   83  Chapter 3: Cancer signature genes are regulated by integrin ligand density but not by cellular tension  3.1 Introduction Although integrin ligand density, substrate stiffness, and tension can drive cancer progression250, the underlying molecular pathways are not completely understood. Transcriptional changes associated with the transition to a progressive cancer phenotype in melanoma and other cancers are not well defined and only recently some key genes and their transcriptional regulation have been elucidated598, 599. For example, the two TFs, AP-2α and CREB/ATF1, have been shown to play an important role during the progression of melanoma, as they, along with their downstream targets, are dysregulated during the transition from the horizontal to the vertical growth phase. Some of these downstream targets, including MMP-2, c-KIT, Cyr61, and MUC18, are known to contribute to cancer progression598, 600.  FN is a major constituent of ECMs and plays an important role in embryogenesis, wound healing, and cancer invasion254, 601, 602. FN is a multi-domain protein and contains a RGD (Arg-Gly-Asp) motif that mediates the binding to integrin such as α5β1 and αvβ3350. Overexpression of FN is associated with various human tumors, including metastatic melanoma587, 588, 603-605. Increased FN deposition correlates with the switch to the invasive growth phase587 and greater invasive capacity588. Moreover, using soluble FN-derived peptides to interfere with the binding of integrins to FN inhibits experimental metastasis of melanoma cells in mice606.  B16F1 cells are highly proliferative mouse melanoma cells. They represent a well-established tumor model607 and express FN-binding integrins such as αvβ3 and α5β1. When injected s.c into C57BL/6 mice, B16F1 cells form tumors within two weeks in situ. When   84  injected i.v., the cells readily form lung metastases within two weeks. These properties make B16F1 cells a commonly used model for tumor growth and tumor cell extravasation. B16F1 cells were derived more than 60 years ago by isolating metastases from lung tissue after the less metastatic C57BL/6-derived melanoma cells (B16F0 cells) had been i.v. injected608. By successively re-injecting and collecting these metastases for 10 rounds, Fidler et al. also created the B16F10 cell line. The F0, F1 and F10 variants of B16 cells are characterized by progressively higher metastatic potential with B16F10 cells being capable of forming almost four times as many metastases after tail vein injections than B16F1 cells609-611. During cancer progression, normal tissue architecture is compromised and cancer cells are exposed to ECMs that are altered in terms of their composition and density. I therefore designed an experimental system to identify gene expression changes that occur when cancer cells adhere to increasing amount of ECM. I used high-throughput RNA sequencing (RNA-seq) to identify genes that are regulated by increased FN density. From these genes I selected Cyr61, MUC18, and TRPM1 as ‘cancer signature genes’ to investigate how adhesome signaling affects gene expression. These three genes were selected based on their strong regulation by FN density and their association with cancer progression in other studies. I then tested the hypothesis that adhesion to FN can drive the characteristics of ECM-dependent tumor progression in B16F1 cells, using the cancer signature genes Cyr61, MUC18, and TRPM1 as a readout. Because cell spreading on ECM can also modulate cytoskeletal tension, I also asked whether tension is important for the ECM-induced regulation of these genes.    85  3.2 Results 3.2.1 Identification of genes regulated by cell adhesion to increasing FN density  Based on the prominent role of FN in melanoma progression, I used B16F1 as a model for melanoma and varied the density of FN that they were plated on. I tested different FN densities ranging from 25 ng/cm2 to 5 µg/cm2 of FN coated on silicone rubber. Uncoated silicone rubber does not allow for cell attachment612. In this experimental setup B16F1 also didn’t attach, even  when cells are cultured in complete medium, which contains 2.5-3 µg/ml of soluble FN613. Therefore, the B16F1 cells were limited to attach to FN that is coupled to a ‘softer’ surface than TC plastic (Young’s elastic modulus ‘E’ 2-4 GPa for TC plastic614 and ≈150 kPa for silicone substrates used in standard cell-stretch systems615, 616, 617), which better reflects the rigidity of physiological substrates. I identified that B16F1 cells exhibited distinct morphologies depending on the FN density to which they are attached. In particular densities of 0.025 (low), 0.25 (medium) or 2.5 (high) µg/cm2 of FN resulted in reproducible morphological phenotypes (Figure 10). At low FN density, most of the B16F1 cells were able to adhere, but did not spread. On medium FN density, all of the cells attached and the majority were able to spread. At high FN density, all of the cells spread extensively and flattened. These density-dependent changes in cell morphology also indicated that soluble FN or other adhesion substrates present in the medium did not coat the silicone rubber substrate.    86   Figure 10: Cell adhesion to ECM regulates cell morphology. Cells were cultured on 0.025 (low), 0.25 (medium) or 2.25 (high) µg/cm2 FN for 24 h before imaging. Representative images from 9 independent experiments are shown. Size bar, 100 µm.   To identify FN density regulated genes, RNA from B16F1 cells that had been plated on low, medium and high FN was subjected to RNA-seq. RNA-seq is a high throughput technology that takes a quantitative snapshot of the RNAs present at a given time. In contrast to microarray methods, which are limited to the detection of known RNAs, RNA-seq directly determines the cDNA sequence, thereby allowing for the detection of novel genes, splice variants and changes in non-coding RNAs (e.g. microRNAs, long non-coding RNAs)618, 619. These RNA snapshots, also called transcriptomes, provide information about the RNA molecules present in a population of cells but do not distinguish changes due to de novo transcription, transcriptional attenuation, or post-transcriptional events such as RNA degradation. Nevertheless, the analysis of transcriptomes, also referred to as expression profiling, provides insight into genes that are being actively expressed. This RNA-seq was conducted as pilot experiment without repeats. The RNA-seq data revealed that increasing FN density caused changes in gene expression. Table 14 to Table 17 summarize the top 20% up and downregulated genes that were revealed by this initial RNA-seq experiment.    87  Table 14: Top 20% upregulated genes from low to medium FN Gene symbol Gene name Fold change Adj. p-value Pigk phosphatidylinositol glycan anchor biosynthesis, class K 4.33 <0.0001 Ahnak AHNAK nucleoprotein (desmoyokin)  3.71 <0.0001 MUC18 melanoma cell adhesion molecule  3.61 <0.0001 Fech ferrochelatase  3.16 0.0011 Phip pleckstrin homology domain interacting protein  2.95 0.0008 Dusp3 dual specificity phosphatase 3 (vaccinia virus phosphatase VH1-related)  2.60 0.0007 Pik3ca phosphatidylinositol 3-kinase, catalytic, alpha polypeptide  2.54 0.0119 Synpo synaptopodin  2.53 0.0229 Tmem184b transmembrane protein 184b 2.53 0.0229 Tsc22d1 TSC22 domain family, member 1  2.47 0.0195 Abcc1 ATP-binding cassette, sub-family C (CFTR/MRP), member 1  2.39 0.0332 Actr10 ARP10 actin-related protein 10 homolog (S. cerevisiae) 2.39 0.0332 Ap1s1 adaptor protein complex  AP-1, sigma 1  2.36 0.0119 Btaf1 BTAF1 RNA polymerase II, B-TFIID transcription factor-associated, (Mot1 homolog, S. cerevisiae)  2.34 0.0448      88  Table 15: Top 20% upregulated genes from low to high FN Gene symbol Gene name Fold change Adj. p-value MUC18 melanoma cell adhesion molecule 5.06 <0.0001 Ahnak AHNAK nucleoprotein (desmoyokin) 3.61 <0.0001 Phip pleckstrin homology domain interacting protein 3.25 0.0001 Pigk phosphatidylinositol glycan anchor biosynthesis, class K 3.20 0.0004 Synpo synaptopodin 3.09 0.0022 Fech ferrochelatase 2.82 0.0057 Srxn1 sulfiredoxin 1 homolog (S. cerevisiae) 2.77 <0.0001 Abcc1 ATP-binding cassette, sub-family C (CFTR/MRP), member1 2.76 0.0054 Tsc22d1 TSC22 domain family, member 1 2.67 0.0055 Maoa monoamine oxidase A 2.66 0.0000 Pik3ca phosphatidylinositol 3-kinase, catalytic, alpha polypeptide 2.61 0.0101 Gm2904 predicted pseudogene 2904 2.60 0.0033 Esd esterase D/formylglutathione hydrolase 2.60 0.0033 Gm5456 predicted gene 5456 2.60 0.0033 Slc7a11 solute carrier family 7 (cationic amino acid transporter, y+ system), member 11 2.56 <0.0001 Mdm2 transformed mouse 3T3 cell double minute 2 2.54 0.0013 Pgd phosphogluconate dehydrogenase 2.48 <0.0001 Dusp3 dual specificity phosphatase 3 (vaccinia virus phosphatase VH1-related) 2.46 0.0013 Itga6 integrin alpha 6 2.43 0.0017 Cyr61 cysteine rich protein 61 2.42 0.0049 Prdx6 peroxiredoxin 6 2.40 <0.0001 Txnrd1 thioredoxin reductase 1 2.37 <0.0001 1110057K04Rik 2.35 0.0465 Limch1 LIM and calponin homology domains 1 2.33 0.0004     89  Table 16: Top 20% downregulated genes from low to medium FN Gene symbol Gene name Fold change Adj. p-value Vwa3a von Willebrand factor A domain containing 3A  -4.29 0.0029 Mgll monoglyceride lipase  -4.18 0.0029 Phlpp2 PH domain and leucine rich repeat protein phosphatase 2 -3.75 0.0029 Lrrc46 leucine rich repeat containing 46  -3.48 0.0100 Gm10639 predicted gene 10639  -3.19 0.0007 Dcaf10 DDB1 and CUL4 associated factor 10  -3.08 0.0492 Lgals3bp lectin, galactoside-binding, soluble, 3 binding protein  -2.46 0.0123 Myl4 myosin, light polypeptide 4 -2.32 0.0119 Gramd1a GRAM domain containing 1A  -2.32 0.0452 Trpm1 transient receptor potential cation channel, subfamily M, member 1  -2.29 <0.0001            90  Table 17: Top 20% downregulated genes from low to high FN Gene symbol Gene name Fold change Adj. p-value Mgll monoglyceride lipase -6.33 <0.0001 Gm10639 predicted gene 10639 -4.51 0.0003 Trpm1 transient receptor potential cation channel, subfamily M, member 1 -3.64 <0.0001 Oca2 oculocutaneous albinism II -3.02 0.0005 Gsta1 glutathione S-transferase, alpha 1 (Ya) -2.93 <0.0001 A130022J15Rik -2.81 <0.0001 Tyr tyrosinase -2.80 <0.0001 Fbln5 fibulin 5 ] -2.67 <0.0001 Mknk2 MAP kinase-interacting serine/threonine kinase 2 -2.58 <0.0001 Ptgds prostaglandin D2 synthase (brain) -2.48 <0.0001 Vps11 vacuolar protein sorting 11 (yeast) -2.47 0.0225 Tspan10 tetraspanin 10 -2.43 <0.0001                 91  3.2.2 Identification of cancer signature genes Two of the highest genes regulated by FN density identified by RNA-seq, MUC18 and TRPM1, are known markers of melanoma progression. MUC18 is a transmembrane glycoprotein that is expressed on melanomas but rarely on premalignant lesions 620-624. It has recently been proposed as a ‘molecular warning of cancer progression’ in early stage melanoma, where the presence of MUC18 on circulating tumor cells predicts a clinically apparent disease, and the absence is related to stable disease or disease-free status625. TRPM1, a non-selective cation channel originally named melastatin, is steadily lost during melanoma progression and is partially or completely lost in metastatic melanoma626, 627.  Both, MUC18 and TRPM1 were amongst the top 20% up-regulated (MUC18) and down-regulated (TRPM1) genes from low to medium and low to high FN density. Also regulated by increasing FN density (top 20% upregulated from low to high FN) was Cysteine-rich protein 61 (Cyr61). Positive correlations between Cyr61 expression and the stage, tumor size and poor survival have been found in several cancers including breast cancer628-632, prostate cancer633, glioma634 and squamous cell carcinoma635. In melanoma cells, Cyr61 is often aberrantly expressed. However it is not clear at which stage during cancer progression Cyr61 acts as a tumor suppressor636 or tumor promoter637, 638.  Given the connection of these three genes, now referred to as “cancer signature genes” to cancer progression, in particular for melanoma, I first validated the expression levels identified by RNA-seq using qRT-PCR (Figure 11). When B16F1 cells adhered to the same FN densities (low, medium. high), Cyr61 mRNA levels increased ~2.5-fold from low to medium FN density and ~4.3-fold from low to high FN density. Similarly, MUC18 mRNA levels increased ~3.8-fold when the FN density was changed from low to medium, and ~5.8-fold when the FN density was   92  changed from medium to high. TRPM1 mRNA levels were downregulated in a dose-dependent manner (~2.7-fold for low versus medium FN and ~4.5-fold for low versus high FN). Taken together, the qRT-PCR data confirmed that Cyr61, MUC18 and TRPM1 mRNA expression is dose-dependent regulated by FN density.        Figure 11: Cell adhesion to ECM regulates gene expression of Cyr61, MUC18 and TRPM1. Cells were cultured as in A and mRNA levels were determined by qRT-PCR. Fold change of signature genes between low versus medium, and low versus high FN. Mean and SD of 4 (Cyr61) or 5 (MUC18 and TRPM1) independent experiments. Asterisks indicate significant differences (Student’s paired t-test between medium/low FN (black bars) or high/low (grey bars), **, p < 0.01; ***, p < 0.001; ****, p < 0.0001).   3.2.3 Adhesion stimulates FA signalling in B16F1 cells   In order to be able to test whether cell adhesion regulates the expression of the signature genes, I compared B16F1 cells cultured under non-attached conditions to cells attached to tissue culture (TC) plastic. When cultured on TC plastic, B16F1 adopted a different morphology than   93  when they were cultured on TC plastic that was coated with poly-2-hydroxyethyl methacrylate (polyHEMA), a polymer that prevents cell attachment639. Cells cultured directly on plastic attached and spread, whereas cells added to polyHEMA coated wells were unable to adhere and grew in multicellular clumps that adhered to each other (Figure 12A). To confirm that integrin-dependent signalling was only minimally activated in cells that were kept in suspension for 24h and that culturing cells on TC plastic stimulated the phosphorylation of integrin adhesion proteins, I assessed total protein tyrosine-phosphorylation as well as the phosphorylation of FAK, paxillin and p130Cas on specific tyrosines that are associated with activation and/or recruitment to integrin-adhesions. In comparison to cells that could not attach, cells that adhered to tissue culture plates had higher total phospho-Tyrosine (pTyr) (Figure 12B) as well as higher levels of phosphorylation of FAK-Tyr397, paxillin-Tyr118, and p130Cas-Tyr410 (Figure 12C). This indicates adhesion to TC plastic stimulates adhesome signalling, which does not occur when B16F1 cells form clumps in suspension. This validated the use of these conditions to determine whether adhesome signalling regulates the three signature genes.   94   Figure 12: Adhesion to TC plastic results in adhesome signalling. (A) B16F1 cells were cultured on polyHEMA or TC plastic overnight. Size bars, 100 µm. (B,C) Total cell lysates were prepared from cells cultured as in A. Total cell lysates were probed for pTyr (B) and the indicated (phospho)proteins (C). Molecular weight markers in kDa are indicated. Each panel is representative of  > 3 independent experiments.   3.2.4 Adhesion is a strong regulator of the cancer signature gene expression  Because adhesion to TC plastic induced strong phosphorylation of key adhesome proteins, I tested whether this also regulates the expression of the cancer signature genes. Compared to expression levels of cells cultured on polyHEMA, B16F1 cells plated on TC plastic   95  had a ~4-fold upregulation of Cyr61 mRNA and MUC18 mRNA, as well as a ~5.5-fold downregulation of TRPM1 mRNA (Figure 13A). Moreover, the upregulation of MUC18 protein expression upon adhesion to TC plastic was also detectable by immunoblotting (Figure 13B and C). This suggests that cell adhesion regulates the expression of the three signature genes.   Figure 13: Adhesion to TC plastic regulates cancer signature gene expression. (A) Fold change of signature genes in B16F1 cells plated on TC plastic or polyHEMA for 24 h. mRNA levels were determined by qRT-PCR. Mean and standard deviation (SD) for 3 independent experiments. (B) Immunoblots of MUC18 from total cell lysates from B16F1 cells on polyHEMA or TC plastic for 24 h. (C) Quantification of immunoblots from 3 independent experiments.    I showed that cell adhesion is a strong regulator of the cancer signature genes Cyr61, MUC18 and TRPM1, and that the expression of these genes is regulated by the ECM component FN in a dose-dependent manner. One molecular mechanism by which the signature genes may be regulated is through mechanobiological forces generated during spreading of the B16F1 cells   96  on ECM. For cells to spread on ECM surfaces, they actively undergo cycles of tension generation and release to establish a tensional balance between cytoskeletal forces, membrane tension and surface area640. Cytoskeletal forces depend on substrate stiffness, ECM ligand density and spreading area. For example, in human pulmonary artery endothelial cells, cell spreading leads to an increase in FA numbers, with lower average forces per adhesion but higher total forces than cells that have spread to a lesser degree. Conversely, increasing integrin ligand density while confining spreading area increases the number of FAs and produces larger total forces641. Therefore, I hypothesized that mechanical forces, with their ability to activate integrin signalling pathways344, 533, 642, may be partly responsible for the FN density-induced changes in gene expression of the three signature genes.   To test this hypothesis, I used three approaches to modulate mechanical forces in B16F1 cells. First, I increased external mechanical load by stretching cells that were adhered to FN-coated flexible silicone rubber plates. Second, I decreased internal cytoskeletal tension using blebbistatin, a drug that inhibits actomyosin contractility643, 644. Third, I inhibited the activity of the small GTPase Rap1. Rap1 has been shown to control the cell’s ability to respond to mechanically induced tension416.   3.2.5 Mechanobiological force is a weak regulator of the cancer signature genes   3.2.5.1 Stretch-induced mechanobiological forces do not significantly regulate cancer signature gene expression in 2D tissue culture systems  Over the past two decades different in vitro systems have been used to study the response of cells to altered mechanical loads645. Many of these studies have used fibroblasts from tendons,   97  ligaments, hearts and lungs, as well as osteoblastic, endothelial cells and smooth muscle cells. These cell types are naturally exposed to varying mechanical loads in vivo. Experimental alteration of the mechanical loads in these cell types has revealed that stretch- or shear-induced tension is sensed by FA and results in changes in FA composition646 and reorganization of the cytoskeleton that modulates cell shape, orientation, proliferation, and gene expression647-649. For example, mechanical stretching of integrin-bound talin that is induced by external forces promotes the recruitment of vinculin to FAs and stabilizes talin at these adhesion sites495. Increased levels of vinculin staining at FAs can therefore be used as an indicator for cellular responses to applied stretch. Cell stretching also regulates the expression of ECM components (e.g. collagen I and FN) and ECM-modifying enzymes (e.g. MMPs) at both the transcriptional and translational level650, 651. Whether this tension-induced regulation is inhibitory or stimulatory for a given gene is dependent on cell type, the duration and strength of the applied mechanical load, and the ECM substrate to which the cells adhere204, 645. Interestingly, one of the cancer signature genes, Cyr61, has previously been shown to be induced by stretch in vascular smooth muscles in vitro and by pressure overload in vivo 652. It is not known if Cyr61 is regulated in B16F1 cells or in other cancer cell lines in response to stretch. To test the impact of external mechanical load on the regulation of the three signature genes, B16F1 cells were mechanically stretched using a Flexcell® tension system. Stretched and unstretched cells were then analyzed for changes in expression of the cancer signature genes. Previous data from our lab416  showed that when B16F1 cells were plated on FN-coated silicone rubber plates, allowed to adhere overnight, and then subjected to 10% equibiaxial (uniform radial and circumferential strain) stretch for 4 h, increased vinculin staining intensity at β1-integrin adhesion sites was observed. This suggests that   98  increasing external mechanical load by stretching is able to induce tension dependent changes in the integrin adhesome in B16F1 cells, consistent with previous findings in other cell types646.  To determine if stretch-induced changes in the integrin adhesome correlated with regulation of the cancer signature genes, I compared Cyr61, MUC18 and TRPM1 mRNA levels from unstretched and stretched B16F1 cells. Cells were allowed to adhere to FN-coated silicone rubber plates overnight, subjected to either 10% equibiaxial static stretch or cycles of 2 min on and off stretch for 24 h. Static stretch did not alter MUC18 and TRPM1 mRNA level and caused only a slight decrease (~1.5-fold) in Cyr61 mRNA (Figure 14, left). Cyclic stretching also did not significantly alter the RNA levels of the cancer signature genes (Figure 14, right). Taken together, these data indicate that increasing mechanical load by stretching may regulate Cyr61 expression, but not the expression of MUC18 or TRPM1 in B16F1 cells that are adhered to FN.      99   Figure 14: Stretch does not regulate cancer signature gene expression. B16F1 cells were cultured on FN-coated silicone rubber plates overnight, and then subjected to equibiaxial 10% strain for 24 h. With static stretch (left panel) or cyclic stretch (right panel), mRNA levels were determined by qRT-PCR. Data are represented as the fold change in mRNA level for stretched cells compared to unstretched cells. A value of 1 or -1 = no change. For both panels, the mean and range for 2 independent experiments are shown.   3.2.5.2 Stretch-induced mechanobiological forces do not significantly regulate cancer signature gene expression in 3D tissue culture systems Recently, three-dimensional (3D) matrices have been used to create more physiological models for analyzing cell-matrix interactions than 2D tissue culture653-666. Although adhesomes in 2D and 3D may differ in composition, their underlying function as mediators of outside-in signalling appears to be important in both 2D and 3D settings. For example, increased ECM stiffness in 3D promotes the activation of signalling pathways that are important for tumor cell growth and invasion. Furthermore, vinculin acts as a force-dependent sensor connecting the actomyosin network with the ECM in 3D667, a mechanism that is well established in 2D.   100  Recently, it has been shown in B16F1 cells that applying stretch in 3D matrices induces the formation of FA-like complexes at the edges of the cell, with vinculin accumulating at these sites416. This suggests that in B16F1 cells, adhesomes regulate ECM-dependent pathways in 3D. Therefore I tested whether applying external mechanical forces on B16F1 cells in 3D matrices can regulate the cancer signature genes. To create a 3D stretchable matrix, I embedded B16F1 cells in a 3D collagen I/FN matrix that could be stretched uniaxially. These matrices were subjected to 10% static stretch for 24 h, which was sufficient to induce orientation of the cells along the stretch axis. This correlated with the formation of vinculin-positive FA-like structures at the end of F-actin stress fibers416.  To assess whether applying mechanical stretch in 3D regulated the mRNA levels of the three signature genes, I compared cells that were stretched (static stretch) for 24 h to cells in identical matrices that were left unstretched. Compared to unstretched cells the mRNA level for Cyr61 was upregulated ~1.6-fold, and for MUC18 ~1.7-fold, whereas TRPM1 mRNA levels were downregulated ~1.4-fold. This suggests that in 3D, mechanical forces are a regulator of the signature genes. However, the effect of mechanical stretch in 3D is weaker than increasing FN ligand density in 2D.   101   Figure 15: Mechanical stretch in 3D is a weak regulator of the signature genes.  B16F1 cells were embedded in collagen I/FN gels. Gels were either subjected to 10% uniaxially stretch for 24 h, or left unstretched, and then analyzed for mRNA level by qRT-PCR. Mean and SD for 5 independent experiments. Asterisks indicate significant differences (Student’s paired t-test, *, p < 0.05; **, p < 0.01).     3.2.5.3 Inhibiting myosin II activity disrupt stress fibers and FA maturation, and weakly regulates Cyr61 and MUC18 gene expression As a complementary loss-of-function approach to test whether expression of the cancer signature genes is modulated by internal mechanical load, I used blebbistatin, a small molecular inhibitor of myosin II activity that reduces actomyosin-based cellular contractility forces668. Actomyosin contractility is the primary mechanism for developing mechanical forces within a cell669. Non-muscle myosin II, which is present in all non-muscle eukaryotic cells, is a hexameric actin-binding protein that consists of two heavy chains, two light chains and two regulatory light   102  chains. Its activity is controlled by phosphorylation and self-assembly into myosin filaments. As a motor protein, myosin molecules can slide along and thereby generate tension between actin filaments that are bridged by myosin complexes. Inhibiting myosin activity decreases adhesion size and alters recruitment of adhesome proteins to these adhesions.  To confirm that blebbistatin treatment disrupted cellular tension in B16F1 cells, I assessed its effect on cell morphology, actin cytoskeleton organization, and FA maturation. In mouse embryonic fibroblasts (MEFs), inhibition of myosin II activity redistributes pTyr-containing FAs to the outer periphery of cells, inhibits actin bundle formation, and reduces vinculin recruitment to FAs670. Therefore, I used these three markers to visualize the effects of blebbistatin on B16F1 cells. As predicted, treating B16F1 cells with blebbistatin overnight led to altered cell morphology. Compared to control (DMSO treated) cells, blebbistatin-treated cells retracted and adopted either a dentritic morphology with long protrusions and no apparent polarization or disrupted the actin cytoskeleton organization in cells that were still able to spread and polarize (Figure 16A and B). In control cells, pTyr- and vinculin-containing FAs were evident at the end of F-actin bundles (Figure 16B, top panel). Treatment with blebbistatin however, reduced adhesion size, altered the spatial distribution of FAs, and changed actin organization such that there were fewer actin bundles and the smaller pTyr-containing FAs were confined to the cell periphery (Figure 16B, bottom panel). Consistent with previous studies671-675, less vinculin was recruited to FAs in blebbistatin-treated cells (Figure 16B, bottom panel), suggesting that blebbistatin disrupts actomyosin-dependent contractility and FA maturation in B16F1 cells.    103   Figure 16: Blebbistatin alters cell morphology, as well as FA size and spatial distribution in B16F1 cells. (A) Phase contrast images of DMSO-treated cells and cells that were treated overnight with 50 µM blebbistatin. Scale bar, 100 µm. (B) B16F1 cells were cultured on FN, treated as in A, and then stained for F-actin, pTyr, and vinculin. Arrows point to co-localization of vinculin and pTyr (yellow). Scale bar, 10µm.     Next, I asked whether blebbistatin impacted the expression of the cancer signature genes. Cyr61 mRNA levels were downregulated by ~1.8-fold and ~2.3-fold when cells were treated with 50 µM and 100 µM blebbistatin, respectively (Figure 17). MUC18 gene expression was not significantly altered by 5 µM or 50 µM blebbistatin but was upregulated by ~1.7-fold at 100 µM   104  blebbistatin (Figure 17). TRPM1 expression was not altered by any of these blebbistatin concentrations (Figure 17). Thus consistent with the stretching data, actomyosin activity is less important as a regulatory mechanism for the expression of the cancer signature genes than cell adhesion and integrin ligand density.   Figure 17: Myosin II activity weakly regulates the expression of Cyr61 and MUC18. B16F1 cells were cultured with the indicated concentrations of blebbistatin overnight and then analyzed for mRNA levels by qRT-PCR. Mean and SD of 3 independent experiments. Asterisks indicate significant differences (Student’s paired t-test, *, p < 0.05; ns, not significant).    3.2.5.4 The Rap GTPases regulate B16F1 cell morphology but are weak regulators of the signature genes I have shown above that the expression of the signature genes is regulated by adhesion-induced signalling and by FN density, whereas mechanical loads and cytoskeletal tension play only a minor role in regulating Cyr61 and have little or no effect on the expression of MUC18 or TRPM1. The Rap GTPases are key modulators of integrin-activation, cell spreading and   105  cytoskeletal dynamics in B16F1 cells415. This raised the possibility that Rap activation may regulate the cancer signature genes via one of these mechanisms. To test this, I used a loss-of-function approach in which the expression of a Rap-specific GAP is used to block the activation of the Rap1 and Rap2 GTPases, as done previously676-679. To modulate Rap activation I used a B16F1 cell line previously created by stable transfection with a plasmid expressing RapGapII, a Rap-specific GAP415. Compared to vector control cells, RapGapII cells have similar expression of Rap1 and Rap2 protein (Figure 18A). However, Rap activation assays in which a GST-RalGDS fusion protein is used to selectively pull down activated Rap showed that RapGAPII expression greatly reduced the amount of activated Rap1 and Rap2 in cells cultured on TC plastic for 4 h or overnight (Figure 18A). Blocking Rap activation also altered cell morphology. After plating B16F1 on TC plastic for 4 h, RapGapII-expressing cells did not spread to the same extent as vector cells, with a 25% reduction in cell spreading area (Figure 18B and C).  Next, I asked whether inhibition of Rap activity affected expression of the cancer signature genes. When RapGapII was used to block Rap1/Rap2 activation, Cyr61 mRNA level was slightly decreased (~1.4-fold), but was not statistically significant. MUC18 and TRPM1 mRNA levels were also affected only to a small degree, with a ~1.5-fold downregulation of MUC18 mRNA level and a ~1.25-fold upregulation of TRPM1mRNA level (Figure 18D). This trend is consistent with my earlier findings that cell adhesion and spreading, which is inhibited by RapGapII upregulates Cyr61 and MUC18 expression and downregulates TRPM1 expression. Although Rap activity regulates the expression of the three signature genes, this is only a minor effect compared to cell adhesion and FN density.      106     Figure 18: Blocking Rap activation weakly affects cancer signature gene expression. (A) Rap activation assay after RapGapII and control B16F1 cells were plated on TC plastic for 4 h or overnight. Dashed lines indicate cropped blots for representation purpose (B) Images were acquired after cells had been plated on TC plastic for 4 h. (C) Cell area were determined from images as in B (Median with interquartile range, >50 cells, Mann-Whitney test, ***, p < 0.001). (D) Fold change of signature genes in RapGapII vs control on TC plastic determined by qRT-PCR. Mean and SD for 3 independent experiments for Cyr61 and 4 independent experiments for MUC18 and TRPM1. Panel A is representative of 2 independent experiments. Panel B is representative of 3 independent experiments. Panel C is from 3 experiments. Asterisks indicate significant differences (Student’s paired t-test, *, p < 0.05; **, p < 0.01; ****, p < 0.0001, ns = not significant).   107    A more physiological model of the tumor microenvironment than TC plastic is 3D ECM matrices. Compared to FN-coated glass or TC plastic, 3D matrices such as collagen I/FN gels differ in the presentation of cues (cell adhesion, mechanical forces and diffusible factors680) that impact cell functions. The most striking difference between cells grown on 2D surfaces and in 3D ECM matrices is their morphology. For example, cells grown on FN-coated glass or TC plastic are flat, and can adhere and spread freely on the horizontal plane, but have no possibility for spreading in the vertical dimension. This pre-determines an apical-basal polarity because adhesion molecules (e.g. integrins) are only engaged on the side of the cell-2D substrate interface. In contrast, cells embedded in a collagen I/FN gels are not subjected to prescribed polarity and their adhesions are not restricted to one side but are present in all three dimensions. Furthermore, lower stiffness (kPa range in 3D instead of GPa range in 2D), and sterical hindering of spreading in all three dimensions may impact integrin adhesome composition, integrin signalling and cytoskeletal organization. Consequently, cells in 3D ECM matrices adopt a round or stellate morphology with long actin-rich extensions680. Rap activation is important for B16F1 cells invasion into collagen/FN gels as well as for tumor growth and tumor metastasis415. When cultured in collagen/FN gels, RapGapII-expressing B16F1 formed smaller colonies and had reduced outgrowth into the surrounding matrix than vector control cells (Figure 19A, white arrows, and Freeman et al.416).  The differences in cell morphology and growth observed in 2D cultures versus 3D matrices suggest that 3D collagen/FN gels may be a better model for exploring how Rap activation affects cancer progression. Therefore, I tested whether blocking Rap activation altered   108  the expression of the cancer signature genes when B16F1 cells were cultured in 3D collagen I/FN gels. Similar to what was observed on TC plastic, blocking Rap activation had only minor effects on signature gene expression. Cyr61 and MUC18 mRNA level were slightly lower compared to control cells, whereas TRPM1 mRNA levels were not significantly different (Figure 19B). However, it remains to be determined, whether increasing Rap activity in B16F1 cells is able to regulate the signature genes.     109   Figure 19: Blocking Rap activity is a weak regulator of the signature gene expression. (A) Vector control and RapGAPII-expressing B16F1 cells were cultured in collagen I/FN matrices for 24 h or 48 h and stained for F-actin (red) and nuclei (blue, DAPI). Scale bar, 100 µm. White arrows point to vector cells extending membrane processes into the matrix. Representative images are shown from 4 independent experiments with similar results. (B) Fold change of signature genes in RapGapII vs control cells cultured as in A for 24 h or 48 h. Mean with range from 2 independent experiments for 24 h time point and mean and SD for 4 independent experiments for the 48 h time point. Asterisks indicate significant differences (Student’s paired t-test, *, p < 0.05; ns, not significant).      110  3.3 Summary and discussion  In this chapter Cyr61, MUC18 and TRPM1 were used as cancer signature genes to establish a readout for ECM-induced changes in gene expression in B16F1 melanoma cells. All three genes have previously been linked to cancer progression in the literature. Cyr61  is a member of the CCN (acronym for Connective tissue growth factor, Cyr61 and Nephroblastoma overexpressed gene) family and expressed in multiple cell types. It is a multifunctional protein that plays essential roles in embryogenesis, inflammation and wound healing681-683. As for other CCN proteins, Cyr61 contains a secretory peptide and binds tightly to the ECM after being secreted684. Interestingly, it can also bind directly to the extracellular domain of integrins in an RGD independent manner685-690, suggesting a role for Cyr61 in “outside-in” signaling of integrins691.  MUC18 has been implicated in disease progression692 in melanoma and may promote the development of multiple cancer hallmarks including cell survival, angiogenesis, cell motility, invasion, metastasis, and tumor-promoting inflammation (reviewed in Lei et al.692). MUC18 contributes to melanoma cell metastasis by promoting adhesion693, 694, increasing extravasation, and promoting the growth of new foci623, 695. Recently MUC18 has been identified as a marker of the progression of melanoma and may serve as a predictor of recurrences, disease progression, or risk of relapse. Taken together, this qualifies MUC18 as a cancer signature gene in the context of melanoma.  TRPM1 was originally identified in B16F1 cells as a melanoma metastasis suppressor gene. Duncan et al. found that TRPM1 mRNA is highly expressed in the poorly metastatic B16F0 variants but greatly reduced in the highly metastatic B16F10 variant626.    111   Clinical studies have used TRPM1 as a prognostic tool for disease-free survival of melanoma patients in addition to traditional prognostic factors such as tumor thickness, mitotic rate, ulceration, and tumor-infiltrating lymphocytes696-698. Down-regulation of TRPM1 in melanoma cells correlates with a shorter disease-free survival period for patients with Stage I and II melanomas (American Joint Committee on Cancer (AJCC) staging697). Despite the utility of TRPM1 as a diagnostic marker, its molecular functions in melanoma are not known. Interestingly, besides coding for the TRPM1 protein, the TRPM1 gene also encodes the micro RNA, miR-211 within its sixth intron699, 700. miR-211 shares the same promoter as TRPM1 and is co-regulated with TRPM1 by the microphthalmia TF (MITF). mir-211 is thought to function as a tumor suppressor, consistent with the association between downregulated TRPM1/miR-211 mRNA and increased tumor aggressiveness699, 701. The current hypothesis is that the TRPM1 protein regulates melanogenesis and Ca2+ homoeostasis in normal melanocytes, whereas miR-211 acts as a tumor suppressor702. Loss of either or both may promote the transition from normal melanocyte to melanoma. This makes the TRPM1 mRNA an excellent marker for melanoma aggressiveness and disease progression.   Here, I investigated how adhesion to stiff surfaces, integrin ligand density and cytoskeletal tension affect these three cancer signature genes (Figure 20). Cell adhesion to TC plastic, and to increasing densities of FN, was a strong regulator of cancer signature gene expression, whereas modulating mechanical loads by cell stretching, inhibition of actomyosin contractility or blocking Rap activity had only weak regulatory effects. Thus, in our system the expression of the signature genes appears to be regulated by adhesion independent of mechanical loads.   112   Figure 20: Cell adhesion and adhesion to increasing FN density strongly regulate the signature genes.   The observed changes in the mRNA levels of the cancer signature genes could be due to several mechanisms. These include changes in mRNA stability and processing, as well as activation or inhibition of transcription factors that target the cancer signature genes. The next step could be to test whether changes in the mRNA levels of Cyr61, MUC18 and TRPM1 are regulated by de novo mRNA synthesis rather than changes in mRNA stability or processing. To do so, I could chemically block transcription in B16F1 cells and test whether this abolishes the observed changes. Several TFs have been identified for Cyr61, MUC18 and TRPM1. Cyr61 transcription is regulated by the TF of the CREB, Sp-1 and AP-1 families and is induced downstream of RhoA, JNK, ERK and p38 MAP kinase activity703, all of which can be activated   113  by integrin-dependent signalling (see 1.9.1) Furthermore, the promoter region also contains binding sites for the ATF, E2F, HNF3b, NF1, NFкB, and SRF TFs. MUC18 mRNA levels are regulated by the TFs AP-2, CREB704 and ZBTB7A705. Transcription of TRPM1 is controlled primarily by MITF706, a TF whose expression is controlled by CREB. Whether CREB is activated downstream of integrin signalling during cell-ECM adhesion and mechanical stretch in B16F1 cells and regulates the cancer signature genes in these cells needs to be determined.    Cell adhesion to TC plastic was sufficient to induce adhesome signalling, as indicated by high phosphorylation levels of FAK, paxillin and p130Cas. This may be due to the physical properties of TC plastic, such as negative charge and hydrophilicity, which act as a pseudosubstrate that promotes cell attachment and spreading707. To factor out additional effects caused by cell-cell contact, I allowed for cell-cell contact in all experimental conditions. However, as my experiments were carried out in 10% FCS, which contains FN and vitronectin, I cannot exclude the possibility that the cells adhered to ECM proteins that were able to coat the TC plastic during the 24 h duration of the experiments. Nevertheless, TC plastic can serve as a platform to induce adhesion signalling that supports cell spreading and regulates signature gene expression.  In contrast to TC plastic, B16F1 cells adhered poorly to silicone rubber surfaces, even in the presence of serum, unless the surface was coated with FN. Low FN density (0.025µg/cm2) was sufficient to allow for cell attachment. This is in accordance with Cornelissen et al.612, who found that uncoated PDMS membranes (such as the ones used here) did not support the adhesion of endothelial cells but that coating these membranes with 2.8 µg/cm2 FN allowed for cell attachment at 24 h post seeding. However, neither the effect of different FN densities on cell adhesion nor the effect on cell spreading was addressed in this study.    114  Increasing the FN density on the silicone rubber surfaces regulated the cancer signature genes in the same manner as plating the cells on TC plastic. Interestingly, the fold changes induced by adhesion to TC plastic were as high as the fold changes induced from low to high FN on silicone rubber. This may relate to the spreading of the cells, because the morphology of cells plated on high FN (spread out and flat) resembles that of cells cultured on TC plastic. This suggests that B16F1 cells adopt similar phenotypes when cultured on compliant surfaces with high integrin ligand density and on surfaces of high stiffness. However, additional experiments in which the ligand density is the same but surface stiffness is varied (e.g. FN embedded in polyacrylamide gels with different degrees of crosslinking, as in Chaudhri et al.337) could clarify the contributions of ligand density versus substrate stiffness on the expression of the signature genes.   One way to test if adhesion-dependent signalling initiated by high FN density and high substrate rigidity is quantitatively and qualitatively similar is to compare the compositions of adhesomes formed under these conditions, e.g. the phosphorylation states of FAK, paxillin and p130Cas at adhesion sites, as done in Robertson et al.380.  Increased spreading, caused either by adhesion to rigid TC plastic or to a high density integrin ligand, also leads to increased cell-ECM contact area, which results in an increased number of adhesion complexes as well as increased intracellular tension641, 708. It is not clear whether adhesion signalling, or increased tension, or the combination thereof, regulate expression of the cancer signature genes. To address this, I tested whether mechanobiological mechanisms also regulate the signature genes. I approached this in three ways. First I induced extracellular tension by mechanically stretching the cells on FN coated silicone rubber surfaces. Second, I used blebbistatin to inhibit actomyosin contractility in order to reduce internal   115  mechanical loads. Third, I blocked the activation of the Rap GTPase, which is a mechanosensor that modulates FA dynamics in response to changes in mechanical loads in B16F1 cells. In contrast to modulating adhesion, modulating mechanical loads had significantly less effect on the expression levels of the cancer signature genes than increasing FN density, despite having strong effects on cell morphology, FA number, organization, and composition. To further address whether Rap activation can alter the expression of the signature gene, in the future I will express a constitutive active form of Rap1 (Rap1V12) in B16F1 cells.  Inducing cellular tension by mechanical stretching in 2D did not cause significant changes in expression of the signature genes. This was surprising, particularly for Cyr61, which is regulated by mechanical stretch in bladder smooth muscle cells709. However, in that study, mRNA levels only increased for the duration of the 1 h cyclic stretch and decreased to the levels of unstretched cells at 2-16 h of stretch. During static stretch, Cyr61 mRNA levels were only increased for the first 30 min. These differences in experimental conditions between this study and my experiments could account for the different results. Longer exposure to stretch, as used here, could allow cells to adapt over time. Experiments by Freeman et al.416 showed that short-term exposure (10 min to 4 h) of static stretch with the same magnitude as in this study increased vinculin recruitment to adhesion site. However, whether cells are able to adapt over 24 h needs to be addressed and the strength (for static stretch) or amplitude and frequency (for cyclic stretch) may have to be changed to subject the cells to prolonged tension. Initial gene expression changes may have been missed as the cells reached a new equilibrium. To address this, a time course that includes both short and long duration of stretching might detect initial changes in gene expression that were not detected with only endpoint readout.    116  Both sustained increases in tension and short term changes in tension are biologically relevant during tumor progression. During tumorigenesis, the primary tumor may be subjected to sustained changes in the microenvironment that increase tension, for example, during migration to regions of stiffer substrate. In contrast, when tumor cells leave the primary tumor and migrate to secondary sites, tumor cells may encounter transient changes in the microenvironment as they intravasate into vessels and extravasate into other tissues. Because both sustained and transient conditions could be important to drive gene expression changes that lead to cancer progression, future studies should investigate the time course of ECM-induced changes in gene expression in more detail, and on global scale (i.e. transcriptomics). Compared to stretching in 2D, which had little or no effect on the cancer signature genes, stretching B16F1 cells that were embedded in collagen 1/FN gels weakly regulated the three signature genes in the same manner as increasing FN density in 2D. The difference between 2D and 3D stretching may reflect differences between adhesome composition, dynamics, and signalling in cells on 2D ECM versus in 3D matrices710. For example, integrin-mediated adhesions in 2D are restricted to one plane of the cell, whereas in 3D culture these adhesions form all around the cell surface. Furthermore, applied stretch on 2D substrates leads to force transmission along parallel stress fibers that are located close to the adhesion plane, whereas in cells embedded in 3D matrices, mechanically applied forces travel through the midline of the cells680. As well, adhesion to 2D surfaces promotes the formation of actin-rich lamellipodia whereas cells embedded in 3D matrices adopt a stellate morphology. This implies that cells in 2D and 3D have different actin dynamics, which in turn may differentially regulate cytoskeleton-associated transcriptional co-activators such as ABPs. Taken together, the differences in the spatial distributions of adhesions, force propagation, and actin dynamics between 2D surfaces   117  and 3D matrices could result in differences in the activation of integrin-regulated TFs that control expression of the cancer signature genes.  As a complement to increasing tension, I reduced actomyosin contractility by using a chemical inhibitor of myosin II. Reducing internal mechanical load also had little or no effect on the expression of MUC18 and TRPM1. However, Cyr61 mRNA levels were decreased by blebbistatin in a dose-dependent manner, indicating that actomyosin contractility, or subsequent adhesion maturation events regulate Cyr61 mRNA levels to some extent. Importantly, neither increased extracellular mechanical load nor decreased internal mechanical load altered signature gene mRNA levels to the same extent as adhesion to FN or TC plastic. Consistent with these findings, modulating Rap activity also had a limited effect on the cancer signature gene expression. Thus, in B16F1 cells, the cancer signature genes appear to be primarily regulated by integrin adhesion signalling, independent of the mechanobiological processes that are mediated by ECM-integrin-cytoskeletal connections.  Integrin signalling is mediated by tension-dependent and tension-independent pathways, which may depend on different subsets of adhesome proteins. For example, the recruitment of some adhesome proteins, e.g. FAK, zyxin, α-actinin and vinculin, to FAs is promoted by actomyosin contractility, whereas other adhesome proteins such as paxillin and talin may be recruited to FAs independently of cellular tension671. The results in this chapter suggest that the signature genes are regulated primarily by integrin signalling that is independent of tension. By knocking down individual adhesome proteins such as talin or FAK, or by inhibiting their activity, I could determine whether expression of the signature genes is dependent on signalling that is downstream of specific adhesome components (see chapter 5). Furthermore, experiments comparing changes in gene expression and adhesome composition that are induced by adhesion   118  versus modulating mechanical load (e.g. stretching) would enable us to distinguish between ‘adhesion ‘ pathways and ‘tensional’ pathways, and allow us to identify key transcriptional programs that are dependent on each of these pathways (see chapter 4). The results from such studies could allow us to identify ECM-dependent signalling pathways that contribute to cancer progression, or to other diseases that are influenced by changes in the ECM such as atherosclerosis or liver cirrhosis. In the long term, this could lead to the discovery of novel disease markers or therapeutic approaches.   119  Chapter 4: FN density and cell stretching result in overlapping but distinct patterns of gene regulation  The ECM surrounding cancer cells often exhibits characteristic changes including altered composition, increased density of ECM proteins, and increased ECM stiffness that is caused by increased crosslinking of ECM proteins711. This phenomenon, which is referred to as desmoplasia, is a major contributor to tumorigenesis and tumor progression, and may also drive metastasis242, 251, 301. Importantly, it is a marker of poor prognosis for several types of cancer265. Cell-ECM contact governs many aspects of cellular behavior. This includes the formation of adhesions and their turnover, which affects cell migration, cell survival and cell division. Interaction with the ECM also drives changes in gene expression that regulate cell proliferation and cell fate. Dysregulation of these ECM-regulated stimuli can drive cancer progression by altering the expression of genes that promote cell proliferation, survival, EMT, and metastasis. Therefore, understanding the effects of ECM-stimulated signalling on the regulation of gene networks could identify key regulatory nodes and suggest novel therapeutic strategies for treating cancer. ECM ligand density, ECM composition, and ECM stiffness have been shown to separately and collectively regulate cell behavior337. For example, in cell culture systems where ECM stiffness can be modulated independently from ligand density, it has been shown that increasing ECM stiffness is sufficient to drive mammary epithelial cells towards a malignant phenotype. Conversely, increasing ECM ligand density on a stiff matrix reverses the ability of ECM stiffness to promote this phenotype. Together, ligand density, composition, and stiffness determine the mechanical load on adherent cells. Gene expression changes in response to these mechanical loads can determine how cells react to these stimuli. Mechanical transduction can   120  regulate the expression of genes that are involved in remodelling of the ECM, which in turn affects the forces that the cell will be subjected to650, 712. For example, biaxial stretch applied to human ligament fibroblasts increases MMP-2 and TIMP-2 mRNA levels, two enzymes that can remodel the ECM and thereby affect external mechanical load650. Similarly, shear stress applied to colon cancer cells regulates genes that are linked to the ERK, JNK and NFкB pathways, which regulate characteristic responses to externally-applied forces712. Identifying genes that are regulated by these mechanical forces, and understanding the contributions of ligand density, composition and stiffness to regulate these genes in cancerous and pre-cancerous cells, could identify novel targets for molecular therapy of cancer. In chapter 3, an initial RNA-seq analysis comparing the transcriptomes of B16F1 cells cultured on increasing amounts of FN led to the identification of three signature genes that were regulated by increasing ECM density but only weakly by mechanical stretch. To gain a better understanding of how ECM density versus tension affects gene expression in B16F1 cells, I repeated the RNA-seq transcriptome profiling of B16F1 cells that were plated on increasing densities of FN. Samples were run in two lanes at PE75s (paired-end 75 bp) on a HISeq2500 in high output mode. For each library ~100 million reads were generated and ~75 million aligned within gene boundaries. This resulted in an increased coverage of the genome, and with that confidence in the results. It also allows for detection of fold changes for genes with low expression and enables the detection of changes in exon usage.  In addition, I used RNA-seq to assess transcriptional changes that were induced by increasing tension, with the goal of distinguishing between genes that were regulated by increased ligand density versus induced tension. To increase tension, I applied mechanical stretching forces to cells that were embedded in 3D collagen I/FN matrices. Comparing the   121  global transcriptional changes between these two approaches revealed significant differences between the gene sets that are regulated by ligand density and stretch. Of the genes regulated by these two types of micro-environmental changes, there was only ~8% overlap. Furthermore, genes regulated by integrin ligand density were primarily associated with pathways that regulate ECM degradation, cell motility, and migration, whereas the genes regulated by mechanical stretch were primarily associated with metabolism and hypoxia pathways. This suggests that integrin ligand density and mechanical stretching of cells regulate different transcriptional programs.  4.1 Increasing FN density and applying stretch regulate different set of genes To identify the transcriptional changes that were induced by altering FN density or applying stretch, I used RNA-seq. To increase ECM ligand density on a more compliant surface than TC plastic, I cultured B16F1 cells on silicone rubber plates that had been coated with low, medium or high densities of FN (as in chapter 3). B16F1 cells adhere and spread on this FN-coated substrate in a dose-dependent manner (Figure 10A), illustrating that the densities influenced cell behavior. To be considered a ‘FN density-regulated gene’ a particular mRNA had to have significantly different expression levels between low to medium FN density or between low to high FN density (Figure 21A). Significant gene expression changes were identified using t-test with correction for multiple comparisons and only genes upregulated by > 1.5-fold or downregulated by < 1.5-fold were considered for subsequent analysis. To include genes that were sensitive to either a weak integrin-ligand stimulus (low to medium FN) or a strong integrin-ligand stimulus (low to high FN) I combined the significant regulated genes from both sets (termed density-regulated genes). Genes that were significantly   122  regulated in the opposite direction between low-to-medium FN and low-to-high FN were ignored for the time being. Based on these criteria, I identified 687 genes that were downregulated and 435 genes that were upregulated as the FN density was increased (Figure 21A).   To apply stretch to B16F1 cells in an in vivo-like setting, I embedded B16F1 cells in 3D collagen I/FN gels. These 3D gels were attached to silicone anchors so that the majority of the gel was free-floating. To generate external mechanical stress, a uniaxial stretch was applied to the gels for 24 h. The applied tension caused deformation of the gel, which changed from a rectangular to a parabolic shape. Importantly, the cells embedded in these gels aligned their protrusions along the direction of the applied stretch, indicating that they were responding to the external forces416. To identify genes that were regulated by stretch, I compared the transcriptomes of B16F1 cells that were cultured in a collagen I/FN gel with or without the 10% uniaxial stretch. This resulted in the identification of 546 stretch-regulated genes, of which 180 were downregulated and 366 were upregulated (Figure 21B). This shows that both FN density and applied stretch regulate gene expression in B16F1 cells.     123   Figure 21: Changes in mRNA levels upon FN density or cellular stretch.  (A) Schematics of B16F1 cells cultured on low, medium or high FN density. Significant transcriptional changes between low to medium FN density or between low to high FN density were combined to generate the set of FN density-regulated genes. (B) The schematic indicates the direction of stretch (arrows) that was applied to B16F1 cells that were embedded in 3D collagen I/FN gels. All comparisons were done in duplicate. Significantly regulated genes were identified by multiple t-tests. Expression changes with adjusted p-values < 0.05, using the correction for multiple comparisons (Holm-Sidak method), were defined as ‘significantly different’.      In Table 18 to Table 23 the top 10% fold changes for upregulated and downregulated genes of each category (FN low to med, low to high, stretch vs no stretch averaged for 2 biological repeats) are summarized with adjusted p-values. The signature genes are highlighted in each table. I also validated the changes in signature gene expression by qRT-PCR from the same cDNA that was submitted for RNA-seq. The fold changes in Cyr61, MUC18, and TRPM1 mRNA levels were comparable to previous experiments (Compare Figure 22A to Figure 11 in Chapter 3). The RNA-seq data and qRT-PCR data exhibited the same trends in gene expression   124  changes (Figure 22B). Fold changes were similar for the FN density-associated responses but were more variable for changes in gene expression induced by stretch.   Signature gene Medium/low FN High/low FN Stretch/no stretch  Rep 1 Rep2 Rep 1 Rep2 Rep1 Rep2 Cyr61 2.074885 2.130407 3.410323 2.994299 3.816811 2.181426 MUC18 2.543912 2.597994 4.077191 3.907998 3.399707 1.841229 TRPM1 -1.800422 -1.716498 -2.705100 -2.484157 -1.583195 -1.285492  Figure 22: Validation of fold changes in signature genes by qRT-PCR. Signature gene expression in total RNA from the same cells was analyzed by qRT-PCR and RNA-Seq. (A) Fold change of signature genes between medium versus low FN (black bars), and high versus high low (gray bars). The bars represent the average value of the fold changes and the values for each of the two replicates are represented by the red dots. (B) Total RNA from the same experiments was analyzed RNA-seq. Fold changes for the independent biological replicates, replicate 1 and replicate 2, are shown.     125  Table 18: Top 10% upregulated genes from low to medium FN  Gene symbol  Gene name  Fold change  Adjusted P-value Krt80 keratin-80 3.39 <1.00E-35 Sucnr1 succinate receptor 1 2.93 7.254E-33 Ccdc37 coiled-coil domain containing 37 2.88 3.9E-35 Ifi47  2.86 1.48E-34 Hist1h3a histone cluster 1, H3a 2.86 2.809E-30 Gm17567  2.75 9.222E-20 Prkcq protein kinase C, theta 2.72 1.244E-31 Gm11627  2.71 3.454E-31 Fbxw15 F-box and WD-40 domain protein 18 2.63 9.576E-29 Cfap70 cilia and flagella associated protein 70 2.6 1.133E-28 1700001P01Rik  2.58 1.597E-17 Igfbp7 insulin-like growth factor binding protein 7 2.57 1.476E-27 MUC18 melanoma cell adhesion molecule 2.57 2.908E-28 Fbxw24 F-box and WD-40 domain protein 24 2.53 2.212E-26 Rassf4 Ras association (RalGDS/AF-6) domain family member 4 2.53 4.509E-27 Slc35f1 solute carrier family 35, member F1 2.44 1.668E-25 Igf2bp1 insulin-like growth factor 2 mRNA binding protein 1 2.43 4.118E-25 Gm10337  2.38 8.325E-15 Gm17655  2.38 2.135E-22 Fam184b family with sequence similarity 184, member B 2.35 2.648E-23 Ramp3 receptor (G protein-coupled) activity modifying protein 3 2.31 2.895E-22 Ptchd1 patched domain containing 1 2.28 4.748E-19 Tagln transgelin 2.27 2.836E-21 Kank4 KN motif and ankyrin repeat domains 4 2.27 1.066E-21 Fbxw18 F-box and WD-40 domain protein 18 2.26 1.77E-21 Gm9922  2.24 8.985E-21 Adh7 alcohol dehydrogenase 7 (class IV), mu or sigma polypeptide 2.21 1.906E-20 Frmd3 FERM domain containing 3 2.2 2.67E-20   126   Gene symbol  Gene name  Fold change  Adjusted P-value Gal3st2 galactose-3-O-sulfotransferase 2 2.18 4.747E-19 Art4 ADP-ribosyltransferase 4 (Dombrock blood group) 2.12 1.492E-18 Sema3d sema domain, immunoglobulin domain (Ig), short basic domain, secreted, (semaphorin) 3D 2.11 4.302E-18 Tecrl trans-2,3-enoyl-CoA reductase-like 2.1 8.612E-18 Cyr61 cysteine-rich, angiogenic inducer, 61 2.1 4.242E-18 Slc40a1 solute carrier family 40 (iron-regulated transporter), member 1 2.1 4.631E-18 Gm6594  2.09 1.514E-15 1700104B16Rik  2.07 1.293E-16 Masp1 mannan-binding lectin serine peptidase 1 (C4/C2 activating component of Ra-reactive factor) 2.07 1.592E-17 Thbs1 thrombospondin 1 2.07 1.442E-17 Aldh1a3 aldehyde dehydrogenase 1 family, member A3 2.07 1.825E-17 Gm21953  2.07 1.89E-16 Gm11037  2.07 2.903E-16 Adgrf2 adhesion G protein-coupled receptor F2 2.06 2.272E-17       127  Table 19: Top 10% upregulated genes from low to high FN  Gene symbol  Gene name  Fold change  Adjusted P-value Sucnr1 succinate receptor 1 7.25 <1.00E-35 Cfap70 cilia and flagella associated protein 70 5.66 <1.00E-35 Ccdc37 coiled-coil domain containing 37 5.31 <1.00E-35 Krt80 keratin 80 4.86 <1.00E-35 Rassf4 Ras association (RalGDS/AF-6) domain family member 4 4.72 <1.00E-35 Adh7 alcohol dehydrogenase 7 (class IV), mu or sigma polypeptide 4.57 <1.00E-35 Ifi47 interferon gamma inducible protein 47 4.49 <1.00E-35 Tagln transgelin 4.39 2.72E-12 Prkcq protein kinase C, theta 4.31 <1.00E-35 Igfbp7 insulin-like growth factor binding protein 7 4.01 <1.00E-35 MUC18 melanoma cell adhesion molecule 3.98 <1.00E-35 Serpine1 serpin peptidase inhibitor, clade E (nexin, plasminogen activator inhibitor type 1), member 1 3.79 <1.00E-35 Tecrl trans-2,3-enoyl-CoA reductase-like 3.51 3.27E-35 Gm11627  3.48 8.30E-35 Slc40a1 solute carrier family 40 (iron-regulated transporter), member 1 3.4 4.13E-35 Rgs16 regulator of G-protein signaling 16 3.31 2.11E-33 Masp1 mannan-binding lectin serine peptidase 1 (C4/C2 activating component of Ra-reactive factor) 3.27 8.36E-33 Gal3st2 galactose-3-O-sulfotransferase 2 3.26 5.70E-32 Gm11037  3.24 6.51E-27 Cyr61 cysteine-rich, angiogenic inducer, 61 3.19 1.11E-31 Asb15 ankyrin repeat and SOCS box containing 15 3.18 3.72E-30 Thbs1 thrombospondin 1 3.17 2.23E-31 Fhl2 four and a half LIM domains 2 3.15 8.75E-31 Gsta3 glutathione S-transferase alpha 3 3.08 7.84E-30   128   Gene symbol  Gene name  Fold change  Adjusted P-value 2310030G06Rik  2.96 1.28E-27 Prss23 protease, serine, 23 2.93 3.86E-27 Igf2bp1 insulin-like growth factor 2 mRNA binding protein 1 2.92 3.98E-27 Cdc42ep1 CDC42 effector protein (Rho GTPase binding) 1 2.92 2.60E-27 Aldh1a3 aldehyde dehydrogenase 1 family, member A3 2.91 3.84E-27 Fbxw15  2.9 1.56E-26 Ptchd1 patched domain containing 1 2.87 2.93E-26 Gbp2 guanylate binding protein 2, interferon-inducible 2.86 4.31E-26 Mal mal, T-cell differentiation protein 2.85 5.41E-26 Fbxw24 F-box and WD-40 domain protein 24 2.82 1.11E-25 Duox1 dual oxidase 1 2.82 3.38E-25 Slc7a11 solute carrier family 7 (anionic amino acid transporter light chain, xc- system), member 11 2.81 3.74E-25 Fam184b family with sequence similarity 184, member B 2.81 2.05E-25 Adgrf2 adhesion G protein-coupled receptor F2 2.81 6.14E-25 Prl2c2  2.81 8.19E-25 Rasl11a RAS-like, family 11, member A 2.8 5.23E-24 Gm14221 prolactin family 2, subfamily c, member 2 2.77 1.16E-16 Ctgf connective tissue growth factor 2.76 1.05E-24 Slc35f1 solute carrier family 35, member F1 2.75 1.46E-24 Neb nebulin 2.75 5.44E-23       129  Table 20: Top 10% upregulated genes from no stretch to stretch  Gene symbol  Gene name  Fold change  Adjusted P-value Car9 carbonic anhydrase 9 7.16 <1.00E-35 Stc1 stanniocalcin 1 6.13 <1.00E-35 1600014C23Rik  5.74 <1.00E-35 Selenbp1 selenium binding protein 1 5.69 <1.00E-35 Selenbp2 selenium binding protein 2 5.68 <1.00E-35 Tnfrsf9 tumor necrosis factor receptor superfamily, member 9 5.42 <1.00E-35 Aire autoimmune regulator 5.37 <1.00E-35 Gm4978  4.99 <1.00E-35 D830013O20Rik  4.35 <1.00E-35 Adm adrenomedullin 4.29 <1.00E-35 Ero1l ERO1-like (S. cerevisiae) 3.93 <1.00E-35 Ankrd37 ankyrin repeat domain 37 3.92 <1.00E-35 Masp1 mannan-binding lectin serine peptidase 1 (C4/C2 activating component of Ra-reactive factor) 3.86 <1.00E-35 Pde1b phosphodiesterase 1B, calmodulin-dependent 3.82 <1.00E-35 Espn espin 3.64 <1.00E-35 Bnip3 BCL2/adenovirus E1B 19kDa interacting protein 3-like 3.61 <1.00E-35 Loxl2 lysyl oxidase-like 2 3.52 <1.00E-35 Creb5 cAMP responsive element binding protein 5 3.5 <1.00E-35 Kcnk2 potassium channel, subfamily K, member 2 3.45 <1.00E-35 Maff v-maf musculoaponeurotic fibrosarcoma oncogene homolog F (avian) 3.43 <1.00E-35 Cdc42ep2 CDC42 effector protein (Rho GTPase binding) 2 3.42 <1.00E-35 Gm21451  3.35 <1.00E-35 Plekha2 pleckstrin homology domain containing, family A (phosphoinositide binding specific) member 2 3.32 <1.00E-35   130   Gene symbol  Gene name  Fold change  Adjusted P-value Rasd2 RASD family, member 2 3.19 <1.00E-35 Gm6316  3.18 <1.00E-35 Fhad1 forkhead-associated (FHA) phosphopeptide binding domain 1 3.06 3.468E-36 Proser2 proline and serine rich 2 3.01 7.261E-37 Ier3 immediate early response 3 3 4.516E-36 Cyr61 cysteine-rich, angiogenic inducer, 61 2.99 4.442E-34 Lgals3 lectin, galactoside-binding, soluble, 3 2.99 4.744E-35 Rcor2 REST corepressor 2 2.9 4.29E-34 Hdac9 histone deacetylase 9 2.87 1.062E-34 Rem2 RAS (RAD and GEM)-like GTP binding 2 2.86 5.142E-34 Unc13a unc-13 homolog A (C. elegans) 2.86 3.109E-32 Serpinb9b serine (or cysteine) peptidase inhibitor, clade B, member 9b 2.85 7.296E-34       131  Table 21: Top 10% downregulated genes from low to medium FN  Gene symbol  Gene name  Fold change  Adjusted P-value Kcnj13 potassium inwardly-rectifying channel, subfamily J, member 13 -6.58 <1.00E-35 Gm8074  -4.64 <1.00E-35 Egr1 early growth response 1 -3.51 <1.00E-35 Adck3 aarF domain containing kinase 3 -3.03 2.087E-38 Ip6k3 inositol hexakisphosphate kinase 3 -2.97 3.163E-37 Smtnl1 smoothelin-like 1 -2.97 4.501E-37 Mgll monoglyceride lipase -2.92 4.324E-36 Omg oligodendrocyte myelin glycoprotein -2.82 9.661E-34 Sema7a semaphorin 7A, GPI membrane anchor (John Milton Hagen blood group) -2.75 3.329E-32 3110079O15Rik  -2.71 2.076E-31 Col11a2 collagen, type XI, alpha 2 -2.71 2.111E-31 Wfdc3 WAP four-disulfide core domain 3 -2.66 2.07E-30 Kbtbd11 kelch repeat and BTB (POZ) domain containing 11 -2.65 4.15E-30 Thrsp thyroid hormone responsive -2.61 3.005E-29 Rapgef5 Rap guanine nucleotide exchange factor (GEF) 5 -2.59 8.385E-29 Tcp11l2 t-complex 11, testis-specific-like 2 -2.59 8.426E-29 Npnt nephronectin -2.55 7.736E-28 Aldoc aldolase C, fructose-bisphosphate -2.54 1.209E-27 Cck cholecystokinin A receptor -2.51 3.959E-27 Defb25  -2.51 5.145E-27 Fhdc1 FH2 domain containing 1 -2.49 1.098E-26 Il18 interleukin 18 (interferon-gamma-inducing factor) -2.48 2.342E-26 Hist1h2bm histone cluster 1, H2bm -2.48 2.437E-26 Spp1 secreted phosphoprotein 1 -2.47 2.988E-26 Gm10840  -2.47 3.062E-26 Pgf placental growth factor -2.43 3.52E-25 Sgk2 serum/glucocorticoid regulated kinase 2 -2.42 5.287E-25 Gsta4 glutathione S-transferase alpha 4 -2.39 2.26E-24 Pdk2 pyruvate dehydrogenase kinase, isozyme 2 -2.36 1.212E-23 Ddit4 DNA-damage-inducible transcript 4 -2.32 5.855E-23   132   Gene symbol  Gene name  Fold change  Adjusted P-value Ccdc13 coiled-coil domain containing 137 -2.32 8.321E-23 Egr2 early growth response 2 -2.31 1.028E-22 Lrrc14b leucine rich repeat containing 14B -2.3 1.632E-22 Hhatl hedgehog acyltransferase-like -2.3 2.213E-22 Nipal3 NIPA-like domain containing 3 -2.28 6.344E-22 Pik3ip1 phosphoinositide-3-kinase interacting protein 1 -2.26 1.596E-21 Slc51a solute carrier family 51, alpha subunit -2.25 2.158E-21 Arid3c AT rich interactive domain 3C (BRIGHT-like) -2.25 2.296E-21 Abhd3 abhydrolase domain containing 3 -2.25 2.426E-21 Adamtsl4 ADAMTS-like 4 -2.21 1.378E-20 Cited1 Cbp/p300-interacting transactivator, with Glu/Asp-rich carboxy-terminal domain, 1 -2.21 1.449E-20 Ccdc74a coiled-coil domain containing 74A -2.21 1.84E-20 Ptgds prostaglandin D2 synthase 21kDa (brain) -2.2 3.312E-20 Gm10642  -2.17 1.022E-19 Hsd17b11 hydroxysteroid (17-beta) dehydrogenase 11 -2.16 1.829E-19 Mxd4 MAX dimerization protein 4 -2.15 2.818E-19 Nat8b N-acetyltransferase 8B (GCN5-related, putative, gene/pseudogene) -2.15 3.004E-19 Ankrd37 ankyrin repeat domain 37 -2.15 3.703E-19 A230065H16Rik  -2.15 4.256E-19 Art5 ADP-ribosyltransferase 5 -2.14 4.742E-19 Irf4 interferon regulatory factor 4 -2.13 7.901E-19 Adm adrenomedullin -2.12 1.376E-18 Mafa v-maf musculoaponeurotic fibrosarcoma oncogene homolog A (avian) -2.11 2.602E-18 Col9a3 collagen, type IX, alpha 3 -2.1 3.332E-18 Ighg2c immunoglobulin heavy constant gamma 2C -2.1 3.394E-18 Fam83f family with sequence similarity 83, member F -2.1 3.643E-18 Hpse heparanase -2.1 3.697E-18 Ccng2 cyclin G2 -2.1 4.205E-18   133   Gene symbol  Gene name  Fold change  Adjusted P-value Gpr37 G protein-coupled receptor 37 (endothelin receptor type B-like) -2.1 4.739E-18 Gm11084  -2.09 5.306E-18 9030612E09Rik  -2.07 1.928E-17 Hist1h2ba histone cluster 1, H2ba -2.06 3.033E-17 Gm10639  -2.05 3.82E-17 Dyrk1b dual-specificity tyrosine-(Y)-phosphorylation regulated kinase 1B -2.05 4.782E-17 Scrg1 stimulator of chondrogenesis 1 -2.03 1.362E-16       134  Table 22: Top 10% downregulated genes from low to high FN  Gene symbol  Gene name  Fold change  Adjusted P-value Kcnj13 potassium inwardly-rectifying channel, subfamily J, member 13 -14.59 <1.00E-35 Smtnl1 smoothelin-like 1 -5.85 <1.00E-35 Npnt nephronectin -5.1 <1.00E-35 Mgll monoglyceride lipase -5.01 <1.00E-35 Adck3 aarF domain containing kinase 3 -4.91 <1.00E-35 Col11a2 collagen, type XI, alpha 2 -4.24 <1.00E-35 Gpr37 G protein-coupled receptor 37 (endothelin receptor type B-like) -4.08 <1.00E-35 Ip6k3 inositol hexakisphosphate kinase 3 -3.98 <1.00E-35 Gm8074  -3.88 <1.00E-35 Omg oligodendrocyte myelin glycoprotein -3.87 <1.00E-35 Scrg1 stimulator of chondrogenesis 1 -3.87 <1.00E-35 Rapgef5 Rap guanine nucleotide exchange factor (GEF) 5 -3.69 <1.00E-35 Gsta2 glutathione S-transferase alpha 2 -3.6 3.07E-38 Fhdc1 FH2 domain containing 1 -3.57 6.93E-38 Bcan brevican -3.52 5.68E-37 Nipal3 NIPA-like domain containing 3 -3.46 4.40E-36 Tcp11l2 t-complex 11, testis-specific-like 2 -3.41 2.45E-35 Cck cholecystokinin A receptor -3.41 2.49E-35 Egr1 early growth response 1 -3.4 3.54E-35 Ccdc13 coiled-coil domain containing 137 -3.38 7.61E-35 Hpse heparanase -3.38 9.74E-35 Aldoc aldolase C, fructose-bisphosphate -3.29 2.53E-33 Sema7a semaphorin 7A -3.27 5.60E-33 Cited1 Cbp/p300-interacting transactivator, with Glu/Asp-rich carboxy-terminal domain, 1 -3.21 4.98E-32 Sgk2 serum/glucocorticoid regulated kinase 2 -3.19 1.08E-31   135   Gene symbol  Gene name  Fold change  Adjusted P-value Kbtbd11 kelch repeat and BTB (POZ) domain containing 11 -3.18 1.54E-31 Abhd3 abhydrolase domain containing 3 -3.15 3.90E-31 Hist1h2bn histone cluster 1, H2bn -3.15 4.36E-31 Ptgds prostaglandin D2 synthase 21 -3.14 7.07E-31 Gsta4 glutathione S-transferase alpha 4 -3.02 6.94E-29 Nrip3 nuclear receptor interacting protein 3 -2.96 5.73E-28 Gm10639  -2.95 6.75E-28 Hhatl hedgehog acyltransferase-like -2.91 3.53E-27 Ighg2c immunoglobulin heavy constant gamma 2C -2.9 5.51E-27 Megf6 multiple EGF-like-domains 6 -2.88 1.19E-26 Cabp7 calcium binding protein 7 -2.86 2.32E-26 Thrsp thyroid hormone responsive -2.85 3.05E-26 Scml4 sex comb on midleg-like 4 -2.83 7.39E-26 Bglap2 Osteocalcin-2 precursor -2.83 9.01E-26 Rlbp1 retinaldehyde binding protein 1 -2.81 1.60E-25 Oc90 otoconin 90 -2.81 1.89E-25 Ttbk1 tau tubulin kinase 1 -2.8 2.77E-25 Fxyd1 FXYD domain containing ion transport regulator 1 -2.77 6.08E-25 Asb10 ankyrin repeat and SOCS box containing 10 -2.74 2.32E-24 3110079O15Rik  -2.71 7.33E-24 Gm6821  -2.7 9.44E-24 Hsd17b11 hydroxysteroid (17-beta) dehydrogenase 11 -2.69 1.26E-23 Irf4 interferon regulatory factor 4 -2.69 1.54E-23 Gm21956  -2.67 3.58E-23 Rasl10b RAS-like, family 10, member B -2.66 4.16E-23 Arhgef37 Rho guanine nucleotide exchange factor (GEF) 37 -2.66 4.85E-23 Spa17 sperm autoantigenic protein 17 -2.65 7.87E-23 Azgp1 alpha-2-glycoprotein 1, zinc-binding -2.63 1.63E-22   136   Gene symbol  Gene name  Fold change  Adjusted P-value Gm11175  -2.62 2.14E-22 Col9a3 collagen, type IX, alpha 3 -2.61 2.82E-22 Spp1 secreted phosphoprotein 1 -2.6 4.68E-22 Adamtsl4 ADAMTS-like 4 -2.59 7.25E-22 Fam83f family with sequence similarity 83, member F -2.59 7.37E-22 Trpm1 transient receptor potential cation channel, subfamily M, member 1 -2.59 7.52E-22 Ccdc74a coiled-coil domain containing 74A -2.58 8.92E-22 Paqr6 progestin and adipoQ receptor family member VI -2.58 9.23E-22 Wnk2 WNK lysine deficient protein kinase 2 -2.57 1.21E-21 Hist1h2bm histone cluster 1, H2bm -2.55 2.67E-21 Sox9 SRY (sex determining region Y)-box 9 -2.55 2.99E-21 Cer1 cerberus 1, cysteine knot superfamily, homolog -2.54 4.79E-21       137  Table 23: Top 10% downregulated genes from no stretch to stretch  Gene symbol  Gene name  Fold change  Adjusted P-value Nupr1 nuclear protein, transcriptional regulator, 1 -3.52 <1.00E-35 Kcnj13 potassium inwardly-rectifying channel, subfamily J, member 13 -2.69 1.128E-30 G730046D07Rik  -2.45 1.63E-25 Slc7a11 solute carrier family 7 (anionic amino acid transporter light chain, xc- system), member 11 -2.43 4.423E-25 Tmem217 transmembrane protein 217 -2.39 2.833E-24 Gm13199  -2.36 1.156E-23 Chac1 ChaC, cation transport regulator homolog 1 -2.11 3.266E-18 Herpud1 homocysteine-inducible, endoplasmic reticulum stress-inducible, ubiquitin-like domain member 1 -2.07 2.431E-17 Acot10 Acyl-coenzyme A thioesterase 10, mitochondrial -2.06 4.144E-17 Bricd5 BRICHOS domain containing 5 -2.01 4.908E-16 Apobec1 apolipoprotein B mRNA editing enzyme, catalytic polypeptide 1 -2 6.443E-16 Gm17655  -2 6.66E-16 Atf5 activating transcription factor 5 -1.96 3.682E-15 Gm5039  -1.95 6.519E-15 Eef1a2 eukaryotic translation elongation factor 1 alpha 2 -1.94 1.36E-14 Steap1 six transmembrane epithelial antigen of the prostate 1 -1.93 1.484E-14 Cck cholecystokinin -1.9 5.993E-14 ENSMUSG00000095326  -1.9 7.882E-14 Nupr1 nuclear protein, transcriptional regulator, 1 -3.52 <1.00E-35      138  4.2 The majority of the top 10% genes regulated by FN density show a dose-dependent regulation pattern To further characterize the gene expression pattern for the top 10% of FN density regulated genes from low to medium to high FN, I calculated the ratio of the fold changes from high/low divided by the fold changes from medium/low FN density (the top 10% genes each, see Supplementary Table 24 to Supplementary Table 28, and Figure 23). The majority of the top 10% regulated genes (77.8%), including the signature genes, were consistently up- or downregulated from low to medium to high FN density (Figure 23, green). A smaller proportion (18.8%) was more strongly regulated from low to medium FN density than from low to high FN density, but were consistent in the direction of regulation (either up or down, Figure 23, blue). Three genes (2%) were regulated to the same level between low to medium and low to high FN density (Figure 23, white), and another 2 genes were regulated in opposite directions (Figure 23, purple and yellow). In summary, there was a high degree of overlap between genes that were regulated in one manner between low and medium FN and those that were regulated in the same manner between low and high FN density.    139   Figure 23: Comparison of gene expression changes between B16F1 on low to medium and low to high FN density. Ratio of high/low FN density fold changes over medium/low FN density fold changes from top 10% of both upregulated genes and downregulated genes. A Ratio >1 represents dose-dependent up- or downregulation from low to medium to high FN density (green). All three signature genes can be found in this category (green, red and blue dots). A ratio of 1 indicates the same fold change between low to medium and low to high FN density. A ratio > 0 and > 1 indicates a greater fold change from low to medium than low to high FN density in the same direction (blue). A ratio of 1 indicates the same fold change from low to medium as from low to high FN density. A ratio < 0 and > -1 indicates a stronger fold change from low to medium FN density than a fold change from low to medium in the opposite direction (purple). A ratio of > -1 indicates a smaller fold change from low to medium FN density than the fold change from low to high in the opposite direction (yellow). Sample expression patterns for each category can be found color coded on the right. Genes that belong to each group A-E are listed in Supplementary Table 24 to Supplementary Table 28.    140  4.3 FN density-regulated and stretch-regulated genes have limited overlap  Because increasing FN density and applying stretch both increase intracellular load and affect integrin adhesome signalling, I expected that these two perturbations would regulate an overlapping set of genes. To assess this, I determined the sets of genes that were regulated only by FN density alone (either low to medium or low to high), stretch alone, or regulated in the same direction by both approaches (Figure 24). Changes in FN density resulted in the downregulation of 680 genes whereas stretch downregulated 180 genes. Of these genes, 636 genes were downregulated only by FN, 129 were downregulated only by stretch, and 51 were downregulated by both. Similarly, of the 369 FN density-upregulated genes, 66 genes were also upregulated by stretch, identifying 300 genes that were upregulated only by stretch. Only a few genes were regulated in opposing ways by FN density and stretch. These will be considered separately in the future. Interestingly, only 117 genes (~7.5% of the total expression changes) were regulated both by increasing the FN density and applying stretch. Thus, ligand density and tensile stress may regulate different pathways that control the mRNA levels of distinct sets of genes.   141   Figure 24: FN density and stretch regulate distinct sets of genes. Numbers of genes that were upregulated or downregulated under the two experimental conditions. Genes regulated by ligand density are in green, genes regulated by stretch are in pink. The overlapping areas represent genes that are regulated by both perturbations.  4.4 Increasing FN density and applying stretch regulate genes that cluster in different pathways Given that only ~7.5% of the gene expression changes were regulated by both increased ligand density and applied stretch, I asked whether the gene expression changes were associated with distinct or overlapping functions. A pathway analysis revealed that the downregulated genes did not result in highly significant enrichment of pathways or gene ontology (GO) biological   142  processes. Therefore I concentrated my analysis on the upregulated genes. When I grouped the regulated genes by pathways and GO functions, most genes that were upregulated by FN density were annotated as either adhesion and motility genes or ECM regulators. In contrast, the genes significantly upregulated by stretch belonged primarily to the Hypoxia-Inducible Factor (HIF)-1 regulatory pathway or were associated with metabolic processes such as glycolysis and gluconeogenesis (Figure 25, Supplementary Table 29, and Supplementary Table 30). This suggests that cellular responses to FN density and stretch are distinct from each other.    143   Figure 25: FN density and mechanical stretch regulate distinct pathways. Clusters of FN genes upregulated by density (green) and stretch (pink) were grouped according to their pathways and GO process annotations. Analysis was carried out using the software suite GeneAnalytics©. Scores are –log2(p-value) of the corrected p-value using false discovery rate. Scores > 4.3 correspond to p < 0.05, and are considered significant. Scores > 13.3 correspond to p < 0.0001, scores > 16.6 correspond to p < 0.00001, scores > 19 correspond to p < 0.000001.     144  4.5 FN density and stretch regulate distinct pathways and distinct members of shared pathways Both increasing FN density and mechanical stretch resulted in the upregulation of genes that belong to signalling pathways that are associated with mechanotransduction and integrin signalling. Figure 26 depicts the genes of selected pathways that were regulated by FN density or mechanical stretch alone, or by both perturbations. Some pathways were almost exclusively upregulated by one of the two experimental modulations. For example, only increasing FN density upregulated gene clusters associated with the Hippo pathway. Conversely, only mechanical stretch upregulated gene clusters associated with HIF-1 signalling. Furthermore, comparing pathways that were regulated by both FN density and stretch (e.g. degradation of ECM, ERK signalling, FA) revealed that the subsets of genes regulated by the two approaches were overlapping but not identical. For example, of the 72 genes of the ERK pathway that were upregulated by either FN density or stretch, 12 genes were shared, 33 genes were upregulated only by increasing FN density, and 27 were upregulated only by mechanical stretch. Because only a small fraction of genes within a given pathway were regulated by both FN density and mechanical stretch, it suggests that these two stimuli activate overlapping but non-redundant signalling pathways that result in the activation of different sets of TFs. This may in turn lead to different cellular responses.          145   Figure 26: Pathways or GO biological processes with different sets of genes regulated by FN density, stretch or both stimuli.  Gene symbols in green are only upregulated by FN density, the ones in pink are upregulated only by stretch, and the symbols in grey are upregulated by both FN density and stretch.    4.6 FN density and mechanical stretch may activate distinct sets of TFs To assess whether the distinct gene clusters upregulated by FN density versus stretch reflected the activation of distinct TFs, I used the ToppCluster594 software to search for enriched TF binding sites (TFBS). Although some enriched TFBSs were shared, the gene sets whose expression was preferentially regulated by FN were enriched in different TFBSs than the gene   146  sets whose expression was increased upon mechanical stretch. Genes that were regulated by both FN density and stretch were enriched in the TFBSs for members of the activator protein 1 (AP-1), activating TF (ATF), and SRF TF families (Figure 27, grey). In contrast, genes that were upregulated only in response to increased FN density were enriched for binding sites for members of the BACH, Zic family member (ZIC), Growth factor independent 1 transcription repressors (GFI), Nuclear factor of activated T-cells (NFAT), and estrogen receptor (ER) TF family. TFBSs that were enriched in genes upregulated only by stretch included members of the GATA, Nuclear factor (INF1), HIF-1, and pregnane X receptor (PXR) TFs. This suggests that increasing FN density and mechanical stretch activate overlapping as well as distinct TF networks, which may explain the distinct patterns of gene changes induced by these two microenvironmental changes.    147   Figure 27: Increasing FN density and mechanical stretch upregulate gene clusters that are enriched for distinct TFBS motifs.  Putative TF binding sites to enriched in genes (human orthologues) that are upregulated by either FN density (gene set indicated in green) or mechanical stretch (gene set indicated in pink). TFBSs that are enriched in both gene sets are indicated in grey. The analysis is based on conserved cis-elements within the genes as calculated by ToppCluster594. Selected TF families are represented by different colors.   Which TFs controls the expression of the signature genes in the two experimental set ups, increasing FN density and stretch, needs to be further explored. Cyr61 transcription is regulated by members of the CREB, Sp-1, Egr-1 and AP1 families652, 703, and contains binding sites for ATF, E2F, HNF3b, NF1, NFкB, and SRF TFs in its promoter region713. In melanoma cells   148  transcription of MUC18 is suppressed by the TFs AP2714, ZBTB7A705, and MITF715, and promoted by members of the CREB/ATF1 family716. Recently, there is evidence that in hepatocellular carcinoma, CREB regulates the transcription of MUC18 in a YAP-dependent manner717, which links the regulation of MUC18 expression to the Hippo pathway. In addition, CREB TF family members also have been identified as positive regulator for MITF718, which is the primary regulator of TRPM1 transcription706. Interestingly, although CREB could theoretically affect all three signature genes, TFBSs for CREB were not enriched in either the FN density- or stretch-regulated genes (Figure 27). This is not surprising as gene transcription is often regulated by a network of TFs, enhancers and repressors and varies between cell types. However, TFBSs for AP1, ATF and SRF which are TFs that also drive Cyr61 transcription were enriched in both gene sets. To identify the TFs that regulate the transcription of the signature genes in response to stretch or FN density, next generation sequencing, e.g. ChIP-sequencing (chromatin immunoprecipitation with parallel DNA sequencing to identify protein binding sites of DNA-associated protein) could be employed.    4.7 Summary and perspectives  Summary of findings The microenvironment of a cancer cell plays an important role during cancer progression250. Abnormal ECM microenvironments often include increased density integrin ligands, such as FN or collagen as well as increased ECM stiffness, which leads to increased cellular tension. In this chapter I addressed whether increased FN density and increased tension   149  regulate different transcriptional programs that are related to cancer progression. To test this I compared the transcriptomes of B161 cells that had spread on increasing FN densities with the transcriptome of B16F1 cells that were subjected to mechanical stretching. This analysis revealed only a small overlap between the genes that were regulated by FN density and by mechanical stretch. Pathway enrichment analysis showed that the genes that were upregulated by increasing FN density belonged primarily to pathways involved in ECM remodelling, cell adhesion and migration, whereas the genes that were upregulated upon mechanical stretch were enriched for pathways related to metabolism and the HIF pathway (Figure 28). This suggests that increased integrin ligand density and increased ECM stiffness result in the activation of different transcription regulators. Moreover, these findings are consistent with the finding that the three cancer signature genes are strongly regulated by increased integrin ligand density and less sensitive to modulation of cytoskeletal tension (see chapter 3).   Figure 28: Transcriptome analysis of B16F1 cells plated on increasing FN density or subjected to mechanical stretch. The analysis showed that these modulations affect mRNA levels of genes associated with distinct pathways.    150  There were also several pathways that were affected by both FN density and stretch. However, a comparison of the upregulated pathways revealed that each perturbation regulated different sets of genes within the same pathway. For example, the pathway ECM organization was upregulated by both FN density and stretch, but distinct types of proteins were regulated. The genes upregulated by increased FN density were primarily metalloproteinases such as: members of the disintegrin and metalloproteinase with a thrombospondin type 1 motif family (ADAMTS1-5); the metalloproteinases MMP1 and Tolloid-Like 1 (TLL1); the metalloproteinase regulator TIMP1 ECM proteins, including multiple forms of collagen (COL1A1, COL7A1, COL8A1); enzymes involved in collagen synthesis such as precollagen (PCOLCE); Hyaluronan And Proteoglycan Link Protein 1 (HAPLN); and the matricellular proteins osteonectin (secreted protein acidic and rich in cysteine (SPARC)) and thrombospondin-1 (THBS1). In contrast, the genes in this ECM pathway that were upregulated by mechanical stretch were primarily cell adhesion molecules, including members of the integrin family (ITGA8, ITGB8, ITGB2), the adhesion proteins NCAM-1 and ICAM-2, and Syndecan 4 (SDC4). Interestingly, both FN density and stretch increased the mRNA levels of FN (FN1). Nevertheless, increasing FN density and mechanical stretch primarily regulated the expression of different pathways and different genes within the same pathways, some of which I will discuss in the following sections.  FN density and mechanical stretch both regulate distinct sets of ECM genes The upregulation of collagens in B16F1 cells that were plated on increasing FN density may reflect the fact that cell spreading can induce the expression of collagen. For example, cell spreading in 2D has previously been associated with type I collagen synthesis719. Increased collagen, together with FN density-dependent increase of MMPs and ADAMs, suggests that   151  increasing FN density leads to ECM remodelling. Remodelling of collagen into more straightened and parallel fibers is often seen near tumors and can contribute to tumor progression276-278. Whether the observed increase in collagen mRNAs in B16F1 cells plated on high FN density leads to increased collagen secretion or to the remodelling of collagen fibers needs to be determined.   Interestingly, recent findings by Zhang et al.720 suggest that exposing cells to FN results in increased MMP expression, and that FN alone might activate fibroblasts to remodel or repair ECM. Furthermore, they found that MMP gene expression and activity was less sensitive to cyclic stretch than to increased FN density. Although their experiments differed from those presented here in terms of FN form (soluble versus surface-bound), type of stretch (2D cyclic versus 3D static), and cell type (primary fibroblasts versus B16F1 cells), some parallels can be drawn. Similar to their results, I found that in B16F1 cells the mRNA levels of several MMPs were upregulated to a greater extent by increasing FN density than by cellular stretch in B16F1 cells. In contrast, both FN and stretch increased FN mRNA levels. Therefore my results, together with the findings of Zhang et al.720, suggest that ECM density and composition regulate the expression of ECM components as well as the MMPs that remodel the ECM, and that this occurs independently of mechanical stretch. Further studies using other ECM matrices or tumor-derived ECMs could validate how FN density-dependent gene expression changes compare to changes caused by other tumor-associated ECM modifications such as increased collagen composition and crosslinking, which are associated with more aggressive tumors275.       152  FN density regulates the Hippo pathway I also found that increasing FN density upregulated several genes of the Hippo pathway, whereas applying mechanical stretch did not. The Hippo pathway regulates the activity of the TF YAP/TAZ. Previous work has shown that the Hippo pathway is regulated by cell adhesion to FN721, and that the resulting increase in YAP/TAZ activity has tumor-promoting activity in several mouse models722. In B16F1 cells, increasing the FN density resulted in upregulation of the genes that encode for both negative (LATS2, RASSF1, RASSF6) and positive (AJUBA, TEAD1, TEAD4) regulators of TAZ activity. However, the finding that five direct target genes of TAZ (GTGF, FGF-1, Cyr61, Myc, CCND1) were upregulated in response to increasing FN density suggests that increased integrin ligand density promotes TAZ-dependent transcription. Consistent with this, Kim et al.721 found that adhesion to FN induces nuclear translocation and accumulation of YAP in MCF-10A breast cancer cells. Furthermore, it has been proposed that F-actin stress fibers, which would form in cells adhering to high density FN, are critical for the nuclear translocation of YAP/TAZ723. Together, this suggests that adhesion to ECM activates the Hippo pathway and allows for TAZ-dependent gene expression.  Interestingly, in B16F1 cells, mechanical stretch did not result in the enrichment of genes in the Hippo pathway, although previous research had shown that YAP/TAZ enters the nucleus upon cell stretching724. However, those findings were obtained using cells that were stretched while forming a confluent monolayer in which YAP/TAZ is normally sequestered at cell-cell junctions. Because I applied stretch to a sparse population of B16F1 cells, and B16F1 cells do not form functional cell-cell junctions, this may account for the different results. Nevertheless, my results suggest that increasing FN density enhances YAP/TAZ-dependent transcription and may therefore mediate ECM-density effects on cancer cells.   153  HIF-1 pathway  When B16F1 cells were subjected to stretch, the pathways that were predominantly upregulated were the HIF-1 pathway and various metabolic pathways. This transcriptional response is similar to hypoxia signalling and is in accordance with other studies that found that the HIF-1 is upregulated by stretch725-730. Although HIF-1 is a TF that is critical for tissues to adapt to hypoxia731-733, mechanical stretch-dependent functions are independent of hypoxia and indicate a functional role for HIF-1 under normoxic conditions734. HIF-1 has been implicated in regulating many genes involved in metabolic pathways and it is an important regulator of glycolysis734, 735. In line with this, transcript levels for glycolytic enzymes and enzymes of the citric acid cycle such as pyruvate dehydrogenase kinase 1 (PDK1) and lactate dehydrogenase A (LDHA) were increased when stretch was applied to B16F1 cells. This suggests that in B16F1 cells static stretch induces a HIF-1-dependent metabolic switch that is associated with increased glycolytic capacity under normoxic conditions. It is possible that the 3D collagen I/FN matrix is a more hypoxic environment than when cells are cultured on a 2D substrate and that this primes cells for a hypoxic response under stretch. Whether the 3D matrix in this particular experimental set-up is a hypoxic environment, or whether stretching B16F1 cells results in HIF-1-dependent increases in glycolysis, could be determined by comparing a direct readout for glycolysis (e.g. lactate levels, acidification of media) in B16F1 control cells cultured in 2D versus 3D in the presence of absence of inhibitors of HIF-1 activity.  Consistent with HIF-1 activation promoting normal cell survival under stress condition, its activation in tumors also promotes cell survival and is therefore associated with poor prognosis. Besides reprograming energy metabolism736, HIF-1 is a master transcriptional activator for critical genes that support most, if not all, steps of tumorigenesis and tumor   154  progression including angiogenesis737, 738, proliferation and apoptosis739, invasion and metastasis738, 740, and establishing metastatic niches740, 741. In addition, HIF-1 mediates resistance to radiation and chemotherapy742 743. Several HIF-1 inhibitors, as well as inhibitors that target other components of the HIF-1 pathway, are currently in clinical trials as cancer therapeutics and several have been approved for use in combination therapy with other chemotherapeutics744.  Furthermore some genes that are associated with the HIF-1 pathway, and which are responsive to stretch (e.g. fructose-Bisphosphate Aldolase A (ALDOA)745, carbonic anhydrase 9 (CA9)746, 747 and hexokinase domain containing 1 (HKDC1)) have been implicated in cancer and could potentially be regulated by increasing ECM stiffness. It would be interesting to explore whether targeting HIF-1 directly or targeting subsets of HIF-1-regulated genes could result in a better outcome for cancer patients than current treatments.  Because the HIF-1 pathway is affected by mechanical loads in vitro, it would be interesting to address whether HIF-1 inhibitors could be beneficial for the treatment of early cancer stages that manifest as premalignant lesions. Reducing HIF-1-dependent effects that are caused by increased mechanical loads caused by ECM rigidity could reduce tumor invasiveness and metastasis. Similarly, targeting HIF-1 after removal of a primary tumor may be beneficial for preventing reoccurrence and metastasis.   Future perspectives  Although my analysis revealed that increasing ECM ligand density and stretch regulated distinct pathways, further analysis is likely to reveal additional differences. For example, differences in mRNA splicing would be revealed by exome analysis, which I am currently carrying out. I have also obtained RNA-seq data for long non-coding RNAs (lncRNAs).   155  LncRNA, which include long or large intergenic nuclear-retained RNAs, enhancer RNAs, antisense RNAs and pseudogenes, have been identified as drivers of tumor suppressive and oncogenic functions in certain cancer types such as breast and prostate cancer. Their functions and mechanisms of action are currently being studied748. Analyzing mRNA isoforms and identifying changes in lncRNA levels that occur in response to increasing ligand density or stretch may yield further insight into how changes in the tumor microenvironment impact tumor cells and promote tumor progression. Therefore, this could lead to the discovery of new biomarkers and RNA-based targets for future therapies.     156  Chapter 5: The adhesome proteins talin and FAK regulate expression of the cancer signature genes  5.1 Introduction In chapter 3, I showed that the regulation of the cancer signature gene mRNA levels in B16F1 cells is strongly regulated by integrin-dependent adhesions and in chapter 4 I showed that integrin-dependent adhesions regulate transcriptome-wide gene expression changes. In this chapter, I address how two members of the integrin adhesome, talin and FAK regulate expression of the signature genes. Integrin adhesions link the ECM to the cytoskeleton and initiate signalling through an elaborate network of proteins that are collectively known as the ‘integrin adhesome’. The scaffolding role of the adhesome supports the physical integration of the ECM-bound cell and the cytoskeleton, whereas the signalling role of the adhesome activates downstream targets such as kinases, phosphatases, and G protein regulators. Together, the scaffolding and signalling functions regulate the response of the cell to the ECM and determine cell structure, behaviour and fate208, 383 (section 1.5). Recently, several diseases, including cancer, vascular and musculoskeletal diseases, have been associated with defects in adhesome function that are either caused by mutations or dysregulated activity of adhesome proteins (e.g. FAK, integrins, PI3K). However, how adhesome function relates to disease, and drives cancer progression specifically, is not fully understood. To address this, I investigated the role of two adhesome proteins, talin and FAK, in regulating the cancer signature genes Cyr61, MUC18 and TRPM1 in B16F1 cells. Talin and FAK are two key players in adhesion assembly and turnover749. Talin is a scaffolding protein that binds to integrins, FAK and actin, and activates integrins750, 751. FAK is a scaffolding protein and   157  tyrosine kinase that phosphorylates substrates such as paxillin and regulates adhesion dynamics. Both proteins have been implicated in cancer progression511, 512, 515, 517, 519, 577, 752-754, and several FAK inhibitors are in clinical trials for cancer treatment755. However how these proteins contribute to cancer progression is not entirely understood. To test whether talin- and FAK-dependent signalling is important for regulating the cancer signature genes in B16F1 cells, I used types of loss-of-function approaches. I reduced the expression of talin (talin 1 and talin 2) or FAK using siRNAs. With this tool I discovered that talin and FAK differentially regulate the mRNA levels of the three signature genes.  5.2 Talin is a regulator of MUC18 and TRPM1 gene expression  5.2.1 Talin is important for cell spreading in B16F1 cells To test whether talin is important for regulating the expression levels of the three signature genes, I used siRNAs (one single siRNA per targeted gene) targeting of both talin 1 and talin 2 to transiently knock down both talin isoforms. This resulted in the efficient and reproducible reduction in talin protein levels (Figure 29A). I found that talin knockdown (KD) B16F1 cells spread less extensively when plated on FN overnight compared to control siRNA-transfected cells (Figure 29B). This difference in cell spreading was determined by comparing the area size of F-actin at the nearest plane to the coverslip between cells from both populations.  Consistent with other data450, talin KD cells also had significantly lower levels of FAK-pTyr397 than control cells, suggesting reduced levels of FAK kinase activity (Figure 29C) in cells cultured on TC plastic. Preliminary data also suggests that talin KD cells have lower levels of paxillin-Tyr118 (Supplementary figure 1). Thus knocking down talin alters cell spreading and   158  adhesion signalling in B16F1 cells, processes that are likely involved in regulation of the cancer signature genes.   Figure 29: Knocking down talin reduces cell spreading and FAK phosphorylation. (A) B16F1 cells were transfected with control siRNA or siRNAs specific for talin 1 and talin 2 and cultured on TC plastic. After 48 h, talin 1 and talin 2 levels in cell lysates were evaluated by immunoblotting with an antibody that recognizes both isoforms. (B) Cell spreading area of control and talin KD cells that were plated on FN (2 µg/cm2) overnight. Asterisks indicate significant differences (Median with interquartile range, >45 cells per experimental group per experiment, Mann-Whitney test, ****, p < 0.0001). (C) Immunoblotting for FAK-pTyr397 and FAK in lysates from cells cultured as in A. Dashed lines indicate cropped blots for representation purpose. (D) Quantification of immunoblot of FAK-pTyr397 normalized to total FAK under   159  condition as in A. Mean and range from 2 independent repeats. Panel A is representative of 5 independent experiments. Panel B is representative of 3 independent experiments. Panel C is representative of 2 independent experiments. Molecular weight markers in kDa are indicated.    5.2.2 Talin regulates MUC18 and TRPM1 expression in B16F1 cells   To assess whether talin regulates expression of the cancer signature genes, I compared mRNA levels of the signature genes from control cells and talin KD cells that were cultured for 24 h on TC plastic. The experiments to assess mRNA levels were carried out on TC plastic without additional FN. However, serum-containing medium contains FN. Additional coating of the TC plastic with FN did not alter mRNA levels of the signature genes (Supplementary figure 2). Consistent with the regulation of mRNA levels in response to adhesion and integrin ligand density, MUC18 mRNA levels were ~4-fold lower and TRPM1 mRNA levels were ~9-fold higher in talin KD cells than in control cells. In contrast, Cyr61 mRNA levels did not change (Figure 30A). When protein levels for MUC18 were compared between control and KD cells, expression levels were reduced in KD cells (Figure 30B). This suggested that talin plays a key role in regulating the expression of the signature genes, MUC18 and TRPM1.     160   Figure 30: Talin expression regulates MUC18 and TRPM1 expression. (A) Fold change of signature genes in talin KD B16F1 cells compared to control siRNA cells that were cultured for 24h on TC plastic. mRNA levels were determined by qRT-PCR. Mean and SD for 7 independent experiments. Asterisks indicate significant differences (Student’s t-test, **, p < 0.01; ****, p < 0.0001; ns, not significant). (B) Immunoblots of MUC18 in total cell lysates of control cells and talin KD cells. 1 experiment. Molecular weight markers in kDa are indicated.   5.2.3 Talin 1 is important for the regulation of MUC18 and TRPM1  Next, I tested the hypothesis that talin 1 and talin 2 have distinct roles in regulating the expression of MUC18 and TRPM1 by individually knocking down either talin 1 or talin 2 (Figure 31A) using specific siRNAs. Probing cell lysates with an antibody that recognizes both talin 1 and talin 2 revealed a dramatic reduction of talin (barely detectable band) in cells that were transfected with siRNAs against both talin 1 and talin 2 (Figure 31A, lane 2). In contrast, when cells were transfected with either talin 1 or talin 2 specific siRNAs, a residual band of the non-silenced talin form was visible (Figure 31A, lane 3 and 4). This showed that B16F1 cells express both talin 1 and talin 2 and that silencing one form did not suppress the expression of the other.    161  To determine whether individually knocking down either talin 1 or talin 2 regulated the mRNA levels of MUC18 or TRPM1, I compared their mRNA levels between control, double KD, and single KD B16F1 cells on TC plastic. In comparison to control cells and double KD cells, knocking down talin 1 was sufficient to reduce MUC18 expression to the same extent (~4-fold) as the double KD (~4-fold). In contrast knocking down talin 2 did not affect MUC18 gene expression. Knocking down talin 2 increased TRPM1 mRNA levels by 2-fold, whereas knocking down talin 1 caused a ~5-fold increase of TRPM1, approximately half the increase seen when both talins were knocked down (~9-fold) (Figure 31B). This suggests that talin 1 and talin 2 may have additive effects in regulating TRPM1 mRNA levels. Taken together, it appears that talin 1 is more important than talin 2 for regulating MUC18 and TRPM1 expression.   Figure 31: Talin 1 is a stronger regulator of MUC18 and TRPM1 expression than talin 2. (A) B16F1 cells were transfected with control siRNA or siRNAs specific for talin 1, talin 2, or both. After 48 h on TC plastic, talin 1 and talin 2 levels in cell lysates were evaluated by immunoblotting. (B) Fold change of signature genes in talin KD cells compared to control siRNA cells cultured for 24 h on TC plastic. mRNA levels were determined by qRT-PCR. Mean and SD for 5 independent experiments. Asterisks indicate significant differences (Student’s t-test, *, p < 0.05; **, p < 0.01; ns, not significant). Molecular weight markers in kDa are indicated.   162  5.2.4 Talin 1 and talin 2 both regulate B16F1 cell spreading and cell motility  Given the differences between talin 1 and talin 2 in their ability to regulate the signature genes, I next addressed if talin 1 and talin 2 also have different effects on cell spreading and cell motility in B16F1 cells. When allowed to spread overnight on FN-coated coverslips, talin 1 KD cells exhibited a slight reduction in cell spreading area, compared to control cells, but this effect was not as dramatic as in double KD cells (Figure 32A). In contrast, knocking down talin 2 resulted in increased cell spreading area. This suggests that talin 2 may be a negative regulator of cell spreading and that talin 1 and talin 2 coordinately regulate cell spreading in B16F1 cells.  To assess how knocking down talin 1 or talin 2 or both talins affected cell migration, bead clearing assays were performed. Cells were plated on FN-coated coverslips that also had been coated with fluorescent microbeads415, and then allowed to migrate overnight. During cell movement the cells clear a path, leaving a negative imprint of their migration paths, which provides a visual indicator of cell motility (Figure 32B). Compared to control cells, talin 1 KD cells, talin 2 KD cells, and double talin KD cells all cleared the microbeads from a smaller area (Figure 32B). This suggests that both talin 1 and talin 2 are important for cell migration. However, even when both talins were knocked down, cells were still able to migrate. In fact, the double KD cells cleared a slightly larger area than either talin 1 or talin 2 single KD cells (Figure 32C). This finding is reminiscent of cells undergoing amoeboid motility, which, in contrast to the mesenchymal motility, relies less on integrin-mediated adhesions and less on talin329. Taken together, these data suggest that talin 1 and talin 2 have non-redundant roles in cell spreading, whereas both contribute to adhesion-dependent migration.    163   Figure 32: Talin 1 and talin 2 regulate cell spreading and motility.  (A) Cell spreading area for control and talin KD cells that were plated on FN overnight. (B) Cells were plated on FN-coated (2 µg/cm2) coverslips that had a thin layer of fluorescent microbeads on their surface. During cell movement cells clear a path leaving a negative imprint of their migration paths. (C) Sample paths for each experimental group are shown (right). Size bar 100 µm. Fluorescent microbeads are blue. Representative B16F1 cells for each group are traced in red (based on F-actin staining). Green arrow indicates area cleared by migrating cell, red arrow   164  indicates cell. Asterisks indicate significant differences (Median with interquartile range, >50 cells for A, >28 cells for B, Mann-Whitney test, *, p < 0.05; ***, p < 0.001; ****, p < 0.0001). The data are representative of 2 independent experiments for A and 3 for B.   5.3 FAK regulates expression of the cancer signature genes Overexpression, hyperphosphorylation, and elevated activity of FAK have been reported in a variety of cancers547, and in melanoma these alterations promote an aggressive phenotype756. Similarly, high levels of FAK phosphorylation on Tyr397 and Tyr576 are found in late-stage cutaneous and uveal melanoma, which correlates with robust cell migration and high invasion capacity569. Furthermore, B16F10 cells, which have greater lung-colonizing potential than B16F1cells, have higher levels of FAK and FAK-pTyr than B16F1 cells757. As described in section 1.10.11, FAK regulates the migration and invasion of B16F10 cells566. However, how FAK promotes cancer progression is complex and not fully understood. Because an increase in tumor aggressiveness likely involves not only increased cell migration and invasion but also transcriptional changes that promote other cancer hallmarks, I tested the hypothesis that FAK regulates the mRNA levels of the signature genes.  My findings that the cancer signature genes Cyr61, MUC18 and TRPM1 are regulated by increasing ECM density and cell adhesion, and that MUC18 and TRPM1 expression is dependent on the FAK-interacting protein, talin, suggest that FAK would be important for the regulation of these signature genes. FAK has both kinase-independent scaffolding functions and kinase activity. Using siRNA, I reduced FAK protein level expression, which reduces both kinase-independent and kinase-dependent function of FAK. I first addressed how knocking down FAK influenced B16F1 cell morphology and adhesion organization, and then assessed the effects on the mRNA levels of the cancer signature genes.    165  5.3.1 Knocking down FAK increases cell spreading To reduce both kinase activity and scaffolding function, I used a single siRNA directed against FAK, which efficiently reduced the expression level of FAK such that the protein was undetectable by immunoblotting (Figure 33A). This resulted in increased cell spreading on FN compared to cells transfected with control siRNA (Figure 33B). Knocking down FAK resulted in spreading areas that were 2-fold larger than that for control cells. Thus, FAK is important for regulating cell spreading in B16F1 cells.   Figure 33: Knocking down FAK increases B16F1 cell spreading. (A) B16F1 cells were transfected with control siRNA or siRNA specific for FAK. After 48 h, FAK expression in cell lysates was evaluated by immunoblotting. (B) Cell spreading area for control and FAK KD cells that had been plated on FN (2 µg/cm2) overnight. Median with interquartile range, > 50 cells, Mann-Whitney test, ****, p < 0.0001). Representative of 3 independent experiments with > 50 cells assessed per experiment.    166  5.3.2 Knocking down FAK causes distinct changes in cell morphology changes including loss of cell polarity  Knocking down FAK expression not only caused an increase in cell spreading, but also altered the morphology of the cells (Figure 34). Compared to control cells, which polarize with recognizable front-rear organization (Figure 34B, direction of arrow indicates rear to front), knocking down FAK resulted in a ‘dish-like’ morphology with short cell protrusions (Figure 34C) when plated on FN overnight, consistent with previous finding by Owen et al.758.  In addition, knocking down FAK altered the localization of β1 integrin and vinculin and disrupted the organization of F-actin (Figure 34B). In control cells β1 integrin was present in vinculin-containing adhesion sites and accumulated at the rear of the cell. In contrast, β1 integrin did not exhibit a polarized distribution in FAK KD cells. Instead, small β1 integrin clusters were uniformly distributed within the cell, with slightly greater staining at the center of the cell (Figure 34B).  Actin organization was also disrupted in FAK KD cells. In control cells, F-actin fibers formed along the anterior-posterior axis of the cell and an F-actin rich web was present at the leading edge, which is characteristic for cells undergoing mesenchymal migration. The orientation of F-actin fibers in FAK KD cells was distinctly different from those in control cells. The F-actin fibers lacked front-rear polarization and formed along the periphery of the cells, similarly to the cortical actin filaments seen by Sieg et al.759 in FAK-null fibroblasts (Figure 34B, arrowhead).  In control cells, vinculin clusters were present at the tips of F-actin fibers, presumably serving as a link between F-actin and other adhesome proteins such as talin217, 760. These vinculin clusters were present at the front and rear of the cell as well as underneath the body the cell. In   167  FAK KD cells, vinculin still localized at the tips of F-actin fibers which were now only at the periphery of the cells. Taken together, these results suggest that FAK regulates cell morphology, cell polarity and adhesome formation in B16F1 cells.    Figure 34: Loss of cell polarity and distinct changes in cytoskeletal organization are caused by FAK KD.   (A) Representative cell shapes of control and FAK KD B16F1 cells. (B) Immunolocalization of β1 integrin, vinculin (green) and F-actin (red) in either control or FAK KD B16F1 cells on FN (2 µg/cm2) overnight. Arrows indicate the direction of rear-to-front cell polarity in polarized cells. Arrowheads point to actin at the cell periphery. Size bar, 15 µm.    168  5.3.3 Knocking down FAK regulates signature genes Because knocking down FAK disrupted cell polarity, adhesome localization, and the organization of the F-actin cytoskeleton I asked whether this would also affect the expression of the signature genes. Knocking down FAK in B16F1 cells on TC plastic resulted in a ~2.7-fold increase in TRPM1 mRNA level compared to control cells. However, knocking down FAK resulted in a slight increase in both Cyr61 and MUC18 mRNA. These small increases were variable among the three independent experiments (Figure 35B), and are also reflected in the small magnitude of the effect. Similar results were obtained when control and FAK KD cells were plated on TC plastic that had previously been coated with FN (Supplementary figure 3). Taken together, this indicates that FAK is a stronger regulator for TRPM1 than for Cyr61 and MUC1 expression.   Figure 35: FAK KD differentially regulate mRNA levels of the signature genes.  Fold changes in signature gene mRNA levels for FAK KD cells and control B16F1 cells plated on TC plastic for 24 h. mRNA levels were determined by qRT-PCR. Mean and SD for 3 independent experiments. Student’s paired t-test, *, p < 0.05; **, p < 0.01.     169  5.4 Summary and discussion Integrin adhesomes regulate membrane-proximal and membrane-distal signalling functions that affect cell behaviors including migration, proliferation and ECM remodelling761-763. Alterations in cell-ECM interactions or adhesion signalling contribute to a range of diseases including cancer393, and therefore adhesome proteins are appealing new therapeutic targets. Understanding the role of individual adhesome members, how they regulate adhesion function, and how targeting them affects tumor cell function, could predict their therapeutic value as future therapeutic targets. To better understand how adhesome function contributes to cancer progression, I investigated the role of two key adhesome proteins, talin and FAK, in regulating the expression of the cancer signature genes Cyr61, MUC18 and TRPM1. I found that the mRNA levels for the three cancer signature genes Cyr61, MUC18 and TRPM1 mRNAs were affected by knocking down the adhesome proteins talin and FAK. FAK and talin are key members of the integrin adhesome764 and both play key scaffolding roles in recruiting structural and signalling proteins to the adhesome ( Figure 37A). Therefore it is not surprising that knocking down talin or FAK regulates the expression of the cancer signature genes. However, my results revealed complex regulation of these genes by talin and FAK. As summarized in Figure 36, knocking down talin resulted in downregulation of MUC18 and upregulation of TRPM1, but did not regulate Cyr61. This regulation was primarily dependent on talin 1 rather than talin 2. Knocking down FAK generally had a weaker regulatory effect than knocking down talin and all three signature genes were weakly upregulated in FAK KD cells. This suggests that talin and FAK have overlapping but distinct regulatory effect on the expression of the signature genes.     170   Figure 36: Changes in adhesome composition regulate cancer signature gene expression.  Talin as a regulator of the cancer signature genes expression  I found that the regulation of MUC18 and TRPM1 mRNA levels was dependent on talin. Reduced talin expression altered MUC18 and TRPM1 mRNA levels in a manner similar to what occurred when the cells were forced into suspension or plated on low FN densities. This may reflect the fact that knocking down talin results in reduced activation of integrins through inside-out signalling751, resulting in less adhesion and less integrin signalling ( Figure 37B). The failure to fully activate integrins and the accompanying reduced assembly of adhesomes in talin KD cells may explain why MUC18 and TRPM1 are regulated in the same way as in B16F1 cells that are not adherent or adhered to low FN density.    171  To test whether knocking down talin in B16F1 cells results in reduced integrin activation, I could compare the levels of activated surface integrins between talin KD cells and control cells. However, this approach is limited due to the lack of reliable antibodies that exclusively bind to the activated conformation of most αβ integrin heterodimers. To circumvent this limitation, testing whether talin KD cells exhibit decreased adhesion to diverse ECM substrates, such as FN, laminin, vitronectin, or collagen I, which is mediated by different integrins350, could elucidate whether talin KD reduces the activation of specific integrin heterodimers.  However, even when talin KD cells are able adhere to ECM, as is in the case for FN, it is likely that the phosphorylation of adhesome proteins such as FAK, vinculin and paxillin is reduced because talin is also important for outside-in integrin signalling. Key integrin adhesome pathways such as the PI3K/Akt pathway and the Ras/ERK pathway, as well as activation of GTPases such as Rap1, Rac1 and Cdc42 would likely be reduced in talin KD cells, as it is the case for B16F1 cells that are adhered to low FN density or forced into suspension. This could be tested by comparing control and talin KD cells for the activation of downstream targets of integrin signalling pathways such as ERK, Akt, JNK, and the RhoA, Rac1, Rap1 and Cdc42 GTPases. Taken together, knocking down talin likely impairs outside-in signalling, leading to reduced integrin activation and impaired inside-out signalling, which together results in reduced activation of integrin signalling pathways. Both defects may be the reason for the observed changes in cancer signature gene expression in B16F1 talin KD cells.    172    Figure 37: Model for how talin KD could affect adhesomes. (A) A model showing how talin binds to integrin and F-actin, initiating the formation of an integrin-vinculin-F-actin complex that recruits other adhesome proteins such as FAK, paxillin, p130Cas, and PI3K. PI3K promotes the conversion of PIP2 to PIP3 and activates downstream signalling pathways (e.g. Akt). (B) Talin KD may result in defects in outside-in integrin signalling with reduced integrin activation mimicking cells in suspension or adhesion to low integrin ligand density. (C) Talin-depleted adhesions in cells that do adhere to FN fail to recruit or activate adhesion complex proteins such as vinculin, paxillin and phosphorylation of FAK and therefore alter adhesome signalling, i.e. defects in outside-in signalling.       173  Talin KD cells took longer to adhere to TC plastic or FN than wild type or control cells. They did eventually adhere, although with a significantly reduced spreading area. This could explain why knocking down talin did not completely mimic a non- or less-adherent phenotype. Consistent with this idea, Cyr61 mRNA levels were not affected by talin KD, perhaps because residual talin in the KD cells was sufficient to mediate changes in the expression of Cyr61 but not of MUC18 or TRPM1.  Although talin is best characterized as an integrin-activator and is a major linker between integrins and the cytoskeleton, there are other adhesome proteins and integrin-binding proteins, such as kindlins, filamins and α-actinin765 that may partially compensate for the loss of talin. Although these proteins may allow talin KD cells to adhere in an integrin-dependent manner, they do not provide the mechanical link between ligand-bound integrins and the actin cytoskeleton that is required to initiate FA-dependent signalling pathways and support cell spreading. Moreover, adhesomes that arise from these talin-deficient integrin adhesions likely differ substantially from adhesions that have talin as their main integrin activator, scaffolding protein, and integrin-cytoskeleton link ( Figure 37C). For example, in mammary epithelial cells vinculin, paxillin, and ILK are absent in talin-depleted adhesions450. Although FAK is present in these adhesions, it fails to become phosphorylated on Tyr397, suggesting that it is not activated450. Interestingly, other adhesome proteins and pathways are not altered in talin-deficient adhesions in mammary epithelial cells450. The lack of FAK-pTyr397 in adhesions in talin-depleted cells appears to be a selective effect as the phosphorylation of Src, Akt, and ERK was not reduced in these cells. Importantly talin-depleted cells exhibited reduced cell proliferation, perhaps due to the loss of talin-dependent suppression of the cell cycle inhibitor p21. This illustrates that changes in   174  adhesome composition can affect some pathways without affecting others. To clarify how talin-depleted activation of integrin signalling pathways affects gene expression in B16F1 cells, one could use proteomics approaches to assess (phospho-) adhesome composition, as done by Robertson et al.380. This could be combined with the assessment of the activation states of key integrin signalling pathways. These signalling differences could then be linked to transcriptome differences between cells with and without talin. This could provide a better understanding of how talin regulates signalling pathways and affects gene expression. There has been a longstanding debate as to whether talin 1 and talin 2 have redundant functions. The current view is that the two talins have partially overlapping functions, although this may be tissue-specific and context-dependent504, 509-511. There is some structural information on how talin 1 and talin 2 interact with integrin β tails and the relative affinities of talin 1 and talin 2 for β1A, β1D and β3 integrin tails have been determined462. However, the affinities and regulatory mechanisms that determine which of the 24 unique αβ heterodimers (not including splice variants) are bound by talin 1, talin 2 or both are not known462. Therefore, we lack a complete understanding of how integrin/talin interactions vary in biologically significant ways and how changes in the relative expression levels of talin 1 and talin 2 affect cell behavior or contribute to cancer progression.  I addressed this question by selectively knocking down either talin 1 or talin 2 in B16F1 cells. Selectively knocking down the expression of talin 1 or talin 2 had different effect on the levels of MUC18 and TRPM1 mRNA. I cannot formally rule out the possibility that talin 1 and talin 2 have similar functions and that the effect of either knockdown is due to a decrease in the total amount of talin. However, my findings suggest that the two talin isoforms do not compensate for each other in this context and that talin 1 and talin 2 have non-redundant   175  functions in B16F1 cells. This is also supported by the observation that talin 1 KD and talin 2 KD cells have different morphologies and that cell spreading is decreased by talin 1 KD but increased by talin 2 KD. Therefore B16F1 cells may rely on a finely tuned balance of interactions between the two talins and integrins in order to achieve optimal cell migration and invasion. This idea is further supported by NMR studies that show that even highly similar integrin tail structures have large differences in their affinities for talin and that the structural conformations with which talin 1 and talin 2 bind to integrin tails is significantly different from each other462. This could result in the preferential binding of talin 1 or talin 2 to specific integrins, which would allow the two talins to activate different integrins and assemble different adhesomes. With proteomic and pathway analysis tools, similar to the ones I suggested for the talin double KD above, it would be possible to differentiate between talin 1 and talin 2 functions in integrin function and to better understand the individual effects of the two talin isoforms on cellular behavior. Proteomic analysis would also provide information about the relative levels of talin 1 and talin 2 in B16F1 cells. The antibody that I used to detect both isoforms likely binds to talin 1 with a higher affinity than to talin 2 (sequence comparison and personal communication with manufacturer), making it difficult to draw conclusions about relative expression levels based on immunoblotting.   FAK as a regulator of the cancer signature genes FAK regulates biological process that are important for the pathogenesis of cancer523. Consistent with this, it is overexpressed in several human epithelial tumors544, 766, 767. FAK has both kinase-dependent functions and kinase-independent scaffolding functions, both of which have been linked to tumor progression, cell motility, invasion, survival, changes in gene expression, and   176  cancer cell self-renewal523, 755. In this chapter, I have shown that knocking down FAK, which reduces both the kinase activity and scaffolding functions of FAK, increased the mRNA levels of all three genes. To confirm that these changes are dependent on FAK and not to off-target effects of the siRNA, I could re-introduce FAK into the KD cells. This should reverse the changes in signature gene mRNA levels caused by the knockdown. Alternatively, I could confirm my results by using a second siRNA that targets a different region of the FAK gene.  FAK activity is associated with FA assembly, maturation and turnover768, 769 and the loss of FAK function is associated with more stable adhesions. In B16F1 cells, knocking down FAK resulted in an increase in cell spreading, with loss of cell polarity and leading edge formation and F-actin fibers formed along the periphery of the cells. This suggest that knocking down FAK in B16F1 cells leads to more stable adhesions, however, FRAP analysis of adhesion complexes would be necessary to determine whether knocking down FAK increases FA stability. The morphological changes in FAK KD cells are in contrast to those observed in talin KD cells, where impaired cell spreading potentially allowed the formation of only less mature adhesions. This could account for the differences in the expression of the Cyr61 and MUC18 mRNA levels caused by knocking down talin versus knocking down FAK. Our results suggest that an inability to form mature adhesions result in reduced expression of Cyr61 and MUC18. However, TRPM1 did not follow that pattern suggesting that its regulation may be independent of maturation states of the adhesions. Within focal adhesions, some proteins act as scaffolding molecules whereas a large number of signaling molecules regulate downstream pathways and biological functions. As FAK has both scaffolding and kinase activity, this raises the issue regarding different roles of the signaling and scaffolding roles of FAK in the signature genes. However, kinase activity and   177  scaffolding function are not necessarily independent of each other and may therefore be challenging to distinguish. For example, it has been reported inhibition of the FAK kinase activity may enhance its scaffolding function294. This correlates with an increase in the amount of nuclear FAK in cells that have been treated with this inhibitor575. In the nucleus, FAK can interact with several TFs such as GATA4576 and most notably, suppress p53 function770. However, whether nuclear accumulation of kinase-inactive FAK has positive effects, or acts as a dominant negative that interferes with normal FAK functions, is not known. However, there are several lines of evidence pointing to kinase-driven and scaffold-driven functions of FAK, and studies have shown that FAK KD effects are not always recapitulated by inhibition of FAK kinase activity. For example, it has been shown that FAK expression but not kinase activity is necessary for cell motility in response to growth factors771.  To distinguish between the effects of the FAK scaffolding function versus its kinase activity on the signature gene expression, I could use FAK kinase specific chemical inhibitors or express a catalytically inactive form of FAK in FAK KD cells. This could indicate whether it is the kinase function that is important in regulating the signature gene expression.  In summary, knocking down talin, as well as knocking down FAK had overlapping but distinct effects on the expression of the three signature genes, as well as on cell morphology, the organization of the actin cytoskeleton and the localization of the adhesome proteins β1 integrin and vinculin. This illustrates that alterations in adhesome composition not only affect adhesion structure and localization but are also able to regulate overall cell behavior and gene expression. This may explain why aberrant adhesome signalling or mutations in adhesion proteins can contribute to cancer progression. Therefore, it would be interesting to focus more on how adhesion complexes regulate global cellular functions. Answering this question will help us   178  understand how aberrant adhesome composition and function may lead to diseases and could point to specific proteins as potential therapeutic targets.       179  Chapter 6: Talin 1 and talin 2 differentially regulate B16F1 cell spreading, migration and tumor growth  6.1 Introduction In chapter 5, I showed that talin 1 and talin 2 have different effects on the expression of the signature genes and that they differently regulate cell shape and motility. Knocking down talin 1 resulted in decreased cell spreading, while knocking down talin 2 resulted in increased spreading. In terms of motility, bead clearing assays showed that knocking down either form of talin resulted in decreased motility, whereas knocking down both isoforms of talin also reduced motility, but not to the same extent as the individual knock downs. These observations suggested both overlapping and unique roles of talin 1 and talin 2 in cancer-related processes.   The aim of this chapter is to study the distinct and overlapping functions of talin 1 and talin 2 in cancer related processes in vitro and in vivo. Because there are no talin KO cancer cell lines available, I first generated talin 1 and talin 2 single KO and double KO B16F1 cell lines. To create these cell lines, I used the clustered regularly interspaced short palindromic repeat (CRISPR)/Cas technique. This system is based on a prokaryotic adaptive immune system that was modified to enable editing of mammalian genomes772-776. CRISPR/Cas uses short RNAs (termed CRISPR RNAs (crRNA) or single-chain guided RNA (sgRNA)) as guides for the endonuclease to cleave genomic DNA at a predefined target sequence of interest and create a double-strand break (DSB)777, 778. Following the introduction of a DSB, the cellular DNA repair machinery can repair the damage using either non-homologous end joining (NHEJ) or a homology-directed repair mechanism. NHEJ, the mechanism I relied on, joins the two broken ends together. However, since there is no homologous template for the DNA repair, NHEJ   180  typically leads to the introduction of small insertions and deletions at the site of the break, often inducing frame-shifts that knock out gene function. With these KO cell lines, I then assessed the effects of losing either or both talin isoforms on cell spreading, cell motility, in vitro proliferation, and the ability to grow tumors after s.c. injections.   6.2 Results  6.2.1 Creation of talin 1 and talin 2 single and double KO B16F1 cell lines  To knock out the two isoforms of talin, I designed crRNAs that target the coding region of either isoform according to Hsu et al.779, and optimized them as recommended by Fu et al.780 in order to improve specificity. Talin 1 was targeted in exon 2 and talin 2 was targeted in exon 3, sites that are just downstream of the start codon (Figure 38B). B16F1 cells were transfected with plasmids containing the crRNA for either talin 1 or talin 2, or with both plasmids. Cells were sorted for transient expression of the CRISPR plasmid, which encodes for an orange fluorescent protein, and then single cell clones were generated by sequential dilution. Single colonies that grew up in wells of a 96-well plate were identified, expanded and screened for talin 1 and talin 2 expression by western blotting. It is important to note that the talin 1/2 double KO cells were generated in the same co-transfection experiment as the talin 1 KO cells and were not generated by sequentially knocking out talin 1 and then talin 2. Talin 2 KO cells were generated in a separate transfection. All KO cell lines were generated from a single clone each. Figure 38C shows talin 1 and talin 2 protein expressions for talin 1 KO, talin 2 KO, and double KO B16F1 cell lines and for respective control cell lines. Both control cell lines show bands for talin 1 and talin 2, whereas, talin 1 KO cells lack the band for talin 1, while still exhibiting the band for talin   181  2. Conversely, talin 2 KO cells lack the band for talin 2, while exhibiting the band for talin 1. The double KO clone did not exhibit a detectable band for either talin 1 or talin 2. Several more KO clones are still being generated and validated, and that analysis of the mutations introduced by CRIPSR/Cas9 in all clones is still underway. Note that I cannot exclude the possibility that these KO cells produce shorter forms of talin 1 and talin 2 from alternative start codons505. These shortened proteins would not be detected because the epitopes recognized by the respective antibodies are near the N-termini (for talin 2) or middle (for talin 1) of the proteins. Clones that went through the same process, but which exhibited no alteration in talin 1 or talin 2 expression, were used as control cells for all experiments.      182   Figure 38 CRISPR/Cas strategy to target talin 1 and talin 2 in B16F1 cells.  (A) CRISPR/Cas9 system cleaves 3 base pairs upstream of the protospacer adjacent motif (PAM) sequence and creates a DSB. (B) Talin 1 crRNA target sequences in exon 2 and talin 2 crRNA target sequences in exon 3. Both sequences are located within the first coding exon, 88 base pairs (talin 1) and 99 base pairs (talin 2) downstream from the respective start codon. (C) Total cell lysate single KO, double KO and control B16F1 clones were probed for talin 1 or talin 2 using specific antibodies. Actin was used as a loading control. Molecular weight markers are in kDa.   6.2.2 Talin 1 and talin 2 regulate cell spreading and cell morphology  To test the effect of knocking out the genes encoding either talin 1, talin 2 or both on B16F1 cell spreading, the cells were plated on FN for 90 min and the cell area close to the coverslip was imaged by staining for F-actin. Compared to control cells, talin 1 KO cells exhibited a reduction in cell spreading area (Figure 39B, D). Double KO cells exhibited a more dramatic reduction in cell spreading area that was greater than that caused by knocking down   183  talin 1 only. Knocking out talin 2, however, resulted in an increase in cell spreading area. In addition to exhibiting altered spreading, the KO cells also adopted different cell shapes than control cells (Figure 39A). To quantify this, control and KO cells were analyzed for the shape factor circularity (C), which is the ratio of the area to the perimeter of the cell781, 782. C is a description of the roundness of the cell. A perfect circle has a value of 1 and values decrease as the shape becomes more irregular. Many control cells had an irregular shape with a median C value of ~ 0.4 (Figure 39B). In contrast, both talin 1 and double talin KO cells exhibited an increased circularity, with median C values of ~ 0.7 (Figure 39B). Talin 2 KO exhibited a slight decrease in circularity, compared to their control cells, indicating that they have a more irregular shape than control cells, talin 1 KO and double KO cells. Similar trends were seen across multiple repeats (Figure 39D, E). Thus, talin 1 and talin 2 both modulate cell spreading and cell shape in B16F1 cells.    184      185  Figure 39: Talin 1 and talin 2 have different effects on cell spreading and cell shape. (A) Representative cell shapes of control and KO B16F1 cells that were plated on FN for 90 min. Staining of F-actin was used to determine the outline of the cell and converted to a binary image using ImageJ. (B) Cell spreading area for control and KO B16F1 cells that were plated on FN (1.33 µg/cm2) for 90 min. (C) Circularity (C) of control and KO B16F1 cells, calculated using the formula circularity = 4π(area/perimeter2). A value of C equal to 1 represents a circle. Values < 1 indicate increasingly irregular shapes. Each dot represents one cell. Median with interquartile range, n > 50 cells, ****, p <0.0001; ***, p <0.001; **, p <0.01; ns, not significant, Mann-Whitney test. One representative experiment is shown of 4 independent experiments. (D, E) Mean + SEM of 4 independent experiments from B and C; Paired t-test. *, <0.05; ****, p <0.0001; ***, p <0.001; **, p <0.01.    6.2.3 Talin 1 and talin 2 regulate cell motility and velocity  Because both talin 1 and talin 2 regulate cell area size as well as cell shape, I asked whether talin 1 and talin 2 also regulate cell motility. To test this, control B16F1 cells were mixed with KO cells, plated on FN coated slides, and imaged overnight at 10-minute intervals. KO cells or control were dyed with CellMask™ Orange Plasma membrane stain to distinguish them from each other. To exclude potential artifacts caused by the dye, replicates were performed in which the cell population that was dyed was switched. Compared to control cells, all of the KO cell types showed a 40% decrease in total migration distance (Figure 40A). The loss of either or both talins did not result in a complete loss of motility, suggesting that B16F1 cells also employ talin-independent mechanisms to migrate. The reduced migration distance exhibited by the KO cell lines was due to a reduction in velocity, rather than some cells not migrating at all while others migrated at normal speed. The various control cells migrated 1.2 - 1.5 µm in 10 min in multiple experiments. In contrast, the talin 1 KO cells only migrated on average 0.9 µm, talin 2 KO cells 0.6 µm, and talin double KO cells 0.8 µm in 10 min (Figure 40B). When total migration distances were binned, it became clear that control cells as well as KO cell lines showed a range of distribution of migration distances (Figure 40C). In comparison   186  to control cells, which had a wide range total migrated distances (54 - 443 µm) with a median of 228 µm in 13.5 h migration time, talin 1 KO cells exhibited a reduced range of total migrated distances (20 - 250 µm), with a median of 154 µm (Figure 40C, top). During 17 h of migration, control migrated 30 - 430 µm with a median of 242 µm, while the total range of migration of talin 2 KO cells was 30 - 134 µm with a median of 100 µm (Figure 40C, top). When comparing total migration distance between control and talin double KO cells for 13.3 h, the loss of both talins reduced the range of migration from ~400 µm (27- 425 µm) with a median of 183 µm in control cells to 300 µm (22-321 µm) with a median of 101 µm in talin double KO cells (Figure 40C, bottom). Taken together, the loss of either talin reduces overall migration distance and migration velocity. This suggests that both talin 1 and talin 2 contribute to adhesion-dependent cell migration and do not compensate for each other.    187   Figure 40: The loss of either or both talins results in reduced 2D migration distance and velocity on FN.  (A) Total migration distances on FN (2µg/cm2) for KO cell lines and relevant control cell lines. The migration distance for KO cell lines were normalized to control cells (≡100%). Mean and SEM migration of control cells are represented. (B) Mean with 95% CI of velocities (per 10 min) for each cell line was plotted, ****, p <0.0001, Mann-Whitney test. (C) Total distances migrated   188  were binned into 20-µm bins and relative frequencies for control and talin 1 KO cells (13.5 h total migration time), control and talin 2 KO (17 h total migration time), control and talin double KO (13.3 h total migration time) were blotted. The median migrated distances for the control and KO cell lines are indicated by the red and green stars, respectively. Black and grey lines show Gaussian distributions fitted by non-linear regression. Talin 1 and talin double KO cells were analyzed from 2 independent experiments, talin 2 KO from 3 independent experiments. In each experiment either control or KO cells were dyed with CellMask™ Orange Plasma membrane stain to distinguish them from each other. Replicates were performed in which the cell population that was dyed was switched.     6.2.4 Talin 1 and talin 2 may determine the mode of cell motility in B16F1 cells  Many cancer cells switch between mesenchymal and amoeboid motility319, 783-785 (see section 1.4.2). This plasticity of motility may be a key factor in cancer metastasis because disseminating tumor cells need to migrate through a range of ECM milieus in order to escape the original tumor and establish tumor growth in distant organs. Recent work786-788 suggests that a balance of three cell-intrinsic parameters, protrusion type, actomyosin contractility and the cell’s ability to form adhesions, determine the mode of motility. For example, both knocking down talin 1 and talin 2, or using an integrin β1 integrin-blocking antibody, increases the proportion of normal human dermal fibroblast cells that switch from mesenchymal to amoeboid motility329, suggesting that the switch to amoeboid-like migration is due to the absence of focal adhesions.  B16F1 cells plated on FN coated coverslips generally exhibit mesenchymal motility and assume a spread morphology with pronounced and persistent leading edges (Figure 41, i). However, some of the B16F1 cells also transitioned to a more elongated morphology with pseudopodia-like structures. These cells generally migrated less (Figure 41, ii). Consistent with published results329, talin double KO B16F1 cells migrated predominantly with amoeboid-like motility, characterized by a round shape and a small leading edge (Figure 41, vii and viii). This is consistent with the idea that the loss of both talins drives MET in B16F1 cells and results in   189  amoeboid-like migration, and agrees with findings in other cancer cells329. However, the loss of either talin separately was not sufficient for MAT and did not completely impair mesenchymal motility. Talin 1 KO cells did not switch to amoeboid motility. Even though they formed fewer persistent leading edges (Figure 41, iii), they frequently adopted a cell morphology characterized by both short and long pseudopodia-like protrusions (Figure 4, iv). Like talin 1 KO cells, talin 2 KO cells did not switch to amoeboid motility. However, in contrast to talin 1 KO cells, many talin 2 KO cells did not exhibit leading edges but adapted a star-like morphology with short pseudopodia (Figure 41, v and vi). This suggests that talin 1 and talin 2 have different roles during mesenchymal migration in B16F1 cells.          190     Figure 41: Talin KO and control B16F1 cells have different morphology during 2D migration.  (A) Videos were taken of control cells (top row), talin 1 KO cells (second row), talin 2 KO cells (third row) and talin double KO cells (bottom row) that were plated on FN-coated (2µg/cm2) glass slides (phase contrast imaging). Snapshots of one representative cell for each cell type are displayed between 2 h and 16 h (left to right) after the cells were plated on FN-coated coverslips. Control cells form pronounced leading edge structures in the direction of migration (i), but intermittently transition into a morphology with long randomly-oriented pseudopodia (ii). Talin 1 KO cells form smaller leading edges (iii) and transition to a cell morphology with short   191  pseudopodia (iv). Talin 2 KO cells exhibit star-like morphology with short pseudopodia (v and vi). Talin double KO cells adapt amoeboid morphology with intermittent leading edge structures in the direction of migration (vii and viii). See Supplementary video 1. (B) Enlarged pictures of snapshots i to viii. Size bar, 70 µm.   6.2.5 Talin regulates B16F1 tumor growth in vivo   Talin has been linked to cell cycle progression and proliferation450, but the relative roles of talin 1 and talin 2 in cancer cell proliferation or apoptosis are not known. Therefore, I asked whether the loss of either talin isoform affects cell proliferation and in vivo tumor growth of B16F1 cells.  Talin 1 and talin 2 do not affect cell growth in vitro on tissue culture plates  To determine whether knocking out either or both talins affected cell viability of B16F1 cells in vitro, I performed a time course over three days, measuring cell viability using Alamar Blue (Resazurin sodium salt). The Alamar Blue assay is a well-established method designed to quantitatively measure cell viability in cell lines789. The dye is a redox indicator dye, which is converted into the red-fluorescent resorufin by metabolically-active cells. Therefore, the amount of fluorescence produced is proportional to the number of living cells.  The loss of talin 1 or talin 2, or both talin isoforms did not significantly affect viability of B16F1 cells that were plated on TC plastic ( Figure 42, left). However, when cells were kept in suspension by culturing them on low adhesion plates, the talin double KO cells exhibited reduced viability compared to control cells on day 2 and day 3 (   192  Figure 42, right). This indicates that talins are not essential for cell viability when cells adhere, but are important when the cells are grown in anchorage-independent conditions. Furthermore, it shows that talin 1 and talin 2 can compensate for the loss of each other, because knocking out only talin 1 or only talin 2 did not reduce the viability of cells growing in suspension. To test whether knocking down talin also affects tumor growth, I went on to test the in vivo growth of B16F1 control and KO cells in situ.     Figure 42: The loss of both talin isoforms reduces B16F1 cell viability when cell are forced into suspension. Cells were grown in complete medium either on TC plastic (left) or on plates that prevent cell attachment for 3 days. Cells are assayed for cell viability daily using the metabolic indicator Alamar Blue. Fluorescence values for days 1, 2, and 3 were normalized to the fluorescence value for the corresponding cell line that was determined immediately after plating (Day 0). Mean and SD for 3 independent experiments are shown for all cell lines, except for the double KO on TC plastic which there were 2 independent experiments. ns = not significant; ***, p < 0.001; ****, p < 0.0001; 2-way ANOVA.    193     Talin 2 is important for s.c. tumor growth of B16F1 cells in C57BL/6 mice  To test if the loss of either or both talins affects tumor growth in vivo, I injected 105 control or KO cells into the flanks of C57BL/6 mice. Each mouse was injected with KO cells in the right flank and the relevant control cells in the left flank. At the experimental end point, tumors from both flanks were excised and compared pairwise for weight as an indicator of tumor growth (Figure 43). Compared to control cells, talin 1 KO cells had no reduction of tumor weight with a median of ~0.1 g for both cell types (Figure 43 left). Surprisingly, knocking out talin 2 dramatically reduced the ability of B16F1 cells to form tumors in situ, with most s.c. injections of talin 2 KO resulting in no tumors or tumors that were too small to weigh. In these experiments, the control cells yielded median tumor weights of ~0.12 g (Figure 43, middle). Interestingly, the reduction in tumor growth was less pronounced when both talins were knocked out. Tumors arising from double KO cells had a median weight of ~ 0.05 g, compared to ~0.1g for control cells (Figure 43, right). This suggests that talin 1 is not necessary for tumor growth in vivo but that talin 2 is essential. The finding that talin double KO cells form bigger tumors than talin 2 single KO cells suggests that talin 1 has a negative effect on tumor growth that is overcome when talin 2 is present.    194   Figure 43: Talin 2 KO impairs subcutaneous tumor growth Control or KO B16F1 cells (105) were injected s.c. pairwise into the flank of C57BL/6 mice. n =13 mice for talin 1 KO cells, n= 8 mice for talin 2 KO cells, n = 11 mice for talin double KO cells. At the humane endpoint (13-27 days post injections), both tumors were excised and weighed as an indicator of tumor growth. Dots represent individual tumor weight, with the KO and control cells-derived tumors from the same mouse connected by a line. Red bars represent the median for each group. (Pairwise comparison, Student’s t-test, ns = not significant; ***, p < 0.001).   6.2.6 Re-expressing talin in talin KO clones partially rescues normal cell behavior Because the CRISPR/Cas9 KO cell lines are clonal, it is possible that altered phenotypes they display are a clonal effect rather than an effect based on the loss of expression of talin 1, talin 2 or both. Therefore I reconstituted the double and single KO clones with an egfp-tagged version of talin 1 and talin 2, to see whether restoration of talin expression would revert the phenotype. To do that, egfp-talin 1 and egfp-talin 2 were subcloned into an expression vector with a β-actin promotor (Supplementary figure 4). To create stable cell lines, the β-actin-egfp-talin 1, β-actin-egfp-talin 2 or the β-actin-egfp (control) constructs were transfected into the   195  double KO cells and single KO cells. The cells were then selected with G418 for 2 weeks, and sorted for egfp-positive cells.   To assess the expression level of the egfp-fusion proteins in the KO cells, I performed both western blotting and FACS (Figure 44). For immunoblotting, total cell lysates of control, KO and reconstituted KO cells were probed for egfp, talin 1 and talin 2. The antibody against egfp detected bands that correspond to the estimated size of egfp-talin fusion proteins in the reconstituted KO cells, or to the size of egfp in the cells that only express egfp (Figure 44A top panel). Unfortunately the addition of the 25-kDa egfp protein resulted in a band that co-migrated with a known background band that is detected by the talin 1 antibody (Figure 44A middle panel), making it difficult to compare the amount of egfp-talin 1 in the reconstituted KO cells with the amount of talin 1 in the control cells. In contrast, a talin 2-specific antibody detected a band corresponding the estimated size of a egfp-talin 2 fusion protein in KO cells that had been reconstituted with egfp-talin 2 (Figure 44A, bottom panel).  Single cell analysis by FACS showed that egfp-talin 1 and egfp-talin 2 were expressed at similar level, with a slightly wider spectrum of expression levels for egfp-talin 1 than egfp-talin 2 (Figure 44A top). The level of egfp-talin 1 in the talin 1 KO cells was slightly higher than egfp-talin 2 that was expressed in talin 2 KO cells (Figure 44A bottom panel). In the control cell lines, the egfp fluorescence was at least as high as that for the egfp-talin fusion proteins. Taken together these data confirmed that the egfp-talin fusion proteins were successfully expressed in the talin KO cells.    196   Figure 44: Reconstitution of double and single talin KO cells with egfp-tagged talin proteins.  (A) Western blotting of reconstituted egfp-talin KO cells. Total cell lysates of control cells, KO cells, and reconstituted KO cells were probed for egfp (top), talin 1 (middle), or talin 2 (bottom). Note that the additional 25 kDa of the egfp that is fused to talin 1 resulted in an egfp-talin 1 band that co-migrated with a known background band (*) that is detected by the talin 1 antibody. β-actin serves as loading control. (B) Representative FACS plots of double talin KO cells (top) and single talin KO cells (bottom) that were reconstituted with egfp-talin isoforms or egfp, as indicated.   To assess whether the observed phenotypes in talin KO cells were due to the loss of talin, I performed spreading assays on FN to see whether the reintroduction of talin 1 or talin 2 was able to revert the respective phenotypes. As expected, both egfp-talin 1 and egfp-talin 2 located to adhesions when expressed in either the double KO or respective single talin KO cells. The expression of either egfp-talin 1 or egfp-talin 2 resulted in a partial rescue of the cell spreading defect in the talin double KO cells. In comparison to control cells, which express both talin 1 and talin 2, double KO cells had a significantly reduced spreading on FN (Figure 45B), as seen previously (Figure 39A). Expressing either egfp-talin 1 or egfp- talin 2 in the double KO cells   197  caused the cells to spread to a significantly greater extent than the double KO cells, although not to the same extent as the control cells. Thus spreading defect in the double KO cells can be partially corrected by expressing either talin 1 or talin 2, suggesting that both of these talin isoforms promote cell spreading in B16F1 cells.  There are several possible reasons why re-expressing either egfp-talin 1 or egfp-talin 2 did not fully reconstitute spreading in the talin 1/2 double KO B16F1 cells. Talin 1 and talin 2 may have distinct functions and both are capable of supporting B16F1 cell spreading. Alternatively, talin 1 and talin 2 could have some common functions and the lower levels of total talin in the singly-reconstituted cells than in wild type control cells may not be sufficient to fully restore talin function. If this were the case, then expressing either talin 1 or talin 2 at higher levels in the double KO cells might result in more complete restoration of cell spreading. Another possibility is that the addition of egfp to the talin proteins impairs their function or localization. Finally, one cannot rule out the possibility that the talin 1/2 double KO clone has additional defects or changes compared to the control cells that cannot be corrected by re-expressing talins. Nevertheless, my findings show that re-expressing talin 1 or talin 2 in the double KO cells substantially reverts the cell spreading defect, suggesting that both isoforms are involved in B16F1 cell spreading. Moreover, a role for talin 1 in B16F1 cell spreading on FN is supported by the finding that reconstituting talin 1 single KO cells with egfp-talin 1 completely reverted the spreading defect observed in talin 1 KO cells (Figure 45).    198   Figure 45: Reconstitution of double talin KO cells with either egfp-talin 1 or egfp-talin 2 partially restores cell spreading on FN. (A) Representative WT control, double KO cells, and reconstituted cells were plated on FN (2 µg.cm2) for 90 min. F-actin (red), respective talin isoform (green). (B) Quantification of >100 cells per cell type. Median with interquartile range, ****, <0.0001; Mann-Whitney test. One representative experiment is shown out of 3 independent experiments (top). Mean + SEM of 3 independent experiments; **, <0.01, *, <0.05; Paired t-test (bottom).       199   Figure 46: Reconstitution of talin 1 KO cells with egfp-talin 1 restores cell spreading on FN. (A) Representative WT control, talin 1 KO cells, and reconstituted talin 1 KO cells were plated on FN (2 µg.cm2) for 90 min. F-actin (red), talin 2 isoform (green). (B) Quantification of spreading are of >100 cells per cell type. Median with interquartile range, ****, <0.0001; Mann-Whitney test. One representative experiment is shown of 3 independent experiments (top). Mean + SEM of 3 independent experiments; *, <0.05; Paired t-test (bottom).   In contrast to talin 1, knocking down talin 2 using siRNA or disrupting the talin 2 gene using CRISPR technology, resulted in a slightly increased spreading area when the cells were plated on FN (Figure 32 and Figure 39A). This was reverted when egfp-talin 2 was re-expressed in talin 2 KO cells, as compared to talin-2 KO cells that were transfected with egfp. Note that over multiple repeats the initially observed increase in spreading upon loss or reduction of talin 2   200  was not as strong or as consistent as in initial experiments (compare Figure 39). This may be due to the cells being in culture for a longer period of time in order to generate the reconstituted cell lines. Nevertheless, the effect of talin 2 on cell spreading is different and less pronounced than that of talin 1.   Figure 47: Reconstitution of talin 2 KO cells with egfp-talin 2 does not affect cell spreading on FN.  (A) Representative WT control, talin 2 KO cells, and reconstituted talin 2 KO cells on FN (2 µg.cm2) for 90 min. F-actin (red), talin 2 isoform (green). (B) Quantification of > 100 cells per cell type per experiment. Median with interquartile range, ****, < 0.0001; Mann-Whitney test. One representative experiment is shown of 4 independent experiments (top). Mean + SEM of 4 independent experiments; *, < 0.05; Paired t-test (bottom).     201  6.3 Summary and discussion  In this chapter I tested the hypothesis that talin 1 and talin 2 have distinct roles in cancer-related processes such as cell spreading and migration, cell proliferation and tumor formation. Using an initial set of control and talin KO clones generated using CRISPR/Cas9 technology, I found that the loss of either talin 1, talin 2, or both talins resulted in decreased B16F1 cell motility and velocity, altered cell morphology, and in the case of talin 2, decreased ability of the cells to grow in vivo, as summarized in Figure 48.   Figure 48: Knocking out talin 1, talin 2, or both talins has different effects on processes related to cancer progression.  The CRISPR/Cas9 knockout strategy has recently gained popularity as a novel loss-of-function approach790, 791. Previous screens using CRISPR/Cas9 have identified genes that are essential for cell survival and responsible for drug resistance in cell lines. Furthermore, this technique has been used as a screening tool in a mouse model for tumor evolution792, and identified novel tumor suppressor genes. I used CRISPR/Cas9 to create B16F1 knockout cell lines lacking either talin 1 or talin 2, or both talins, as confirmed using isoform-specific antibodies. These cells lines provide an important new tool for us to investigate the specific contributions of talin 1 and talin 2 in tumor progression.   202   To confirm that the LOF phenotype of talin 1 and talin 2 was not due to a clonal selection, I reintroduced egfp-tagged talin in the respective single KO cells. The single KO clones reverted their respective LOF phenotype upon re-expression of the specific isoform. Expressing egfp-talin 1 in talin 1 KO cells restored their spreading area to a similar size as the spreading area of the control cells. However, further experiments are needed to determine whether this is an effect of total talin expression levels or specific to talin 1. To address this I could overexpress talin 2 in talin 1 KO cells, which would reveal whether talin 2, if expressed at a high level, could compensate for the loss of talin 1. This could explain why expressing egfp-talin 2 in talin 1/2 double KO cells can partially restore cell spreading even though cell spreading is not defective in talin 2 KO cells. Another way to test the importance of the expression levels of talin 1 or 2 on spreading would be to isolate B16F1 cell clones with high, medium and low expression of either talin 1 or talin 2, and then correlate the expression level to the degree of spreading. In addition, I am planning on co-expressing mCherry-talin 1 and egfp-talin 2 in the double KO cells to determine whether high levels of either talin protein can compensate for low expression of the other talin. Ideally I would be able to match talin expression in the reconstituted clones with the levels of endogenous talin 1 and talin 2 in wild type cells in order to clarify whether the observed phenotypes are due to total levels of talin or whether talin 1 and talin 2 have unique roles in B16F1 cells.  Reintroducing egfp-talin 2 into talin 2 KO cells reverted the increased spreading of talin 2 KO back to control levels, therefore reverting the phenotype of increased spreading on FN observed upon talin 2 KD or in talin 2 KO cells. However, this was less consistent than the effect of talin 1 KO, the role of talin 2 in 2D cell spreading on FN seems to be less significant. My   203  preliminary data comparing double talin KO cells reconstituted with either egfp-talin 1 or egfp-talin 2 in 3D collagen I/FN matrices suggest a more important role for talin 2 than talin 1 in 3D cell growth (data not shown).  During cancer progression, cancer cells break away from primary tumors to colonize new organs. Cells have the ability to switch migration modes in response to different ECMs and in response to intrinsic factors such as the ability of cells to form cell-ECM adhesions. Because talin is a regulator of cell-ECM adhesion through its ability to activate integrins, I investigated whether talin 1, talin 2 and talin double KO cells affected B16F1 cell migration. I find that all three KO cell lines exhibited reduced overall motility and reduced velocity.  Talin double KO cells adopted an amoeboid morphology and a mode of motility that resembled the behavior of leukocytes, which migrate independently of integrin-based adhesion. The reduction in cell spreading area in talin double KO cells may cause the actin-myosin contractile machinery to favor a rounded (higher circularity) cell shape with formation of a contractile cortex that enables amoeboid migration. Interestingly, the amoeboid motility of leukocytes is typically faster than mesenchymal migration. However this was not the case for B16F1 talin double KO cells. On average, talin double KO cells moved significantly shorter distances during the same time interval than control cells. Nevertheless, the switch to amoeboid-like migration in B16F1 talin double KO cells most likely depends on the decrease of FAs.  Although talin 1 KO cells also exhibited reduced motility and often switched to a rounder morphology, they did not switch to an amoeboid morphology or amoeboid motility. Instead, the motility of these cells was characterized by the transient formation of leading edges that resembled the actin polymerization-driven leading edges of control cells. The cells often adopted shapes that were more circular than control cells but which had well defined cell edges with   204  several short protrusions that may depend on actin stress fibers. This suggests that the loss of talin 1 results in initial lamellipodia formation, but reduced cytoskeletal reinforcement, insufficient traction forces, decreased movement of the cell body, and reduced motility.  In contrast, the morphology of talin 2 KO cells differed substantially from control cells, talin 1 KO cells and talin double KO cells. Although talin 2 KO cells also exhibited reduced motility, the cells had a larger spreading area and irregular shapes, with many cells adopting a star-like morphology with numerous spiky protrusions. Cells with this spread morphology exhibited highly reduced motility. This suggests that talin 2 is important for FA turnover. If this were the case, then loss of talin 2 should be accompanied by more stable adhesions with a decreased turnover rate. For example, if talin 2 was important for FA disassembly, but not FA assembly, then the loss of talin 2 would result in ‘stuck cells’ that exhibit reduced motility, as I observed. Further comparison of actin-stress fiber distribution, the formation and distribution of FAs and fibrillar adhesions between talin 1 KO and talin 2 KO cells would shed light on the different roles that talin isoforms have on adhesion-dependent migration.  One way to explain the different effects of knocking out talin 1 or talin 2 could be that the two talin isoforms have differential binding affinities for specific integrins. This is supported by NMR data that show that talin 1 and talin 2 have different affinities for β1 or β3 integrin in vitro462. Talin 1 has a higher affinity for integrin β3 and β1A tails than talin 2, whereas talin 2 has a higher affinity for integrin β1B. If that were the case in B16F1 cells, a change in the relative amounts of activated β1 and β3 integrins due to the loss of either talin could affect integrin mediated migration, depending on the substrate. Determining which integrins each talin associates with in B16F1 cells would provide insights into the potential differential regulation of integrins by talin 1 versus talin 2. Also of particular interest would be to investigate the relative   205  locations of talin 1 and talin 2 within the cell during the migration of B16F1 cells. If talin 1 and talin 2 had different roles during migration, for example in the formation of lamellipodia at a leading edge, or in FA turnover at the center or rear of the cell, they may exhibit differential localization in migrating cells. This could explain why talin 1 KO and talin 2 KO B16F1 cells have different phenotypes.  Even if talin 1 and talin 2 bind to the same integrins, they could recruit different subsets of adhesome proteins, resulting in different adhesome signalling output. Several binding partner of talin 1 have been identified and some of these are shared with talin 2. However, little is known about talin 2 binding partners. To address this potential talin isoform specific difference, proteomic of talin 1-containing vs talin 2-containing complexes and studies comparing the adhesomes compositions of control, talin 1 KO and talin 2 KO cells could be done, as described previously for the adhesomes of specific integrin heterodimers793. This could be then correlated with differences in adhesome signalling between the talin 1 KO and talin 2 KO, which could explain the observed phenotypes of these cells. The viability of B16F1 cells that are cultured on TC plates does not seem to be dependent on either talin 1 or talin 2. Interestingly, talin double KO cells but not talin 1 or talin 2 KD cells exhibited decreased cell viability when grown in suspension. This suggests that talin is important for anchorage-independent growth, and that either talin isoform is sufficient for this function.  In contrast, in vivo tumor growth after s.c. injections was reduced in double KO cells and dramatically impaired in talin 2 KO cells. The loss of both talins resulted in a decrease in tumor size in comparison to control cells. Tumor cells from talin double KO cells were on average half the size of tumors derived from control cells. This may be due to reduced integrin activation that   206  results in anoikis. This is consistent with the observation that double KO cells had reduced viability when cells were grown under conditions that prevented adhesion to TC plastic in vitro. The loss of talin 1 alone did not reduce in vivo tumor growth of B16F1 cells. This finding is in line with the observation that knocking down talin 1 in MDA-MB-231 cells results in even larger primary tumors519. This may be due to talin 2 being able to compensate for the loss of talin 1 for in vivo growth in B16F1 cells.  In contrast to talin 1, knocking out talin 2 resulted in a dramatic reduction in tumor size, and a greatly reduced number of tumors that were formed. This indicates that talin 2 is important for in vivo growth but not for in vitro growth, and must be important for challenges that are not present in vitro.  Surprisingly, the talin double KO cells showed only moderately impaired tumor growth in vivo, a less severe phenotype than talin 2 KO cells. This suggests that the loss of talin 1 in a talin 2 KO setting partly restores tumor growth.  Tumor metastasis involves the ability of tumor cells to home to a distant organ, extravasate, and form secondary tumors. Therefore, it will be interesting to test the ability of talin KO cells to form tumors in the lung after i.v. injection, a model that tests tumor cell extravasation and secondary growth. Preliminary results (not shown) indicate that the double KO B16F1 cells were severely impaired in their ability to form tumor nodules in the lung after i.v. injections. Talin 1 KO and talin 2 KO B16F1 both exhibited reduced numbers of tumor nodules, despite only talin 2 impairing the growth of primary s.c. tumors. Several factors may account for these results. To establish lung nodules, the injected B16F1 cells have to accomplish two major tasks. First, they need to adhere to the endothelial cells of lung vessels and extravasate into the underlying parenchyma. For example, in B16 cells, both α4β1-mediated interaction with VCAM-  207  1 on endothelial cells and the interaction of αvβ3 with FN are important for cell adhesion and transmigration794, 795. Second, the extravasated cells need to proliferate in order to grow into a macroscopic tumor nodule. Talin 1 and talin 2 both affect adhesion and migration in B16F1 cells, and most likely the formation and function of invadopodia, a structure that is necessary for extravasation 796, 797. Therefore I predict that reduced numbers of lung nodules would arise from single talin KO cells, and an even greater reduction in the number of tumor nodules formed by double KO cells. In summary, my findings suggest that talin 1 and talin 2 have overlapping but non-redundant functions for cell spreading and cell migration and that they may regulate the switch between mesenchymal and amoeboid cell motility. Furthermore, my results show that in B16F1 cells talin 1 is not essential for tumor growth, whereas the loss of talin 2 or both talins reduces tumor growth in vivo. This is consistent with my preliminary results indicating that talin 2 is more important for growth in 3D collagen/FN matrices than talin 1. Moreover, other preliminary experiments suggest that the loss of both talins leads to a more dramatic decrease in metastatic potential than the loss of only one isoform. However, it remains to be determined if this reflects distinct functions for talin 1 and talin 2 as opposed to impaired common functions due to lower levels of total talin 1 plus talin 2.   If talin 1 and talin 2 have distinct functions, then my findings could be exploited to develop novel therapeutic strategies for anti-cancer drugs by targeting talins. Targeting scaffolding proteins such as talin, which have only protein-protein interaction domains, has been difficult. However, recent advances in computational chemistry have led to the development of several small molecules that target protein-protein interactions and are currently in clinical trials for cancer therapy798-802. Therefore, understanding the contributions of talin 1 and talin 2 to   208  tumor growth and cancer progression, identifying targetable (enzymatic) downstream effectors that mediate signalling downstream of talins in cancer cells, and determining whether targeting both talins or targeting only one talin isoform can restrict tumor growth and metastasis, would help identify new avenues to be exploited therapeutically.         209  Chapter 7: Overall conclusion and future directions  ECM-cell interactions are vital for tissue homeostasis and tumor cells take advantage of this for their survival and disseminations. Aberrant communication between the ECM and cancer cells, caused either by altered ECMs or cell-intrinsic factor such as changes in adhesome composition or adhesome signalling pathways, can contribute to cancer progression. Tumoral and stromal changes further aggravate each other by mechanisms yet to be understood. Some of these changes result in gene expression changes that are indicative of cancer progression, and therefore could serve as biomarkers to detect the switch from premalignant to malignant stages, or are drivers of tumor progression.  This thesis addresses the overall hypothesis that ECM-integrin interactions drive cancer progression in a manner that depends on adhesome signalling and mechanobiological tension. For two of the four result chapters, I used the mRNA expression levels of three cancer signature genes, Cyr61, MUC18 and TRPM1, as a readout for cancer progression. More aggressive cancers are associated with higher expression of Cyr61 and MUC18 and lower expression of TRPM1. Figure 49 provides a graphical summary of the main findings in this thesis organized by chapter.         210   Figure 49: ECM-cell dependent control of cancer progression.  Upregulation of Cyr61and MUC18 and downregulation of TRPM1 are gene expression changes associated with cancer progression. Panels A-D reiterate the graphical summaries of the indicated result chapter. (A) Chapter 3, (B) Chapter 4, (C) Chapter 5, (D) Chapter 6.       211  Summary of main findings:  Chapter 3: Cancer signature genes are regulated by integrin ligand density but not by cellular tension (Summary Figure 49A)   To address the first aim of this chapter I identified genes that were regulated when B16F1 cells adhered to increasing concentrations of FN. Based on qRT-PCR validation and previous implication in cancer progression, three genes were chosen as a readout to further investigate how aspects of cell adhesion, such as acto-myosin contractility and cellular tension, affect gene expression. The first main finding of chapter is that the three signature genes are strongly regulated by B16F1 cell adhesion to TC plastic, a non-compliant surface, and by adhesion to increased FN density in a manner similar to what occurs during cancer progression, specifically increased Cyr61 and MUC18 and decreased TRPM1 expression. Secondly, in B16F1 cells these adhesion-dependent changes in gene expression changes were not dependent on mechanical cytoskeletal forces that are generated during cell spreading on either non-compliant surfaces or surfaces with high ECM ligand density. Therefore, Cyr61, MUC18 and TRPM1 are genes whose mRNA levels are regulated by cell adhesion but not by cytoskeletal tension. This supports the idea that different aspects of cell-ECM interactions, such as adhesion, spreading, and cytoskeletal tension have distinct effects on gene expression.       212  Chapter 4: FN density and cell stretching result in overlapping but distinct patterns of gene regulation (Summary Figure 49B)   Allowing B16F1 cells to spread on increasing FN densities, and applying mechanical stretch to these cells, resulted in the upregulation of genes belonging to distinct pathways, with little overlap. Genes that were upregulated by increased FN densities were primarily annotated as members of adhesion, migration and ECM remodelling pathways as well as the Hippo pathway. In contrast, genes that were upregulated in response to mechanical stretch were annotated primarily as metabolic genes and genes of the HIF-1 pathway.    Chapter 5: The adhesome proteins talin and FAK regulate expression of the cancer signature genes (Summary Figure 49C)  I found that the three signature genes Cyr61, MUC18 and TRPM1 were regulated in a manner that is dependent on talin and FAK.  Knocking down talin mimics to some extent the effect of reduced adhesion or reduced expression of proteins responsible for integrin activation. This was reflected in reduced spreading area in double talin KD cells. Talin 1, rather than talin 2, regulated the expression of the cancer signature genes MUC18 and TRPM1. Knocking down FAK resulted in distinct cell morphologies from knocking down talin and differently regulation of the signature gens. FAK KD exhibited larger cell spreading areas on FN than control cells and upregulated Cyr61, MUC18 and TRPM1. This suggests that disturbing focal adhesion formation or changes in focal adhesion protein composition have distinct effects on gene expression.      213  Chapter 6: Talin 1 and talin 2 differently regulate B16F1 cell spreading, migration and tumor growth (Summary Figure 49D).  I found that talin 1 and talin 2 have overlapping but distinct roles in processes related to cancer progression. Simultaneously knocking down both isoforms of talin severely reduced cell spreading and cell migration, and caused the cells to adopt what appears to be an amoeboid mode of migration, instead of their typical mesenchymal migration. Furthermore, the loss of both talin isoforms resulted in impaired tumor growth after s.c. injection. The loss of only talin 1 also decreased cell spreading and impaired migration but did not affect tumor growth. Talin 2 KO cells exhibited impaired migration and substantially reduced tumor growth but the loss of only talin 2 increased cell spreading. Reconstituting single talin KO cells with the respective talin isoform reverted their spreading phenotypes. The severe spreading defect in double talin KO cells was partially rescued by expressing either talin 1 or talin 2, suggesting that both isoforms promote cell spreading in B16F1 cells. The partial rescue could be due to insufficient expression levels of either talin 1 or talin 2 in the reconstituted cells or due to both talin isoforms being necessary for cell spreading. To address this, we can clone out low, medium or high talin 1 or talin 2 expressing cells and correlate their expression levels with the degree of spreading or co-express talin 1 and talin 2 together in double KO cells. Taken together, my data suggests that talin 1 and talin 2 have non redundant roles in regulating cell morphology, cell migration, and tumor growth in B16F1 cells.   Changes in the composition, organization and stiffness of the ECM can contribute to cancer progression301. I showed that in vitro modelling of high surface stiffness by increasing the density of the ECM component FN regulated the mRNA levels of the three cancer signature   214  genes in a manner that is consistent with cancer progression (i.e. Cyr61 and MUC18 upregulation and TRPM1 downregulation). I also showed that the integrin adhesome components talin and FAK are important for the regulation of these three genes. Finally I showed that the two isoforms of the adhesion protein talin, talin 1 and talin 2, differently regulate expression of the signature genes, and have distinct effect on B16F1 cell growth, survival and migration. This suggests that targeting individual adhesome proteins has potential in the development of anti-cancer therapies. However, it also highlights challenges that could arise when targeting adhesome complexes. For example, highly similar proteins, such as talin 1 and talin 2, may be able to partially compensate for the loss of each other in processes that are important for tumor progression. Also, because ECM-cell adhesion and cell-cell adhesion complexes share subsets of proteins, targeting ECM-cell adhesomes could also affect cell-cell junctions and increase the ability of circulating tumor cells to extravasate into underlying tissue and form secondary growths573.  The drawback of a targeted study that relies on gene readout and the manipulation of single adhesome proteins is the limitation of the scale of manipulation and the scope of the readouts. The three curated cancer signature genes that I focused on may serve as biomarkers but they neither fully reflect the scope of the cellular response to altered cell-ECM conditions nor allow for the discovery of novel targets that could be exploited with therapeutic value. To address this, I took advantage of next generation sequencing techniques to perform an unbiased analysis of transcriptomes for two different models of ECM-dependent changes that cancer cells likely experience during cancer progression: adhesion to increased FN density as a model for increased ECM deposition and applied mechanical stretch as a model for the increased cytoskeletal tension that can result from altered ECM composition or crosslinking. With RNA-  215  seq as an unbiased readout, I showed that these two aspects of altered ECM-cell communication regulated different pathways that were associated with cancer malignancies and metastasis. Interestingly, the transcriptome analysis revealed that stretching primarily resulted in upregulation of genes associated with metabolism (Warburg effect), whereas increasing integrin ligand density upregulated genes that were primarily associated with cell migration, cell motility, and ECM remodelling. It would be of interest to investigate whether these changes are specific for B16F1 cells or also apply to non-malignant cells such as melanocytes or to cells derived from pre-cancerous skin lesions (i.e. dysplastic nevus). Recently it has been shown that normal and pre-malignant mammary epithelial cells differ in their response to increased mechanical tension in their ability to re-localize the mechanosensitive protein zyxin and their reorganization of actin structures (Poon et al. manuscript in preparation). Whether these differences between non-malignant, pre-malignant and malignant cells are also reflected in gene expression changes in response to stretch would be an interesting point to address. The identification of tension-sensitive gene expression changes that are specific for pre-malignant or malignant cells could lead to the discovery of new targets for cancer therapy.   In addition to transcriptome sequencing, advances in proteomics, phosphoproteomics, and signalling pathway analysis as unbiased readouts will provide additional insight into how ECM-cell communication alters global cellular behavior and contributes to malignancies. For example, Robertson et al.380 adapted proteomics techniques to specifically analyze the phosphoproteome of integrin adhesomes. This approach is superior to immunostaining selected adhesome proteins and could provide unbiased information about adhesome composition changes in response to different ECMs. Similarly, using mass spec-based proteomics or   216  CyTOF803, 804 for the analysis of single cells would generate large datasets providing a more comprehensive analysis of adhesome signalling than current immunoblotting approaches can provide. In addition, RNA-seq transcriptome profiling could be complemented by ChIP-sequencing in order to identify changes in promotor activity in response to altered ECM-cell communication. Figure 50 outlines a potential workflow for how these large data approaches could be integrated in order to use multiple levels of information to find new targets for cancer treatment. To understand the link between ECM-cell communication and cancer progression, I would integrate data that is generated on three levels: adhesome composition, signalling, and transcriptome analysis (Figure 50A). This data could then serve as an input for network analysis (Figure 50B) to identify important parameters that determine the outcome of altered ECM-cell interactions. This could then be correlated with known cancer progression markers (Figure 50C) such as gene expression and morphological changes. Changes in input parameters, such as different ECM compositions, increasing stiffness or ECM ligand density or modulating intrinsic cell changes such as knocking out or mutating adhesome proteins (Figure 50D), can then validate whether an identified key parameter is essential and may identify novel targets of therapeutic value (Figure 50E). In the past decades a number of mouse models have been developed to study melanoma development and test new anti-melanoma therapies. These mouse models of melanoma are based on mutations that are associated with human melanoma, in particular mutations in BRAF, KRAS, NRAS, p16INK4A, and TP53805. Whether the ECM-regulated genes identified in this study modulate cancer progression in the context of these driver mutations can now be rapidly assessed using CRISPR/Cas to alter these genes in these existing mouse models of melanoma.   217   Although there are multiple genetically-engineered mouse models of cancer, as well as xenograft models and UV light- and chemical-induced models of cancer806-808 it is still a challenge to recapitulate natural tumor progression from proliferation to invasion and metastasis. It is particularly challenging to study the contribution of ECM-tumor cell interactions to cancer progression in vivo. However, several recently developed mouse models, such as transgenic mice with collagen-dense mammary tissue279 or mice with induced aberrant LOX-mediated collagen crossliking275, focus on establishing a causal link between ECM changes and tumor progression. Elucidating how ECM ligand density and cellular tension affect cell behaviour and gene expression in vitro may help further develop and refine these mouse models.   Figure 50: An unbiased systems biology approach to identify novel therapeutic targets for cancer progression associated with alterations in cell-ECM communication.  (A) Inputs are the three types of ‘omics’ data that can be obtained for interactions between cells and ECM. (B) Network analysis function determines dependencies between multidimensional input parameters. (C) These dependencies can be correlated with cancer progression markers and generate testable hypotheses (D), with the goal of identifying novel therapeutic targets (E).      218  During the last two decades, scientists have discovered the importance of the ECM for tissue homeostasis and have recognized that changes in ECM-cell communication can result in pathogenesis. We are beginning to understand the underlying mechanisms by which altered ECM, characterized by dysplasia, is a hallmark of and contributor to, cancer progression. Recent advances in proteomics and next generation sequencing tools such as RNA-seq will allow us now to take a new, less biased look at the complex network that governs adhesome-dependent regulation of global cellular functions and may help us to identify new strategies for cancer therapies. My thesis work represents a step in this direction.   219  References 1. http://www.who.int/cancer/en/. (September 30th 2015). 2. Advisory, C.C.S.s. & 2015, C.o.C.S.C.C.S. (2015). 3. Vogelstein, B. & Kinzler, K.W. The Multistep Nature of Cancer. Trends in Genetics 9, 138-141 (1993). 4. Farber, E. The Multistep Nature of Cancer Development. Cancer Research 44, 4217-4223 (1984). 5. Hanahan, D. & Weinberg, R.A. The hallmarks of cancer. Cell 100, 57-70 (2000). 6. Hanahan, D. & Weinberg, R.A. Hallmarks of cancer: the next generation. Cell 144, 646-74 (2011). 7. Foulds, L. The Experimental Study of Tumor Progression - a Review. Cancer Research 14, 327-339 (1954). 8. Renan, M.J. How many mutations are required for tumorigenesis? Implications from human cancer data. Mol Carcinog 7, 139-46 (1993). 9. Clark, W.H. Tumor Progression and the Nature of Cancer. British Journal of Cancer 64, 631-644 (1991). 10. Miller, A.J. & Mihm, M.C., Jr. Melanoma. N Engl J Med 355, 51-65 (2006). 11. Siegel, R., Naishadham, D. & Jemal, A. Cancer statistics, 2012. CA Cancer J Clin 62, 10-29 (2012). 12. Atlanta. American Cancer Society Cancer Facts & Figures (2013). 13. Bertolotto, C. Melanoma: from melanocyte to genetic alterations and clinical options. Scientifica (Cairo) 2013, 635203 (2013). 14. Daniotti, M., Oggionni, M., Ranzani, T., Vallacchi, V., Campi, V., Di Stasi, D., Torre, G.D., Perrone, F., Luoni, C., Suardi, S., Frattini, M., Pilotti, S., Anichini, A., Tragni, G., Parmiani, G., Pierotti, M.A. & Rodolfo, M. BRAF alterations are associated with complex mutational profiles in malignant melanoma. Oncogene 23, 5968-77 (2004). 15. Chapman, P.B., Hauschild, A., Robert, C., Haanen, J.B., Ascierto, P., Larkin, J., Dummer, R., Garbe, C., Testori, A., Maio, M., Hogg, D., Lorigan, P., Lebbe, C., Jouary, T., Schadendorf, D., Ribas, A., O'Day, S.J., Sosman, J.A., Kirkwood, J.M., Eggermont, A.M.M., Dreno, B., Nolop, K., Li, J., Nelson, B., Hou, J., Lee, R.J., Flaherty, K.T., McArthur, G.A. & Grp, B.-S. Improved Survival with Vemurafenib in Melanoma with BRAF V600E Mutation. New England Journal of Medicine 364, 2507-2516 (2011).   220  16. Balch, C.M., Gershenwald, J.E., Soong, S.J., Thompson, J.F., Atkins, M.B., Byrd, D.R., Buzaid, A.C., Cochran, A.J., Coit, D.G., Ding, S., Eggermont, A.M., Flaherty, K.T., Gimotty, P.A., Kirkwood, J.M., McMasters, K.M., Mihm, M.C., Jr., Morton, D.L., Ross, M.I., Sober, A.J. & Sondak, V.K. Final version of 2009 AJCC melanoma staging and classification. J Clin Oncol 27, 6199-206 (2009). 17. Girotti, M.R., Saturno, G., Lorigan, P. & Marais, R. No longer an untreatable disease: How targeted and immunotherapies have changed the management of melanoma patients. Molecular Oncology 8, 1140-1158 (2014). 18. Dolan, D.E. & Gupta, S. PD-1 Pathway Inhibitors: Changing the Landscape of Cancer Immunotherapy. Cancer Control 21, 231-237 (2014). 19. Baksh, K. & Weber, J. Immune Checkpoint Protein Inhibition for Cancer: Preclinical Justification for CTLA-4 and PD-1 Blockade and New Combinations. Semin Oncol 42, 363-377 (2015). 20. Flaherty, K.T., Puzanov, I., Kim, K.B., Ribas, A., McArthur, G.A., Sosman, J.A., O'Dwyer, P.J., Lee, R.J., Grippo, J.F., Nolop, K. & Chapman, P.B. Inhibition of Mutated, Activated BRAF in Metastatic Melanoma. New England Journal of Medicine 363, 809-819 (2010). 21. Sosman, J.A., Kim, K.B., Schuchter, L., Gonzalez, R., Pavlick, A.C., Weber, J.S., McArthur, G.A., Hutson, T.E., Moschos, S.J., Flaherty, K.T., Hersey, P., Kefford, R., Lawrence, D., Puzanov, I., Lewis, K.D., Amaravadi, R.K., Chmielowski, B., Lawrence, H.J., Shyr, Y., Ye, F., Li, J., Nolop, K.B., Lee, R.J., Joe, A.K. & Ribas, A. Survival in BRAF V600-Mutant Advanced Melanoma Treated with Vemurafenib. New England Journal of Medicine 366, 707-714 (2012). 22. Sherr, C.J. Principles of tumor suppression. Cell 116, 235-46 (2004). 23. Schwab, M. Genetic principles of tumor suppression. Biochim Biophys Acta 989, 49-64 (1989). 24. Shay, J.W., Wright, W.E. & Werbin, H. Defining the molecular mechanisms of human cell immortalization. Biochim Biophys Acta 1072, 1-7 (1991). 25. Harley, C.B., Futcher, A.B. & Greider, C.W. Telomeres shorten during ageing of human fibroblasts. Nature 345, 458-60 (1990). 26. Kim, N.W., Piatyszek, M.A., Prowse, K.R., Harley, C.B., West, M.D., Ho, P.L., Coviello, G.M., Wright, W.E., Weinrich, S.L. & Shay, J.W. Specific association of human telomerase activity with immortal cells and cancer. Science 266, 2011-5 (1994). 27. Harley, C.B., Vaziri, H., Counter, C.M. & Allsopp, R.C. The telomere hypothesis of cellular aging. Exp Gerontol 27, 375-82 (1992).   221  28. Counter, C.M., Avilion, A.A., LeFeuvre, C.E., Stewart, N.G., Greider, C.W., Harley, C.B. & Bacchetti, S. Telomere shortening associated with chromosome instability is arrested in immortal cells which express telomerase activity. EMBO J 11, 1921-9 (1992). 29. Wyllie, A.H., Kerr, J.F. & Currie, A.R. Cell death: the significance of apoptosis. Int Rev Cytol 68, 251-306 (1980). 30. Rock, K.L. & Kono, H. The inflammatory response to cell death. Annu Rev Pathol 3, 99-126 (2008). 31. Lowe, S.W., Cepero, E. & Evan, G. Intrinsic tumour suppression. Nature 432, 307-15 (2004). 32. Adams, J.M. & Cory, S. The Bcl-2 apoptotic switch in cancer development and therapy. Oncogene 26, 1324-37 (2007). 33. Evan, G. & Littlewood, T. A matter of life and cell death. Science 281, 1317-22 (1998). 34. Ashkenazi, A. & Dixit, V.M. Apoptosis control by death and decoy receptors. Curr Opin Cell Biol 11, 255-60 (1999). 35. Thornberry, N.A. & Lazebnik, Y. Caspases: enemies within. Science 281, 1312-6 (1998). 36. McIlwain, D.R., Berger, T. & Mak, T.W. Caspase functions in cell death and disease. Cold Spring Harb Perspect Biol 5, a008656 (2013). 37. Slee, E.A., Adrain, C. & Martin, S.J. Executioner caspase-3, -6, and -7 perform distinct, non-redundant roles during the demolition phase of apoptosis. J Biol Chem 276, 7320-6 (2001). 38. Herbert, S.P. & Stainier, D.Y. Molecular control of endothelial cell behaviour during blood vessel morphogenesis. Nat Rev Mol Cell Biol 12, 551-64 (2011). 39. Tonnesen, M.G., Feng, X. & Clark, R.A. Angiogenesis in wound healing. J Investig Dermatol Symp Proc 5, 40-6 (2000). 40. Demir, R., Yaba, A. & Huppertz, B. Vasculogenesis and angiogenesis in the endometrium during menstrual cycle and implantation. Acta Histochem 112, 203-14 (2010). 41. Baeriswyl, V. & Christofori, G. The angiogenic switch in carcinogenesis. Semin Cancer Biol 19, 329-37 (2009). 42. Folkman, J. Role of angiogenesis in tumor growth and metastasis. Semin Oncol 29, 15-8 (2002). 43. Boocock, C.A., Charnock-Jones, D.S., Sharkey, A.M., McLaren, J., Barker, P.J., Wright, K.A., Twentyman, P.R. & Smith, S.K. Expression of vascular endothelial growth factor   222  and its receptors flt and KDR in ovarian carcinoma. J Natl Cancer Inst 87, 506-16 (1995). 44. Itakura, J., Ishiwata, T., Shen, B., Kornmann, M. & Korc, M. Concomitant over-expression of vascular endothelial growth factor and its receptors in pancreatic cancer. Int J Cancer 85, 27-34 (2000). 45. Moon, W.S., Park, H.S., Yu, K.H., Jang, K.Y., Kang, M.J., Park, H. & Tarnawski, A.S. Expression of angiopoietin 1, 2 and their common receptor Tie2 in human gastric carcinoma: In implication for angiogenesis. Journal of Korean Medical Science 21, 272-278 (2006). 46. Olive, K.P., Jacobetz, M.A., Davidson, C.J., Gopinathan, A., McIntyre, D., Honess, D., Madhu, B., Goldgraben, M.A., Caldwell, M.E., Allard, D., Frese, K.K., Denicola, G., Feig, C., Combs, C., Winter, S.P., Ireland-Zecchini, H., Reichelt, S., Howat, W.J., Chang, A., Dhara, M., Wang, L., Ruckert, F., Grutzmann, R., Pilarsky, C., Izeradjene, K., Hingorani, S.R., Huang, P., Davies, S.E., Plunkett, W., Egorin, M., Hruban, R.H., Whitebread, N., McGovern, K., Adams, J., Iacobuzio-Donahue, C., Griffiths, J. & Tuveson, D.A. Inhibition of Hedgehog signaling enhances delivery of chemotherapy in a mouse model of pancreatic cancer. Science 324, 1457-61 (2009). 47. Zee, Y.K., O'Connor, J.P., Parker, G.J., Jackson, A., Clamp, A.R., Taylor, M.B., Clarke, N.W. & Jayson, G.C. Imaging angiogenesis of genitourinary tumors. Nat Rev Urol 7, 69-82 (2010). 48. Turner, H.E., Harris, A.L., Melmed, S. & Wass, J.A. Angiogenesis in endocrine tumors. Endocr Rev 24, 600-32 (2003). 49. Nagy, J.A., Chang, S.H., Shih, S.C., Dvorak, A.M. & Dvorak, H.F. Heterogeneity of the tumor vasculature. Semin Thromb Hemost 36, 321-31 (2010). 50. Langley, R.R. & Fidler, I.J. Tumor cell-organ microenvironment interactions in the pathogenesis of cancer metastasis. Endocr Rev 28, 297-321 (2007). 51. Mehlen, P. & Puisieux, A. Metastasis: a question of life or death. Nat Rev Cancer 6, 449-58 (2006). 52. Monteiro, J. & Fodde, R. Cancer stemness and metastasis: therapeutic consequences and perspectives. Eur J Cancer 46, 1198-203 (2010). 53. Nguyen, D.X., Bos, P.D. & Massague, J. Metastasis: from dissemination to organ-specific colonization. Nat Rev Cancer 9, 274-84 (2009). 54. Paget, S. The distribution of secondary growths in cancer of the breast. 1889. Cancer Metastasis Rev 8, 98-101 (1989). 55. Langley, R.R. & Fidler, I.J. The seed and soil hypothesis revisited--the role of tumor-stroma interactions in metastasis to different organs. Int J Cancer 128, 2527-35 (2011).   223  56. Erwing, J. Neoplastic diseases. WB Saunders 3rd ed. (1928). 57. Talmadge, J.E. & Fidler, I.J. AACR centennial series: the biology of cancer metastasis: historical perspective. Cancer Res 70, 5649-69 (2010). 58. Fidler, I.J. The pathogenesis of cancer metastasis: the 'seed and soil' hypothesis revisited. Nat Rev Cancer 3, 453-8 (2003). 59. Shibue, T. & Weinberg, R.A. Metastatic colonization: settlement, adaptation and propagation of tumor cells in a foreign tissue environment. Semin Cancer Biol 21, 99-106 (2011). 60. Fidler, I.J. Metastasis: guantitative analysis of distribution and fate of tumor embolilabeled with 125 I-5-iodo-2'-deoxyuridine. J Natl Cancer Inst 45, 773-82 (1970). 61. Chambers, A.F., MacDonald, I.C., Schmidt, E.E., Koop, S., Morris, V.L., Khokha, R. & Groom, A.C. Steps in tumor metastasis: new concepts from intravital videomicroscopy. Cancer Metastasis Rev 14, 279-301 (1995). 62. Klymkowsky, M.W. & Savagner, P. Epithelial-mesenchymal transition: a cancer researcher's conceptual friend and foe. Am J Pathol 174, 1588-93 (2009). 63. Polyak, K. & Weinberg, R.A. Transitions between epithelial and mesenchymal states: acquisition of malignant and stem cell traits. Nat Rev Cancer 9, 265-73 (2009). 64. Thiery, J.P., Acloque, H., Huang, R.Y. & Nieto, M.A. Epithelial-mesenchymal transitions in development and disease. Cell 139, 871-90 (2009). 65. Yilmaz, M. & Christofori, G. EMT, the cytoskeleton, and cancer cell invasion. Cancer Metastasis Rev 28, 15-33 (2009). 66. Schmalhofer, O., Brabletz, S. & Brabletz, T. E-cadherin, beta-catenin, and ZEB1 in malignant progression of cancer. Cancer Metastasis Rev 28, 151-66 (2009). 67. Yang, J. & Weinberg, R.A. Epithelial-mesenchymal transition: At the crossroads of development and tumor metastasis. Developmental Cell 14, 818-829 (2008). 68. Baum, B. & Georgiou, M. Dynamics of adherens junctions in epithelial establishment, maintenance, and remodeling. J Cell Biol 192, 907-17 (2011). 69. Takeichi, M. Dynamic contacts: rearranging adherens junctions to drive epithelial remodelling. Nat Rev Mol Cell Biol 15, 397-410 (2014). 70. Yonemura, S. Cadherin-actin interactions at adherens junctions. Current Opinion in Cell Biology 23, 515-522 (2011). 71. Bershadsky, A. Magic touch: how does cell-cell adhesion trigger actin assembly? Trends in Cell Biology 14, 589-593 (2004).   224  72. Yonemura, S., Wada, Y., Watanabe, T., Nagafuchi, A. & Shibata, M. alpha-Catenin as a tension transducer that induces adherens junction development. Nature Cell Biology 12, 533-U35 (2010). 73. Baum, B. & Perrimon, N. Spatial control of the actin cytoskeleton in Drosophila epithelial cells. Nature Cell Biology 3, 883-890 (2001). 74. Perez-Moreno, M., Jamora, C. & Fuchs, E. Sticky business: Orchestrating cellular signals at adherens junctions. Cell 112, 535-548 (2003). 75. Drees, F., Pokutta, S., Yamada, S., Nelson, W.J. & Weis, W.I. alpha-catenin is a molecular switch that binds E-cadherin-beta-catenin and regulates actin-filament assembly. Cell 123, 903-915 (2005). 76. Lecuit, T. & Yap, A.S. E-cadherin junctions as active mechanical integrators in tissue dynamics. Nat Cell Biol 17, 533-539 (2015). 77. Kausalya, P.J., Phua, D.C.Y. & Hunziker, W. Association of ARVCF with zonula occludens (ZO)-1 and ZO-2: Binding to PDZ-domain proteins and cell-cell adhesion regulate plasma membrane and nuclear localization of ARVCF. Molecular Biology of the Cell 15, 5503-5515 (2004). 78. Whitehead, J., Vignjevic, D., Futterer, C., Beaurepaire, E., Robine, S. & Farge, E. Mechanical factors activate beta-catenin-dependent oncogene expression in APC(1638N/+) mouse colon. Hfsp Journal 2, 286-294 (2008). 79. Petersen, O.W., Ronnovjessen, L., Howlett, A.R. & Bissell, M.J. Interaction with Basement-Membrane Serves to Rapidly Distinguish Growth and Differentiation Pattern of Normal and Malignant Human Breast Epithelial-Cells. Proceedings of the National Academy of Sciences of the United States of America 89, 9064-9068 (1992). 80. Nejsum, L.N. & Nelson, W.J. A molecular mechanism directly linking E-cadherin adhesion to initiation of epithelial cell surface polarity. Journal of Cell Biology 178, 323-335 (2007). 81. Rodriguezboulan, E. & Nelson, W.J. Morphogenesis of the Polarized Epithelial-Cell Phenotype. Science 245, 718-725 (1989). 82. Tepass, U., Theres, C. & Knust, E. Crumbs Encodes an Egf-Like Protein Expressed on Apical Membranes of Drosophila Epithelial-Cells and Required for Organization of Epithelia. Cell 61, 787-799 (1990). 83. Tepass, U. & Knust, E. Crumbs and Stardust Act in a Genetic Pathway That Controls the Organization of Epithelia in Drosophila-Melanogaster. Developmental Biology 159, 311-326 (1993).   225  84. Bhat, M.A., Izaddoost, S., Lu, Y., Cho, K.O., Choi, K.W. & Bellen, H.J. Discs lost, a novel multi-PDZ domain protein, establishes and maintains epithelial polarity. Cell 96, 833-845 (1999). 85. Kemphues, K.J., Priess, J.R., Morton, D.G. & Cheng, N. Identification of Genes Required for Cytoplasmic Localization in Early C-Elegans Embryos. Cell 52, 311-320 (1988). 86. Tabuse, Y., Izumi, Y., Piano, F., Kemphues, K.J., Miwa, J. & Ohno, S. Atypical protein kinase C cooperates with PAR-3 to establish embryonic polarity in Caenorhabditis elegans. Development 125, 3607-3614 (1998). 87. Petronczki, M. & Knoblich, J.A. DmPAR-6 directs epithelial polarity and asymmetric cell division of neuroblasts in Drosophila. Nature Cell Biology 3, 43-49 (2001). 88. Bilder, D., Li, M. & Perrimon, N. Cooperative regulation of cell polarity and growth by Drosophila tumor suppressors. Science 289, 113-116 (2000). 89. Bilder, D. Epithelial polarity and proliferation control: links from the Drosophila neoplastic tumor suppressors. Genes & Development 18, 1909-1925 (2004). 90. Jeanes, A., Gottardi, C.J. & Yap, A.S. Cadherins and cancer: how does cadherin dysfunction promote tumor progression? Oncogene 27, 6920-6929 (2008). 91. Navarro, C., Nola, S., Audebert, S., Santoni, M.J., Arsanto, J.P., Ginestier, C., Marchetto, S., Jacquemier, J., Isnardon, D., Le Bivic, A., Birnbaum, D. & Borg, J.P. Junctional recruitment of mammalian Scribble relies on E-cadherin engagement. Oncogene 24, 4330-4339 (2005). 92. Qin, Y., Capaldo, C., Gumbiner, B.M. & Macara, I.G. The mammalian Scribble polarity protein regulates epithelial cell adhesion and migration through E-cadherin. Journal of Cell Biology 171, 1061-1071 (2005). 93. Royer, C. & Lu, X. Epithelial cell polarity: a major gatekeeper against cancer? Cell Death and Differentiation 18, 1470-1477 (2011). 94. Berx, G. & van Roy, F. Involvement of members of the cadherin superfamily in cancer. Cold Spring Harb Perspect Biol 1, a003129 (2009). 95. Cavallaro, U. & Christofori, G. Cell adhesion and signalling by cadherins and Ig-CAMs in cancer. Nat Rev Cancer 4, 118-32 (2004). 96. Araki, K., Shimura, T., Suzuki, H., Tsutsumi, S., Wada, W., Yajima, T., Kobayahi, T., Kubo, N. & Kuwano, H. E/N-cadherin switch mediates cancer progression via TGF-beta-induced epithelial-to-mesenchymal transition in extrahepatic cholangiocarcinoma. Br J Cancer 105, 1885-93 (2011).   226  97. Klingener, M., Chavali, M., Singh, J., McMillan, N., Coomes, A., Dempsey, P.J., Chen, E.I. & Aguirre, A. N-Cadherin Promotes Recruitment and Migration of Neural Progenitor Cells from the SVZ Neural Stem Cell Niche into Demyelinated Lesions. J Neurosci 34, 9590-606 (2014). 98. Hatta, K., Takagi, S., Fujisawa, H. & Takeichi, M. Spatial and temporal expression pattern of N-cadherin cell adhesion molecules correlated with morphogenetic processes of chicken embryos. Dev Biol 120, 215-27 (1987). 99. Tomita, K., van Bokhoven, A., van Leenders, G.J.L.H., Ruijter, E.T.G., Jansen, C.F.J., Bussemakers, M.J.G. & Schalken, J.A. Cadherin switching in human prostate cancer progression. Cancer Research 60, 3650-3654 (2000). 100. Li, G. & Herlyn, M. Dynamics of intercellular communication during melanoma development. Molecular Medicine Today 6, 163-169 (2000). 101. Hazan, R.B., Phillips, G.R., Qiao, R.F., Norton, L. & Aaronson, S.A. Exogenous expression of N-cadherin in breast cancer cells induces cell migration, invasion, and metastasis. Journal of Cell Biology 148, 779-790 (2000). 102. Islam, S., Carey, T.E., Wolf, G.T., Wheelock, M.J. & Johnson, K.R. Expression of N-cadherin by human squamous carcinoma cells induces a scattered fibroblastic phenotype with disrupted cell-cell adhesion. Journal of Cell Biology 135, 1643-1654 (1996). 103. Nieman, M.T., Prudoff, R.S., Johnson, K.R. & Wheelock, M.J. N-cadherin promotes motility in human breast cancer cells regardless of their E-cadherin expression. Journal of Cell Biology 147, 631-643 (1999). 104. Peinado, H., Marin, F., Cubillo, E., Stark, H.J., Fusenig, N., Nieto, M.A. & Cano, A. Snail and E47 repressors of E-cadherin induce distinct invasive and angiogenic properties in vivo. Journal of Cell Science 117, 2827-2839 (2004). 105. Qin, Q., Xu, Y., He, T., Qin, C.L. & Xu, J.M. Normal and disease-related biological functions of Twist1 and underlying molecular mechanisms. Cell Research 22, 90-106 (2012). 106. Hao, L., Ha, J.R., Kuzel, P., Garcia, E. & Persad, S. Cadherin switch from E- to N-cadherin in melanoma progression is regulated by the PI3K/PTEN pathway through Twist and Snail. British Journal of Dermatology 166, 1184-1197 (2012). 107. Negrini, S., Gorgoulis, V.G. & Halazonetis, T.D. Genomic instability - an evolving hallmark of cancer. Nature Reviews Molecular Cell Biology 11, 220-228 (2010). 108. Colotta, F., Allavena, P., Sica, A., Garlanda, C. & Mantovani, A. Cancer-related inflammation, the seventh hallmark of cancer: links to genetic instability. Carcinogenesis 30, 1073-1081 (2009).   227  109. Warburg, O. Note on the metabolism of tumours. Biochemische Zeitschrift 228, 257-258 (1930). 110. Warburg, O., Wind, F. & Negelein, E. The metabolism of tumors in the body. Journal of General Physiology 8, 519-530 (1927). 111. Warburg, O. Respiratory Impairment in Cancer Cells. Science 124, 269-270 (1956). 112. Warburg, O. Origin of Cancer Cells. Science 123, 309-314 (1956). 113. DeBerardinis, R.J., Lum, J.J., Hatzivassiliou, G. & Thompson, C.B. The biology of cancer: Metabolic reprogramming fuels cell growth and proliferation. Cell Metabolism 7, 11-20 (2008). 114. Jones, R.G. & Thompson, C.B. Tumor suppressors and cell metabolism: a recipe for cancer growth. Genes & Development 23, 537-548 (2009). 115. Koppenol, W.H., Bounds, P.L. & Dang, C.V. Otto Warburg's contributions to current concepts of cancer metabolism. Nat Rev Cancer 11, 325-37 (2011). 116. Vander Heiden, M.G., Cantley, L.C. & Thompson, C.B. Understanding the Warburg effect: the metabolic requirements of cell proliferation. Science 324, 1029-33 (2009). 117. Feron, O. Pyruvate into lactate and back: from the Warburg effect to symbiotic energy fuel exchange in cancer cells. Radiother Oncol 92, 329-33 (2009). 118. Semenza, G.L. Tumor metabolism: cancer cells give and take lactate. J Clin Invest 118, 3835-7 (2008). 119. Kennedy, K.M. & Dewhirst, M.W. Tumor metabolism of lactate: the influence and therapeutic potential for MCT and CD147 regulation. Future Oncol 6, 127-48 (2010). 120. Adam, J., Yang, M., Soga, T. & Pollard, P.J. Rare insights into cancer biology. Oncogene 33, 2547-56 (2014). 121. Dang, C.V. Links between metabolism and cancer. Genes & Development 26, 877-890 (2012). 122. Jang, M., Kim, S.S. & Lee, J. Cancer cell metabolism: implications for therapeutic targets. Experimental and Molecular Medicine 45 (2013). 123. Gordan, J.D., Lal, P., Dondeti, V.R., Letrero, R., Parekh, K.N., Oquendo, C.E., Greenberg, R.A., Flaherty, K.T., Rathmell, W.K., Keith, B., Simon, M.C. & Nathanson, K.L. HIF-alpha effects on c-Myc distinguish two subtypes of sporadic VHL-deficient clear cell renal carcinoma. Cancer Cell 14, 435-46 (2008). 124. Schulze, A. & Harris, A.L. How cancer metabolism is tuned for proliferation and vulnerable to disruption. Nature 491, 364-73 (2012).   228  125. Chaika, N.V., Yu, F., Purohit, V., Mehla, K., Lazenby, A.J., DiMaio, D., Anderson, J.M., Yeh, J.J., Johnson, K.R., Hollingsworth, M.A. & Singh, P.K. Differential expression of metabolic genes in tumor and stromal components of primary and metastatic loci in pancreatic adenocarcinoma. PLoS One 7, e32996 (2012). 126. Hu, J., Locasale, J.W., Bielas, J.H., O'Sullivan, J., Sheahan, K., Cantley, L.C., Vander Heiden, M.G. & Vitkup, D. Heterogeneity of tumor-induced gene expression changes in the human metabolic network. Nat Biotechnol 31, 522-9 (2013). 127. Kang, B.H., Plescia, J., Dohi, T., Rosa, J., Doxsey, S.J. & Altieri, D.C. Regulation of tumor cell mitochondrial homeostasis by an organelle-specific Hsp90 chaperone network. Cell 131, 257-70 (2007). 128. Caino, M.C., Chae, Y.C., Vaira, V., Ferrero, S., Nosotti, M., Martin, N.M., Weeraratna, A., O'Connell, M., Jernigan, D., Fatatis, A., Languino, L.R., Bosari, S. & Altieri, D.C. Metabolic stress regulates cytoskeletal dynamics and metastasis of cancer cells. J Clin Invest 123, 2907-20 (2013). 129. Ehrlich, P. The collected papers of Paul Ehrlich. Pergamon Press Volume II. (1957). 130. Teng, M.W., Swann, J.B., Koebel, C.M., Schreiber, R.D. & Smyth, M.J. Immune-mediated dormancy: an equilibrium with cancer. J Leukoc Biol 84, 988-93 (2008). 131. Kim, R., Emi, M. & Tanabe, K. Cancer immunoediting from immune surveillance to immune escape. Immunology 121, 1-14 (2007). 132. Smyth, M.J., Crowe, N.Y. & Godfrey, D.I. NK cells and NKT cells collaborate in host protection from methylcholanthrene-induced fibrosarcoma. International Immunology 13, 459-463 (2001). 133. Smyth, M.J., Thia, K.Y., Street, S.E., MacGregor, D., Godfrey, D.I. & Trapani, J.A. Perforin-mediated cytotoxicity is critical for surveillance of spontaneous lymphoma. J Exp Med 192, 755-60 (2000). 134. Smyth, M.J., Thia, K.Y., Street, S.E., Cretney, E., Trapani, J.A., Taniguchi, M., Kawano, T., Pelikan, S.B., Crowe, N.Y. & Godfrey, D.I. Differential tumor surveillance by natural killer (NK) and NKT cells. J Exp Med 191, 661-8 (2000). 135. Shankaran, V., Ikeda, H., Bruce, A.T., White, J.M., Swanson, P.E., Old, L.J. & Schreiber, R.D. IFNgamma and lymphocytes prevent primary tumour development and shape tumour immunogenicity. Nature 410, 1107-11 (2001). 136. Dunn, G.P., Old, L.J. & Schreiber, R.D. The immunobiology of cancer immunosurveillance and immunoediting. Immunity 21, 137-48 (2004). 137. Spiotto, M.T., Yu, P., Rowley, D.A., Nishimura, M.I., Meredith, S.C., Gajewski, T.F., Fu, Y.X. & Schreiber, H. Increasing tumor antigen expression overcomes "ignorance" to   229  solid tumors via crosspresentation by bone marrow-derived stromal cells. Immunity 17, 737-747 (2002). 138. Webb, S.D., Sherratt, J.A. & Fish, R.G. Cells behaving badly: a theoretical model for the Fas/FasL system in tumour immunology. Mathematical Biosciences 179, 113-129 (2002). 139. Doubrovina, E.S., Doubrovin, M.M., Vider, E., Sisson, R.B., O'Reilly, R.J., Dupont, B. & Vyas, Y.M. Evasion from NK cell immunity by MHC class I chain-related molecules expressing colon adenocarcinoma. Journal of Immunology 171, 6891-6899 (2003). 140. Grivennikov, S.I. & Karin, M. Inflammatory cytokines in cancer: tumour necrosis factor and interleukin 6 take the stage. Annals of the Rheumatic Diseases 70, I104-I108 (2011). 141. Hansemann, D.v. Ueber asymmetrische Zellteilung in Epithelkrebsen und deren biologische Bedeutung. Virchows Arch. Patholog. Anat. 199, 299-326 (1890). 142. Boveri, T. Zur frage der enstehung maligner tumoren Gustav Fischer Verlag, Jena (1914). 143. Lengauer, C., Kinzler, K.W. & Vogelstein, B. Genetic instability in colorectal cancers. Nature 386, 623-627 (1997). 144. Nowell, P.C. Citation Classic - the Clonal Evolution of Tumor-Cell Populations. Current Contents/Life Sciences, 19-19 (1988). 145. Winge, O. Zytologische untersuchungen uber die natur maligner tumoren. II. Teerkarzinome bei mausen. Z. Zellforsch. Mikrosk. Anat. 10, 683–735 (1930). 146. Sjoblom, T., Jones, S., Wood, L.D., Parsons, D.W., Lin, J., Barber, T.D., Mandelker, D., Leary, R.J., Ptak, J., Silliman, N., Szabo, S., Buckhaults, P., Farrell, C., Meeh, P., Markowitz, S.D., Willis, J., Dawson, D., Willson, J.K.V., Gazdar, A.F., Hartigan, J., Wu, L., Liu, C.S., Parmigiani, G., Park, B.H., Bachman, K.E., Papadopoulos, N., Vogelstein, B., Kinzler, K.W. & Velculescu, V.E. The consensus coding sequences of human breast and colorectal cancers. Science 314, 268-274 (2006). 147. Wood, L.D., Parsons, D.W., Jones, S., Lin, J., Sjoblom, T., Leary, R.J., Shen, D., Boca, S.M., Barber, T., Ptak, J., Silliman, N., Szabo, S., Dezso, Z., Ustyanksky, V., Nikolskaya, T., Nikolsky, Y., Karchin, R., Wilson, P.A., Kaminker, J.S., Zhang, Z.M., Croshaw, R., Willis, J., Dawson, D., Shipitsin, M., Willson, J.K.V., Sukumar, S., Polyak, K., Park, B.H., Pethiyagoda, C.L., Pant, P.V.K., Ballinger, D.G., Sparks, A.B., Hartigan, J., Smith, D.R., Suh, E., Papadopoulos, N., Buckhaults, P., Markowitz, S.D., Parmigiani, G., Kinzler, K.W., Velculescu, V.E. & Vogelstein, B. The genomic landscapes of human breast and colorectal cancers. Science 318, 1108-1113 (2007). 148. Jones, S., Zhang, X.S., Parsons, D.W., Lin, J.C.H., Leary, R.J., Angenendt, P., Mankoo, P., Carter, H., Kamiyama, H., Jimeno, A., Hong, S.M., Fu, B.J., Lin, M.T., Calhoun, E.S., Kamiyama, M., Walter, K., Nikolskaya, T., Nikolsky, Y., Hartigan, J., Smith, D.R., Hidalgo, M., Leach, S.D., Klein, A.P., Jaffee, E.M., Goggins, M., Maitra, A., Iacobuzio-  230  Donahue, C., Eshleman, J.R., Kern, S.E., Hruban, R.H., Karchin, R., Papadopoulos, N., Parmigiani, G., Vogelstein, B., Velculescu, V.E. & Kinzler, K.W. Core signaling pathways in human pancreatic cancers revealed by global genomic analyses. Science 321, 1801-1806 (2008). 149. Jones, S., Zhang, X., Parsons, D.W., Lin, J.C., Leary, R.J., Angenendt, P., Mankoo, P., Carter, H., Kamiyama, H., Jimeno, A., Hong, S.M., Fu, B., Lin, M.T., Calhoun, E.S., Kamiyama, M., Walter, K., Nikolskaya, T., Nikolsky, Y., Hartigan, J., Smith, D.R., Hidalgo, M., Leach, S.D., Klein, A.P., Jaffee, E.M., Goggins, M., Maitra, A., Iacobuzio-Donahue, C., Eshleman, J.R., Kern, S.E., Hruban, R.H., Karchin, R., Papadopoulos, N., Parmigiani, G., Vogelstein, B., Velculescu, V.E. & Kinzler, K.W. Core signaling pathways in human pancreatic cancers revealed by global genomic analyses. Science 321, 1801-6 (2008). 150. Chin, L., Meyerson, M., Aldape, K., Bigner, D., Mikkelsen, T., VandenBerg, S., Kahn, A., Penny, R., Ferguson, M.L., Gerhard, D.S., Getz, G., Brennan, C., Taylor, B.S., Winckler, W., Park, P., Ladanyi, M., Hoadley, K.A., Verhaak, R.G.W., Hayes, D.N., Spellman, P.T., Absher, D., Weir, B.A., Ding, L., Wheeler, D., Lawrence, M.S., Cibulskis, K., Mardis, E., Zhang, J.H., Wilson, R.K., Donehower, L., Wheeler, D.A., Purdom, E., Wallis, J., Laird, P.W., Herman, J.G., Schuebel, K.E., Weisenberger, D.J., Baylin, S.B., Schultz, N., Yao, J., Wiedemeyer, R., Weinstein, J., Sander, C., Gibbs, R.A., Gray, J., Kucherlapati, R., Lander, E.S., Myers, R.M., Perou, C.M., McLendon, R., Friedman, A., Van Meir, E.G., Brat, D.J., Mastrogianakis, G.M., Olson, J.J., Lehman, N., Yung, W.K.A., Bogler, O., Berger, M., Prados, M., Muzny, D., Morgan, M., Scherer, S., Sabo, A., Nazareth, L., Lewis, L., Hall, O., Zhu, Y.M., Ren, Y.R., Alvi, O., Yao, J.Q., Hawes, A., Jhangiani, S., Fowler, G., San Lucas, A., Kovar, C., Cree, A., Dinh, H., Santibanez, J., Joshi, V., Gonzalez-Garay, M.L., Miller, C.A., Milosavljevic, A., Sougnez, C., Fennell, T., Mahan, S., Wilkinson, J., Ziaugra, L., Onofrio, R., Bloom, T., Nicol, R., Ardlie, K., Baldwin, J., Gabriel, S., Fulton, R.S., McLellan, M.D., Larson, D.E., Shi, X.Q., Abbott, R., Fulton, L., Chen, K., Koboldt, D.C., Wendl, M.C., Meyer, R., Tang, Y.Z., Lin, L., Osborne, J.R., Dunford-Shore, B.H., Miner, T.L., Delehaunty, K., Markovic, C., Swift, G., Courtney, W., Pohl, C., Abbott, S., Hawkins, A., Leong, S., Haipek, C., Schmidt, H., Wiechert, M., Vickery, T., Scott, S., Dooling, D.J., Chinwalla, A., Weinstock, G.M., O'Kelly, M., Robinson, J., Alexe, G., Beroukhim, R., Carter, S., Chiang, D., Gould, J., Gupta, S., Korn, J., Mermel, C., Mesirov, J., Monti, S., Nguyen, H., Parkin, M., Reich, M., Stransky, N., Garraway, L., Golub, T., Protopopov, A., Perna, I., Aronson, S., Sathiamoorthy, N., Ren, G., Kim, H., Kong, S.K., Xiao, Y.H., Kohane, I.S., Seidman, J., Cope, L., Pan, F., Van Den Berg, D., Van Neste, L., Yi, J.M., Li, J.Z., Southwick, A., Brady, S., Aggarwal, A., Chung, T., Sherlock, G., Brooks, J.D., Jakkula, L.R., Lapuk, A.V., Marr, H., Dorton, S., Choi, Y.G., Han, J., Ray, A., Wang, V., Durinck, S., Robinson, M., Wang, N.J., Vranizan, K., Peng, V., Van Name, E., Fontenay, G.V., Ngai, J., Conboy, J.G., Parvin, B., Feiler, H.S., Speed, T.P., Socci, N.D., Olshen, A., Lash, A., Reva, B., Antipin, Y., Stukalov, A., Gross, B., Cerami, E., Wang, W.Q., Qin, L.X., Seshan, V.E., Villafania, L., Cavatore, M., Borsu, L., Viale, A., Gerald, W., Topal, M.D., Qi, Y., Balu, S., Shi, Y., Wu, G., Bittner, M., Shelton, T., Lenkiewicz, E., Morris, S., Beasley, D., Sanders, S., Sfeir, R., Chen, J., Nassau, D., Feng, L., Hickey, E., Schaefer, C., Madhavan, S., Buetow, K., Barker, A., Vockley, J., Compton, C., Vaught,   231  J., Fielding, P., Collins, F., Good, P., Guyer, M., Ozenberger, B., Peterson, J., Thomson, E., Network, C.G.A.R., Sites, T.S., Ctr, G.S., Ctr, C.G.C. & Teams, P. Comprehensive genomic characterization defines human glioblastoma genes and core pathways. Nature 455, 1061-1068 (2008). 151. Ding, L., Getz, G., Wheeler, D.A., Mardis, E.R., McLellan, M.D., Cibulskis, K., Sougnez, C., Greulich, H., Muzny, D.M., Morgan, M.B., Fulton, L., Fulton, R.S., Zhang, Q.Y., Wendl, M.C., Lawrence, M.S., Larson, D.E., Chen, K., Dooling, D.J., Sabo, A., Hawes, A.C., Shen, H., Jhangiani, S.N., Lewis, L.R., Hall, O., Zhu, Y.M., Mathew, T., Ren, Y.R., Yao, J.Q., Scherer, S.E., Clerc, K., Metcalf, G.A., Ng, B., Milosavljevic, A., Gonzalez-Garay, M.L., Osborne, J.R., Meyer, R., Shi, X.Q., Tang, Y.Z., Koboldt, D.C., Lin, L., Abbott, R., Miner, T.L., Pohl, C., Fewell, G., Haipek, C., Schmidt, H., Dunford-Shore, B.H., Kraja, A., Crosby, S.D., Sawyer, C.S., Vickery, T., Sander, S., Robinson, J., Winckler, W., Baldwin, J., Chirieac, L.R., Dutt, A., Fennell, T., Hanna, M., Johnson, B.E., Onofrio, R.C., Thomas, R.K., Tonon, G., Weir, B.A., Zhao, X.J., Ziaugra, L., Zody, M.C., Giordano, T., Orringer, M.B., Roth, J.A., Spitz, M.R., Wistuba, I.I., Ozenberger, B., Good, P.J., Chang, A.C., Beer, D.G., Watson, M.A., Ladanyi, M., Broderick, S., Yoshizawa, A., Travis, W.D., Pao, W., Province, M.A., Weinstock, G.M., Varmus, H.E., Gabriel, S.B., Lander, E.S., Gibbs, R.A., Meyerson, M. & Wilson, R.K. Somatic mutations affect key pathways in lung adenocarcinoma. Nature 455, 1069-1075 (2008). 152. Kinzler, K.W. & Vogelstein, B. Cancer-susceptibility genes - Gatekeepers and caretakers. Nature 386, 761-& (1997). 153. Barnes, D.E. & Lindahl, T. Repair and genetic consequences of endogenous DNA base damage in mammalian cells. Annual Review of Genetics 38, 445-476 (2004). 154. Zhang, J.R. & Powell, S.N. The role of the BRCA1 tumor suppressor in DNA double-strand break repair. Molecular Cancer Research 3, 531-539 (2005). 155. Fishel, R., Lescoe, M.K., Rao, M.R.S., Copeland, N.G., Jenkins, N.A., Garber, J., Kane, M. & Kolodner, R. The Human Mutator Gene Homolog Msh2 and Its Association with Hereditary Nonpolyposis Colon-Cancer. Cell 75, 1027-1038 (1993). 156. Leach, F.S., Nicolaides, N.C., Papadopoulos, N., Liu, B., Jen, J., Parsons, R., Peltomaki, P., Sistonen, P., Aaltonen, L.A., Nystromlahti, M., Guan, X.Y., Zhang, J., Meltzer, P.S., Yu, J.W., Kao, F.T., Chen, D.J., Cerosaletti, K.M., Fournier, R.E.K., Todd, S., Lewis, T., Leach, R.J., Naylor, S.L., Weissenbach, J., Mecklin, J.P., Jarvinen, H., Petersen, G.M., Hamilton, S.R., Green, J., Jass, J., Watson, P., Lynch, H.T., Trent, J.M., Delachapelle, A., Kinzler, K.W. & Vogelstein, B. Mutations of a Muts Homolog in Hereditary Nonpolyposis Colorectal-Cancer. Cell 75, 1215-1225 (1993). 157. Kennedy, R.D. & D'andrea, A.D. DNA repair pathways in clinical practice: Lessons from pediatric cancer susceptibility syndromes. Journal of Clinical Oncology 24, 3799-3808 (2006). 158. Cleaver, J.E. Cancer in xeroderma pigmentosum and related disorders of DNA repair. Nature Reviews Cancer 5, 564-573 (2005).   232  159. Deininger, P. Genetic instability in cancer: caretaker and gatekeeper genes. Ochsner J 1, 206-9 (1999). 160. Loeb, L.A. Mutator Phenotype May Be Required for Multistage Carcinogenesis. Cancer Res 51, 3075-3079 (1991). 161. Loeb, L.A. A mutator phenotype in cancer. Cancer Res 61, 3230-3239 (2001). 162. Wang, Z.H., Cummins, J.M., Shen, D., Cahill, D.P., Jallepalli, P.V., Wang, T.L., Parsons, D.W., Traverso, G., Awad, M., Silliman, N., Ptak, J., Szabo, S., Willson, J.K.V., Markowitz, S.D., Goldberg, M.L., Karess, R., Kinzler, K.W., Vogelstein, B., Velculescu, V.E. & Lengauer, C. Three classes of genes mutated in colorectal cancers with chromosomal instability. Cancer Res 64, 2998-3001 (2004). 163. Negrini, S., Gorgoulis, V.G. & Halazonetis, T.D. Genomic instability--an evolving hallmark of cancer. Nat Rev Mol Cell Biol 11, 220-8 (2010). 164. Maya-Mendoza, A., Ostrakova, J., Kosar, M., Hall, A., Duskova, P., Mistrik, M., Merchut-Maya, J.M., Hodny, Z., Bartkova, J., Christensen, C. & Bartek, J. Myc and Ras oncogenes engage different energy metabolism programs and evoke distinct patterns of oxidative and DNA replication stress. Mol Oncol 9, 601-16 (2015). 165. Burrell, R.A., McClelland, S.E., Endesfelder, D., Groth, P., Weller, M.C., Shaikh, N., Domingo, E., Kanu, N., Dewhurst, S.M., Gronroos, E., Chew, S.K., Rowan, A.J., Schenk, A., Sheffer, M., Howell, M., Kschischo, M., Behrens, A., Helleday, T., Bartek, J., Tomlinson, I.P. & Swanton, C. Replication stress links structural and numerical cancer chromosomal instability. Nature 494, 492-6 (2013). 166. Luo, J., Emanuele, M.J., Li, D., Creighton, C.J., Schlabach, M.R., Westbrook, T.F., Wong, K.K. & Elledge, S.J. A genome-wide RNAi screen identifies multiple synthetic lethal interactions with the Ras oncogene. Cell 137, 835-48 (2009). 167. Jemal, A., Siegel, R., Xu, J.Q. & Ward, E. Cancer Statistics, 2010. Ca-a Cancer Journal for Clinicians 60, 277-300 (2010). 168. Medzhitov, R. Origin and physiological roles of inflammation. Nature 454, 428-435 (2008). 169. Kuper, H., Adami, H.O. & Trichopoulos, D. Infections as a major preventable cause of human cancer (Reprinted from Journal of Internal Medicine, vol 248, pg 171-183, 2000). Journal of Internal Medicine 249, 61-73 (2001). 170. Kuper, H., Adami, H.O. & Trichopoulos, D. Infections as a major preventable cause of human cancer. Journal of Internal Medicine 248, 171-183 (2000). 171. Virchow, R. An address on the value of pathological experiments. Br. Med. J. 2, 198-203 (1889).   233  172. Michaud, M., Martins, I., Sukkurwala, A.Q., Adjemian, S., Ma, Y.T., Pellegatti, P., Shen, S.S., Kepp, O., Scoazec, M., Mignot, G., Rello-Varona, S., Tailler, M., Menger, L., Vacchelli, E., Galluzzi, L., Ghiringhelli, F., di Virgilio, F., Zitvogel, L. & Kroemer, G. Autophagy-Dependent Anticancer Immune Responses Induced by Chemotherapeutic Agents in Mice. Science 334, 1573-1577 (2011). 173. Ehrlich, P. Ueber den jetzigen stand der karzinomforschung. Ned Tijdschr Geneeskd. 5, 73–290 (1909). 174. Burnet, M. Cancer: a biological approach. III. Viruses associated with neoplastic conditions. IV. Practical applications. Br Med J 1, 841-7 (1957). 175. Burnet, M. Cancer; a biological approach. I. The processes of control. Br Med J 1, 779-86 (1957). 176. Dunn, G.P., Bruce, A.T., Ikeda, H., Old, L.J. & Schreiber, R.D. Cancer immunoediting: from immunosurveillance to tumor escape. Nat Immunol 3, 991-8 (2002). 177. DeNardo, D., Andreu, P. & Coussens, L.M. Interactions between lymphocytes and myeloid cells regulate pro- versus anti-tumor immunity. Cancer and Metastasis Reviews 29, 309-316 (2010). 178. Grivennikov, S.I., Greten, F.R. & Karin, M. Immunity, Inflammation, and Cancer. Cell 140, 883-899 (2010). 179. Hussain, S.P. & Harris, C.C. Inflammation and cancer: An ancient link with novel potentials. International Journal of Cancer 121, 2373-2380 (2007). 180. Itzkowitz, S.H. & Yio, X. Inflammation and cancer IV. Colorectal cancer in inflammatory bowel disease: the role of inflammation. Am J Physiol Gastrointest Liver Physiol 287, G7-17 (2004). 181. Whitcomb, D.C. Inflammation and Cancer V. Chronic pancreatitis and pancreatic cancer. Am J Physiol Gastrointest Liver Physiol 287, G315-9 (2004). 182. Baumgarten, S.C. & Frasor, J. Minireview: Inflammation: an instigator of more aggressive estrogen receptor (ER) positive breast cancers. Mol Endocrinol 26, 360-71 (2012). 183. Fouad, T.M., Kogawa, T., Reuben, J.M. & Ueno, N.T. The role of inflammation in inflammatory breast cancer. Adv Exp Med Biol 816, 53-73 (2014). 184. Karnoub, A.E. & Weinberg, R.A. Chemokine networks and breast cancer metastasis. Breast Dis 26, 75-85 (2006). 185. Qian, B.Z. & Pollard, J.W. Macrophage Diversity Enhances Tumor Progression and Metastasis. Cell 141, 39-51 (2010).   234  186. Lu, P.F., Weaver, V.M. & Werb, Z. The extracellular matrix: A dynamic niche in cancer progression. Journal of Cell Biology 196, 395-406 (2012). 187. Costa-Silva, B., Aiello, N.M., Ocean, A.J., Singh, S., Zhang, H., Thakur, B.K., Becker, A., Hoshino, A., Mark, M.T., Molina, H., Xiang, J., Zhang, T., Theilen, T.M., Garcia-Santos, G., Williams, C., Ararso, Y., Huang, Y., Rodrigues, G., Shen, T.L., Labori, K.J., Lothe, I.M., Kure, E.H., Hernandez, J., Doussot, A., Ebbesen, S.H., Grandgenett, P.M., Hollingsworth, M.A., Jain, M., Mallya, K., Batra, S.K., Jarnagin, W.R., Schwartz, R.E., Matei, I., Peinado, H., Stanger, B.Z., Bromberg, J. & Lyden, D. Pancreatic cancer exosomes initiate pre-metastatic niche formation in the liver. Nat Cell Biol 17, 816-26 (2015). 188. Sleeman, J.P. The metastatic niche and stromal progression. Cancer Metastasis Rev 31, 429-40 (2012). 189. Brigati, C., Noonan, D.M., Albini, A. & Benelli, R. Tumors and inflammatory infiltrates: friends or foes? Clin Exp Metastasis 19, 247-58 (2002). 190. Tsung, K., Dolan, J.P., Tsung, Y.L. & Norton, J.A. Macrophages as effector cells in interleukin 12-induced T cell-dependent tumor rejection. Cancer Research 62, 5069-5075 (2002). 191. Schoppmann, S.F., Birner, P., Stockl, J., Kalt, R., Ullrich, R., Caucig, C., Kriehuber, E., Nagy, K., Alitalo, K. & Kerjaschki, D. Tumor-associated macrophages express lymphatic endothelial growth factors and are related to peritumoral lymphangiogenesis. American Journal of Pathology 161, 947-956 (2002). 192. Coussens, L.M. & Werb, Z. Inflammation and cancer. Nature 420, 860-867 (2002). 193. Grivennikov, S.I. & Karin, M. Inflammation and oncogenesis: a vicious connection. Current Opinion in Genetics & Development 20, 65-71 (2010). 194. Sullivan, N.J., Sasser, A.K., Axel, A.E., Vesuna, F., Raman, V., Ramirez, N., Oberyszyn, T.M. & Hall, B.M. Interleukin-6 induces an epithelial-mesenchymal transition phenotype in human breast cancer cells. Oncogene 28, 2940-2947 (2009). 195. Voronov, E., Shouval, D.S., Krelin, Y., Cagnano, E., Benharroch, D., Iwakura, Y., Dinarello, C.A. & Apte, R.N. IL-1 is required for tumor invasiveness and angiogenesis. Proceedings of the National Academy of Sciences of the United States of America 100, 2645-2650 (2003). 196. Nguyen, D.X., Bos, P.D. & Massague, J. Metastasis: from dissemination to organ-specific colonization. Nature Reviews Cancer 9, 274-U65 (2009). 197. Condeelis, J. & Pollard, J.W. Macrophages: Obligate partners for tumor cell migration, invasion, and metastasis. Cell 124, 263-266 (2006).   235  198. Peinado, H., Rafii, S. & Lyden, D. Inflammation joins the "niche". Cancer Cell 14, 347-9 (2008). 199. Huang, S. & Ingber, D.E. Cell tension, matrix mechanics, and cancer development. Cancer Cell 8, 175-176 (2005). 200. Turner, C.H. & Pavalko, F.M. Mechanotransduction and functional response of the skeleton to physical stress: the mechanisms and mechanics of bone adaptation. J Orthop Sci 3, 346-55 (1998). 201. Mullender, M., El Haj, A.J., Yang, Y., van Duin, M.A., Burger, E.H. & Klein-Nulend, J. Mechanotransduction of bone cells in vitro: mechanobiology of bone tissue. Med Biol Eng Comput 42, 14-21 (2004). 202. Owens, G.K. Role of mechanical strain in regulation of differentiation of vascular smooth muscle cells. Circ Res 79, 1054-5 (1996). 203. Williams, B. Mechanical influences on vascular smooth muscle cell function. J Hypertens 16, 1921-9 (1998). 204. Kshitiz, Park, J., Kim, P., Helen, W., Engler, A.J., Levchenko, A. & Kim, D.H. Control of stem cell fate and function by engineering physical microenvironments. Integr Biol (Camb) 4, 1008-18 (2012). 205. Metallo, C.M., Vodyanik, M.A., de Pablo, J.J., Slukvin, II & Palecek, S.P. The response of human embryonic stem cell-derived endothelial cells to shear stress. Biotechnol Bioeng 100, 830-7 (2008). 206. Altman, G.H., Horan, R.L., Martin, I., Farhadi, J., Stark, P.R., Volloch, V., Richmond, J.C., Vunjak-Novakovic, G. & Kaplan, D.L. Cell differentiation by mechanical stress. FASEB J 16, 270-2 (2002). 207. Yim, E.K.F., Pang, S.W. & Leong, K.W. Synthetic nanostructures inducing differentiation of human mesenchymal stem cells into neuronal lineage. Experimental Cell Research 313, 1820-1829 (2007). 208. Geiger, B., Spatz, J.P. & Bershadsky, A.D. Environmental sensing through focal adhesions. Nature Reviews Molecular Cell Biology 10, 21-33 (2009). 209. Arulmoli, J., Pathak, M.M., McDonnell, L.P., Nourse, J.L., Tombola, F., Earthman, J.C. & Flanagan, L.A. Static stretch affects neural stem cell differentiation in an extracellular matrix-dependent manner. Scientific Reports 5 (2015). 210. Engler, A.J., Sen, S., Sweeney, H.L. & Discher, D.E. Matrix elasticity directs stem cell lineage specification. Cell 126, 677-689 (2006). 211. Assoian, R.K. & Klein, E.A. Growth control by intracellular tension and extracellular stiffness. Trends in Cell Biology 18, 347-352 (2008).   236  212. Yeung, T., Georges, P.C., Flanagan, L.A., Marg, B., Ortiz, M., Funaki, M., Zahir, N., Ming, W.Y., Weaver, V. & Janmey, P.A. Effects of substrate stiffness on cell morphology, cytoskeletal structure, and adhesion. Cell Motility and the Cytoskeleton 60, 24-34 (2005). 213. Ingber, D.E. Cellular mechanotransduction: putting all the pieces together again. Faseb Journal 20, 811-827 (2006). 214. Orr, A.W., Helmke, B.P., Blackman, B.R. & Schwartz, M.A. Mechanisms of mechanotransduction. Developmental Cell 10, 11-20 (2006). 215. Janostiak, R., Pataki, A.C., Brabek, J. & Rosel, D. Mechanosensors in integrin signaling: the emerging role of p130Cas. Eur J Cell Biol 93, 445-54 (2014). 216. Sawada, Y., Tamada, M., Dubin-Thaler, B.J., Cherniavskaya, O., Sakai, R., Tanaka, S. & Sheetz, M.P. Force sensing by mechanical extension of the Src family kinase substrate p130Cas. Cell 127, 1015-26 (2006). 217. Yao, M., Goult, B.T., Chen, H., Cong, P., Sheetz, M.P. & Yan, J. Mechanical activation of vinculin binding to talin locks talin in an unfolded conformation. Sci Rep 4, 4610 (2014). 218. Margadant, F., Chew, L.L., Hu, X., Yu, H., Bate, N., Zhang, X. & Sheetz, M. Mechanotransduction In Vivo by Repeated Talin Stretch-Relaxation Events Depends upon Vinculin. Plos Biology 9 (2011). 219. Wells, R.G. & Discher, D.E. Matrix elasticity, cytoskeletal tension, and TGF-beta: the insoluble and soluble meet. Sci Signal 1, pe13 (2008). 220. Resnick, N., Yahav, H., Shay-Salit, A., Shushy, M., Schubert, S., Zilberman, L.C. & Wofovitz, E. Fluid shear stress and the vascular endothelium: for better and for worse. Prog Biophys Mol Biol 81, 177-99 (2003). 221. Wang, J.H., Goldschmidt-Clermont, P., Wille, J. & Yin, F.C. Specificity of endothelial cell reorientation in response to cyclic mechanical stretching. J Biomech 34, 1563-72 (2001). 222. Yang, J.H., Sakamoto, H., Xu, E.C. & Lee, R.T. Biomechanical regulation of human monocyte/macrophage molecular function. Am J Pathol 156, 1797-804 (2000). 223. Jaqaman, K. & Grinstein, S. Regulation from within: the cytoskeleton in transmembrane signaling. Trends in Cell Biology 22, 515-526 (2012). 224. Beadle, C., Assanah, M.C., Monzo, P., Vallee, R., Rosenfeld, S.S. & Canoll, P. The role of myosin II in glioma invasion of the brain. Mol Biol Cell 19, 3357-68 (2008). 225. Ingber, D.E. Cellular tensegrity: defining new rules of biological design that govern the cytoskeleton. J Cell Sci 104 ( Pt 3), 613-27 (1993).   237  226. Jaalouk, D.E. & Lammerding, J. Mechanotransduction gone awry. Nature Reviews Molecular Cell Biology 10, 63-73 (2009). 227. Barry, S.P., Davidson, S.M. & Townsend, P.A. Molecular regulation of cardiac hypertrophy. Int J Biochem Cell Biol 40, 2023-39 (2008). 228. Heydemann, A. & McNally, E.M. Consequences of disrupting the dystrophin-sarcoglycan complex in cardiac and skeletal myopathy. Trends Cardiovasc Med 17, 55-9 (2007). 229. Vollrath, M.A., Kwan, K.Y. & Corey, D.P. The micromachinery of mechanotransduction in hair cells. Annu Rev Neurosci 30, 339-65 (2007). 230. Cheng, C., Tempel, D., van Haperen, R., van der Baan, A., Grosveld, F., Daemen, M.J., Krams, R. & de Crom, R. Atherosclerotic lesion size and vulnerability are determined by patterns of fluid shear stress. Circulation 113, 2744-53 (2006). 231. Garcia-Cardena, G., Comander, J., Anderson, K.R., Blackman, B.R. & Gimbrone, M.A., Jr. Biomechanical activation of vascular endothelium as a determinant of its functional phenotype. Proc Natl Acad Sci U S A 98, 4478-85 (2001). 232. Hebner, C., Weaver, V.M. & Debnath, J. Modeling morphogenesis and oncogenesis in three-dimensional breast epithelial cultures. Annual Review of Pathology-Mechanisms of Disease 3, 313-339 (2008). 233. Tse, J.M., Cheng, G., Tyrrell, J.A., Wilcox-Adelman, S.A., Boucher, Y., Jain, R.K. & Munn, L.L. Mechanical compression drives cancer cells toward invasive phenotype. Proceedings of the National Academy of Sciences of the United States of America 109, 911-916 (2012). 234. Haessler, U., Teo, J.C.M., Foretay, D., Renaud, P. & Swartz, M.A. Migration dynamics of breast cancer cells in a tunable 3D interstitial flow chamber. Integrative Biology 4, 401-409 (2012). 235. Polacheck, W.J., Charest, J.L. & Kamm, R.D. Interstitial flow influences direction of tumor cell migration through competing mechanisms. Proceedings of the National Academy of Sciences of the United States of America 108, 11115-11120 (2011). 236. Tien, J., Truslow, J.G. & Nelson, C.M. Modulation of Invasive Phenotype by Interstitial Pressure-Driven Convection in Aggregates of Human Breast Cancer Cells. Plos One 7 (2012). 237. Suresh, S. Biomechanics and biophysics of cancer cells. Acta Biomaterialia 3, 413-438 (2007). 238. Vallenius, T. Actin stress fibre subtypes in mesenchymal-migrating cells. Open Biology 3 (2013).   238  239. Vicente-Manzanares, M., Ma, X.F., Adelstein, R.S. & Horwitz, A.R. Non-muscle myosin II takes centre stage in cell adhesion and migration. Nature Reviews Molecular Cell Biology 10, 778-790 (2009). 240. Kuo, J.C., Han, X., Yates, J.R., 3rd & Waterman, C.M. Isolation of focal adhesion proteins for biochemical and proteomic analysis. Methods Mol Biol 757, 297-323 (2012). 241. Fritz, G., Just, I. & Kaina, B. Rho GTPases are over-expressed in human tumors. Int J Cancer 81, 682-7 (1999). 242. Paszek, M.J., Zahir, N., Johnson, K.R., Lakins, J.N., Rozenberg, G.I., Gefen, A., Reinhart-King, C.A., Margulies, S.S., Dembo, M., Boettiger, D., Hammer, D.A. & Weaver, V.M. Tensional homeostasis and the malignant phenotype. Cancer Cell 8, 241-54 (2005). 243. Kalluri, R. & Weinberg, R.A. The basics of epithelial-mesenchymal transition. J Clin Invest 119, 1420-8 (2009). 244. Thiery, J.P. Epithelial-mesenchymal transitions in tumour progression. Nat Rev Cancer 2, 442-54 (2002). 245. Dupont, S., Morsut, L., Aragona, M., Enzo, E., Giulitti, S., Cordenonsi, M., Zanconato, F., Le Digabel, J., Forcato, M., Bicciato, S., Elvassore, N. & Piccolo, S. Role of YAP/TAZ in mechanotransduction. Nature 474, 179-83 (2011). 246. McCubrey, J.A., Steelman, L.S., Chappell, W.H., Abrams, S.L., Wong, E.W.T., Chang, F., Lehmann, B., Terrian, D.M., Milella, M., Tafuri, A., Stivala, F., Libra, M., Basecke, J., Evangelisti, C., Martelli, A.M. & Franklin, R.A. Roles of the Raf/MEK/ERK pathway in cell growth, malignant transformation and drug resistance. Biochimica Et Biophysica Acta-Molecular Cell Research 1773, 1263-1284 (2007). 247. Bissell, M.J. & Radisky, D. Putting tumours in context. Nat Rev Cancer 1, 46-54 (2001). 248. Wiseman, B.S. & Werb, Z. Stromal effects on mammary gland development and breast cancer. Science 296, 1046-9 (2002). 249. Bissell, M.J. & Labarge, M.A. Context, tissue plasticity, and cancer: are tumor stem cells also regulated by the microenvironment? Cancer Cell 7, 17-23 (2005). 250. Lu, P., Weaver, V.M. & Werb, Z. The extracellular matrix: a dynamic niche in cancer progression. J Cell Biol 196, 395-406 (2012). 251. Pickup, M.W., Mouw, J.K. & Weaver, V.M. The extracellular matrix modulates the hallmarks of cancer. EMBO Rep 15, 1243-53 (2014). 252. Mammoto, T. & Ingber, D.E. Mechanical control of tissue and organ development. Development 137, 1407-20 (2010).   239  253. Ozbek, S., Balasubramanian, P.G., Chiquet-Ehrismann, R., Tucker, R.P. & Adams, J.C. The evolution of extracellular matrix. Mol Biol Cell 21, 4300-5 (2010). 254. Hynes, R.O. The extracellular matrix: not just pretty fibrils. Science 326, 1216-9 (2009). 255. Egeblad, M., Rasch, M.G. & Weaver, V.M. Dynamic interplay between the collagen scaffold and tumor evolution. Curr Opin Cell Biol 22, 697-706 (2010). 256. Hiraki, Y., Kono, T., Sato, M., Shukunami, C. & Kondo, J. Inhibition of DNA synthesis and tube morphogenesis of cultured vascular endothelial cells by chondromodulin-1. Febs Letters 415, 321-324 (1997). 257. Hiraki, Y., Inoue, H., Iyama, K., Kamizono, A., Ochiai, M., Shukunami, C., Iijima, S., Suzuki, F. & Kondo, J. Identification of chondromodulin I as a novel endothelial cell growth inhibitor - Purification and its localization in the avascular zone of epiphysical cartilage. Journal of Biological Chemistry 272, 32419-32426 (1997). 258. Termine, J.D., Belcourt, A.B., Conn, K.M. & Kleinman, H.K. Mineral and Collagen-Binding Proteins of Fetal Calf Bone. Journal of Biological Chemistry 256, 403-408 (1981). 259. Bonnans, C., Chou, J. & Werb, Z. Remodelling the extracellular matrix in development and disease. Nat Rev Mol Cell Biol 15, 786-801 (2014). 260. Frantz, C., Stewart, K.M. & Weaver, V.M. The extracellular matrix at a glance. Journal of Cell Science 123, 4195-4200 (2010). 261. Poschl, E., Schlotzer-Schrehardt, U., Brachvogel, B., Saito, K., Ninomiya, Y. & Mayer, U. Collagen IV is essential for basement membrane stability but dispensable for initiation of its assembly during early development. Development 131, 1619-28 (2004). 262. Miner, J.H., Li, C., Mudd, J.L., Go, G. & Sutherland, A.E. Compositional and structural requirements for laminin and basement membranes during mouse embryo implantation and gastrulation. Development 131, 2247-56 (2004). 263. Badylak, S.F., Freytes, D.O. & Gilbert, T.W. Extracellular matrix as a biological scaffold material: Structure and function. Acta Biomater 5, 1-13 (2009). 264. Royer, C. & Lu, X. Epithelial cell polarity: a major gatekeeper against cancer? Cell Death Differ 18, 1470-7 (2011). 265. Pandol, S., Edderkaoui, M., Gukovsky, I., Lugea, A. & Gukovskaya, A. Desmoplasia of pancreatic ductal adenocarcinoma. Clin Gastroenterol Hepatol 7, S44-7 (2009). 266. Erler, J.T., Bennewith, K.L., Nicolau, M., Dornhofer, N., Kong, C., Le, Q.T., Chi, J.T.A., Jeffrey, S.S. & Giaccia, A.J. Lysyl oxidase is essential for hypoxia-induced metastasis. Nature 440, 1222-1226 (2006).   240  267. Slattery, M.L., John, E., Torres-Mejia, G., Stern, M., Lundgreen, A., Hines, L., Giuliano, A., Baumgartner, K., Herrick, J. & Wolff, R.K. Matrix Metalloproteinase Genes Are Associated with Breast Cancer Risk and Survival: The Breast Cancer Health Disparities Study. Plos One 8 (2013). 268. McCawley, L.J. & Matrisian, L.M. Matrix metalloproteinases: they're not just for matrix anymore! Curr Opin Cell Biol 13, 534-40 (2001). 269. Egeblad, M. & Werb, Z. New functions for the matrix metalloproteinases in cancer progression. Nat Rev Cancer 2, 161-74 (2002). 270. Botta, G.P., Reginato, M.J., Reichert, M., Rustgi, A.K. & Lelkes, P.I. Constitutive K-RasG12D activation of ERK2 specifically regulates 3D invasion of human pancreatic cancer cells via MMP-1. Mol Cancer Res 10, 183-96 (2012). 271. Zeng, Z.S., Cohen, A.M. & Guillem, J.G. Loss of basement membrane type IV collagen is associated with increased expression of metalloproteinases 2 and 9 (MMP-2 and MMP-9) during human colorectal tumorigenesis. Carcinogenesis 20, 749-755 (1999). 272. Margulis, A., Nocka, K.H., Wood, N.L., Wolf, S.F., Goldman, S.J. & Kasaian, M.T. MMP dependence of fibroblast contraction and collagen production induced by human mast cell activation in a three-dimensional collagen lattice. American Journal of Physiology-Lung Cellular and Molecular Physiology 296, L236-L247 (2009). 273. Daniels, J.T., Cambrey, A.D., Occleston, N.L., Garrett, Q., Tarnuzzer, R.W., Schultz, G.S. & Khaw, P.T. Matrix metalloproteinase inhibition modulates fibroblast-mediated matrix contraction and collagen production in vitro. Investigative Ophthalmology & Visual Science 44, 1104-1110 (2003). 274. Peyrol, S., Raccurt, M., Gerard, F., Gleyzal, C., Grimaud, J.A. & Sommer, P. Lysyl oxidase gene expression in the stromal reaction to in situ and invasive ductal breast carcinoma. American Journal of Pathology 150, 497-507 (1997). 275. Levental, K.R., Yu, H.M., Kass, L., Lakins, J.N., Egeblad, M., Erler, J.T., Fong, S.F.T., Csiszar, K., Giaccia, A., Weninger, W., Yamauchi, M., Gasser, D.L. & Weaver, V.M. Matrix Crosslinking Forces Tumor Progression by Enhancing Integrin Signaling. Cell 139, 891-906 (2009). 276. Friedl, P. & Wolf, K. Tube travel: The role of proteases in individual and collective a cancer cell invasion. Cancer Research 68, 7247-7249 (2008). 277. Provenzano, P.P., Eliceiri, K.W., Campbell, J.M., Inman, D.R., White, J.G. & Keely, P.J. Collagen reorganization at the tumor-stromal interface facilitates local invasion. Bmc Medicine 4 (2006). 278. Condeelis, J. & Segall, J.E. Intravital imaging of cell movement in tumours. Nature Reviews Cancer 3, 921-930 (2003).   241  279. Provenzano, P.P., Inman, D.R., Eliceiri, K.W., Knittel, J.G., Yan, L., Rueden, C.T., White, J.G. & Keely, P.J. Collagen density promotes mammary tumor initiation and progression. BMC Med 6, 11 (2008). 280. Provenzano, P.P. & Keely, P.J. Mechanical signaling through the cytoskeleton regulates cell proliferation by coordinated focal adhesion and Rho GTPase signaling. Journal of Cell Science 124, 1195-1205 (2011). 281. Chambard, J.C., Lefloch, R., Pouyssegur, J. & Lenormand, P. ERK implication in cell cycle regulation. Biochimica Et Biophysica Acta-Molecular Cell Research 1773, 1299-1310 (2007). 282. Bae, Y.H., Mui, K.L., Hsu, B.Y., Liu, S.L., Cretu, A., Razinia, Z., Xu, T.N., Pure, E. & Assoian, R.K. A FAK-Cas-Rac-Lamellipodin Signaling Module Transduces Extracellular Matrix Stiffness into Mechanosensitive Cell Cycling. Science Signaling 7 (2014). 283. Wells, R.G. The role of matrix stiffness in regulating cell behavior. Hepatology 47, 1394-1400 (2008). 284. Tilghman, R.W., Cowan, C.R., Mih, J.D., Koryakina, Y., Gioeli, D., Slack-Davis, J.K., Blackman, B.R., Tschumperlin, D.J. & Parsons, J.T. Matrix rigidity regulates cancer cell growth and cellular phenotype. PLoS One 5, e12905 (2010). 285. Schrader, J., Gordon-Walker, T.T., Aucott, R.L., van Deemter, M., Quaas, A., Walsh, S., Benten, D., Forbes, S.J., Wells, R.G. & Iredale, J.P. Matrix stiffness modulates proliferation, chemotherapeutic response, and dormancy in hepatocellular carcinoma cells. Hepatology 53, 1192-205 (2011). 286. Erler, J.T., Bennewith, K.L., Cox, T.R., Lang, G., Bird, D., Koong, A., Le, Q.T. & Giaccia, A.J. Hypoxia-Induced Lysyl Oxidase Is a Critical Mediator of Bone Marrow Cell Recruitment to Form the Premetastatic Niche. Cancer Cell 15, 35-44 (2009). 287. Mouw, J.K., Yui, Y., Damiano, L., Bainer, R.O., Lakins, J.N., Acerbi, I., Ou, G.Q., Wijekoon, A.C., Levental, K.R., Gilbert, P.M., Hwang, E.S., Chen, Y.Y. & Weaver, V.M. Tissue mechanics modulate microRNA-dependent PTEN expression to regulate malignant progression. Nature Medicine 20, 360-+ (2014). 288. Ruoslahti, E. & Reed, J.C. Anchorage dependence, integrins, and apoptosis. Cell 77, 477-8 (1994). 289. Frisch, S.M., Vuori, K., Ruoslahti, E. & Chan-Hui, P.Y. Control of adhesion-dependent cell survival by focal adhesion kinase. J Cell Biol 134, 793-9 (1996). 290. Gilmore, A.P., Metcalfe, A.D., Romer, L.H. & Streuli, C.H. Integrin-mediated survival signals regulate the apoptotic function of Bax through its conformation and subcellular localization. Journal of Cell Biology 149, 431-445 (2000).   242  291. Zahir, N., Lakins, J.N., Russell, A., Ming, W., Chatterjee, C., Rozenberg, G.I., Marinkovich, M.P. & Weaver, V.M. Autocrine laminin-5 ligates alpha6beta4 integrin and activates RAC and NFkappaB to mediate anchorage-independent survival of mammary tumors. J Cell Biol 163, 1397-407 (2003). 292. Lewis, J.M., Truong, T.N. & Schwartz, M.A. Integrins regulate the apoptotic response to DNA damage through modulation of p53. Proceedings of the National Academy of Sciences of the United States of America 99, 3627-3632 (2002). 293. Golubovskaya, V.M. & Cance, W.G. FAK and p53 protein interactions. Anticancer Agents Med Chem 11, 617-9 (2011). 294. Lim, S.T., Chen, X.L., Lim, Y., Hanson, D.A., Vo, T.T., Howerton, K., Larocque, N., Fisher, S.J., Schiaepfer, D.D. & Llic, D. Nuclear FAK promotes cell proliferation and survival through FERM-enhanced p53 degradation. Molecular Cell 29, 9-22 (2008). 295. Zhou, J., Schmid, T., Schnitzer, S. & Brune, B. Tumor hypoxia and cancer progression. Cancer Letters 237, 10-21 (2006). 296. Hynes, R.O. Cell-matrix adhesion in vascular development. J Thromb Haemost 5 Suppl 1, 32-40 (2007). 297. Bonanno, E., Iurlaro, M., Madri, J.A. & Nicosia, R.F. Type IV collagen modulates angiogenesis and neovessel survival in the rat aorta model. In Vitro Cell Dev Biol Anim 36, 336-40 (2000). 298. Schor, A.M., Schor, S.L. & Allen, T.D. Effects of culture conditions on the proliferation, morphology and migration of bovine aortic endothelial cells. J Cell Sci 62, 267-85 (1983). 299. Montesano, R., Orci, L. & Vassalli, P. In vitro rapid organization of endothelial cells into capillary-like networks is promoted by collagen matrices. J Cell Biol 97, 1648-52 (1983). 300. Jones, R.A., Kotsakis, P., Johnson, T.S., Chau, D.Y., Ali, S., Melino, G. & Griffin, M. Matrix changes induced by transglutaminase 2 lead to inhibition of angiogenesis and tumor growth. Cell Death Differ 13, 1442-53 (2006). 301. Malik, R., Lelkes, P.I. & Cukierman, E. Biomechanical and biochemical remodeling of stromal extracellular matrix in cancer. Trends Biotechnol (2015). 302. Alix-Panabieres, C. & Pantel, K. OPINION Challenges in circulating tumour cell research. Nature Reviews Cancer 14, 623-631 (2014). 303. Kim, M.Y., Oskarsson, T., Acharyya, S., Nguyen, D.X., Zhang, X.H., Norton, L. & Massague, J. Tumor self-seeding by circulating cancer cells. Cell 139, 1315-26 (2009). 304. Pantel, K. & Brakenhoff, R.H. Dissecting the metastatic cascade. Nat Rev Cancer 4, 448-56 (2004).   243  305. Baccelli, I., Schneeweiss, A., Riethdorf, S., Stenzinger, A., Schillert, A., Vogel, V., Klein, C., Saini, M., Bauerle, T., Wallwiener, M., Holland-Letz, T., Hofner, T., Sprick, M., Scharpff, M., Marme, F., Sinn, H.P., Pantel, K., Weichert, W. & Trumpp, A. Identification of a population of blood circulating tumor cells from breast cancer patients that initiates metastasis in a xenograft assay. Nature Biotechnology 31, 539-U143 (2013). 306. Fidler, I.J. Tumor heterogeneity and the biology of cancer invasion and metastasis. Cancer Res 38, 2651-60 (1978). 307. Mego, M., Giordano, A., De Giorgi, U., Masuda, H., Hsu, L., Giuliano, M., Fouad, T.M., Dawood, S., Ueno, N.T., Valero, V., Andreopoulou, E., Alvarez, R.H., Woodward, W.A., Hortobagyi, G.N., Cristofanilli, M. & Reuben, J.M. Circulating tumor cells in newly diagnosed inflammatory breast cancer. Breast Cancer Res 17, 2 (2015). 308. Pukazhendhi, G. & Gluck, S. Circulating tumor cells in breast cancer. J Carcinog 13, 8 (2014). 309. Thiery, J.P. & Lim, C.T. Tumor Dissemination: An EMT Affair. Cancer Cell 23, 272-273 (2013). 310. Bierie, B. & Moses, H.L. TGF beta: the molecular Jekyll and Hyde of cancer. Nature Reviews Cancer 6, 506-520 (2006). 311. Rice, G.E. & Bevilacqua, M.P. An Inducible Endothelial-Cell Surface Glycoprotein Mediates Melanoma Adhesion. Science 246, 1303-1306 (1989). 312. Okahara, H., Yagita, H., Miyake, K. & Okumura, K. Involvement of Very Late Activation Antigen-4 (Vla-4) and Vascular Cell Adhesion Molecule-1 (Vcam-1) in Tumor-Necrosis-Factor-Alpha Enhancement of Experimental Metastasis. Cancer Research 54, 3233-3236 (1994). 313. Schlesinger, M. & Bendas, G. Vascular cell adhesion molecule-1 (VCAM-1)-An increasing insight into its role in tumorigenicity and metastasis. Int J Cancer (2014). 314. Schlesinger, M., Roblek, M., Ortmann, K., Naggi, A., Torri, G., Borsig, L. & Bendas, G. The role of VLA-4 binding for experimental melanoma metastasis and its inhibition by heparin. Thromb Res 133, 855-62 (2014). 315. Friedl, P. Prespecification and plasticity: shifting mechanisms of cell migration. Curr Opin Cell Biol 16, 14-23 (2004). 316. Friedl, P. & Wolf, K. Plasticity of cell migration: a multiscale tuning model. J Cell Biol 188, 11-9 (2010). 317. Gardel, M.L., Schneider, I.C., Aratyn-Schaus, Y. & Waterman, C.M. Mechanical Integration of Actin and Adhesion Dynamics in Cell Migration. Annual Review of Cell and Developmental Biology, Vol 26 26, 315-333 (2010).   244  318. Sanz-Moreno, V. & Marshall, C.J. Rho-GTPase signaling drives melanoma cell plasticity. Cell Cycle 8, 1484-7 (2009). 319. Pankova, K., Rosel, D., Novotny, M. & Brabek, J. The molecular mechanisms of transition between mesenchymal and amoeboid invasiveness in tumor cells. Cellular and Molecular Life Sciences 67, 63-71 (2010). 320. Lammermann, T. & Sixt, M. Mechanical modes of 'amoeboid' cell migration. Curr Opin Cell Biol 21, 636-44 (2009). 321. Friedl, P., Borgmann, S. & Brocker, E.B. Amoeboid leukocyte crawling through extracellular matrix: lessons from the Dictyostelium paradigm of cell movement. J Leukoc Biol 70, 491-509 (2001). 322. Sanz-Moreno, V. & Marshall, C.J. The plasticity of cytoskeletal dynamics underlying neoplastic cell migration. Curr Opin Cell Biol 22, 690-6 (2010). 323. Friedl, P. Prespecification and plasticity: shifting mechanisms of cell migration. Current Opinion in Cell Biology 16, 14-23 (2004). 324. Friedl, P. & Wolf, K. Plasticity of cell migration: a multiscale tuning model. Journal of Cell Biology 188, 11-19 (2010). 325. Friedl, P., Zanker, K.S. & Brocker, E.B. Cell migration strategies in 3-D extracellular matrix: Differences in morphology, cell matrix interactions, and integrin function. Microscopy Research and Technique 43, 369-378 (1998). 326. Thiery, J.P. Epithelial-mesenchymal transitions in tumour progression. Nature Reviews Cancer 2, 442-454 (2002). 327. Lammermann, T. & Sixt, M. Mechanical modes of 'amoeboid' cell migration. Current Opinion in Cell Biology 21, 636-644 (2009). 328. Sanz-Moreno, V. & Marshall, C.J. Rho-GTPase signaling drives melanoma cell plasticity. Cell Cycle 8, 1484-1487 (2009). 329. Liu, Y.J., Le Berre, M., Lautenschlaeger, F., Maiuri, P., Callan-Jones, A., Heuze, M., Takaki, T., Voituriez, R. & Piel, M. Confinement and low adhesion induce fast amoeboid migration of slow mesenchymal cells. Cell 160, 659-72 (2015). 330. Ulrich, T.A., Pardo, E.M.D. & Kumar, S. The Mechanical Rigidity of the Extracellular Matrix Regulates the Structure, Motility, and Proliferation of Glioma Cells. Cancer Research 69, 4167-4174 (2009). 331. Ruprecht, V., Wieser, S., Callan-Jones, A., Smutny, M., Morita, H., Sako, K., Barone, V., Ritsch-Marte, M., Sixt, M., Voituriez, R. & Heisenberg, C.P. Cortical Contractility Triggers a Stochastic Switch to Fast Amoeboid Cell Motility. Cell 160, 673-685 (2015).   245  332. Pathak, A. & Kumar, S. Independent regulation of tumor cell migration by matrix stiffness and confinement. Proc Natl Acad Sci U S A 109, 10334-9 (2012). 333. Ulrich, T.A., de Juan Pardo, E.M. & Kumar, S. The mechanical rigidity of the extracellular matrix regulates the structure, motility, and proliferation of glioma cells. Cancer Res 69, 4167-74 (2009). 334. Pathak, A. & Kumar, S. Biophysical regulation of tumor cell invasion: moving beyond matrix stiffness. Integr Biol (Camb) 3, 267-78 (2011). 335. Wen, J.H., Vincent, L.G., Fuhrmann, A., Choi, Y.S., Hribar, K.C., Taylor-Weiner, H., Chen, S. & Engler, A.J. Interplay of matrix stiffness and protein tethering in stem cell differentiation. Nat Mater 13, 979-87 (2014). 336. Engler, A., Bacakova, L., Newman, C., Hategan, A., Griffin, M. & Discher, D. Substrate compliance versus ligand density in cell on gel responses. Biophys J 86, 617-28 (2004). 337. Chaudhuri, O., Koshy, S.T., Branco da Cunha, C., Shin, J.W., Verbeke, C.S., Allison, K.H. & Mooney, D.J. Extracellular matrix stiffness and composition jointly regulate the induction of malignant phenotypes in mammary epithelium. Nat Mater 13, 970-8 (2014). 338. Kumar, S. Cellular mechanotransduction: stiffness does matter. Nat Mater 13, 918-20 (2014). 339. Zaidel-Bar, R., Itzkovitz, S., Ma'ayan, A., Iyengar, R. & Geiger, B. Functional atlas of the integrin adhesome. Nat Cell Biol 9, 858-67 (2007). 340. Zaidel-Bar, R., Ballestrem, C., Kam, Z. & Geiger, B. Early molecular events in the assembly of matrix adhesions at the leading edge of migrating cells. Journal of Cell Science 116, 4605-4613 (2003). 341. Zaidel-Bar, R., Cohen, M., Addadi, L. & Geiger, B. Hierarchical assembly of cell-matrix adhesion complexes. Biochemical Society Transactions 32, 416-420 (2004). 342. Tamkun, J.W., Desimone, D.W., Fonda, D., Patel, R.S., Buck, C., Horwitz, A.F. & Hynes, R.O. Structure of Integrin, a Glycoprotein Involved in the Transmembrane Linkage between Fibronectin and Actin. Cell 46, 271-282 (1986). 3