Characterisation of TRIM16 As a Putative Tumour Suppressor Protein and Drug Target in Melanoma

Characterisation of TRIM16 as a putative tumour suppressor protein and drug target in melanoma

Selina K. Sutton

This thesis is submitted in fulfilment of the requirements for the degree of Doctor of Philosophy at the University of New South Wales, Australia

Children’s Cancer Institute Australia for Medical Research School of Women’s and Children’s Health, Faculty of Medicine University of New South Wales, Australia

March 2014

i PLEASE TYPE THE UNIVERSITY OF NEW SOUTH WALES Thesis/Dissertation Sheet

Sumame or Family name: Sutton

First name: Selina Other name/s:

Abbreviation for degree as given in the University calendar: ph9

School: Women's and Childrens Health Faculty: Medicine

Title: Characterization of TRIM16 as a putative tumour suppressor protein and drug target in melanoma

Abstract 350 words The incidence of cutaneous malignant melanoma is rapidly increasing. Melanoma is an extremely aggressive malignancy making up only 4% of total skin cancers, but it is responsible for 80% of skin cancer deaths. The clinical benefit of BRAF-MEK-inhibitor therapy in BRAF mutant melanoma remains temporary, and the mechanisms of metastasis with advanced disease are incompletely understood. The tripartite motif (TRIM) family members have been implicated in the pathogenesis of multiple cancers functioning as both oncogenes and tumour suppressors. Intriguingly, in the histological progression of squamous cell carcinoma (SCC) from normal skin, TRlM16 was significantly reduced in vivo and overexpression of TRlM16 reduced SCC cell migration in vitro. Furthermore, TRlM16 has demonstrated some of the features of a tumour suppressor protein in neuroblastoma through down-regulation of protein binding partners: cytoplasmic vimentin and nuclear E2F1. Taken together, these data suggest that TRIM16 acts to repress cancer cell replication and migration. However, the role of TR|M16 in melanoma is presently unknown. We developed a keratinocyte-specific TRlM16 knockout mouse model for the evaluation of TR|M16 in the development of SCC. Surprisingly, these mice demonstrated large cutaneous melanocytic lesions, indicating a possible role in melanoma. We thep used human melanoma cell lines and primary tissues to show that low TR|M16 expression associates with enhanced cell migration in vitro, metdstatic disease in vivo, and poor prognosis in a large cohort of melanoma patients with lymph node metastasis. TR|M16 protein expression is increased with vemurafenib BRAF inhibitor treatment in vitro and is partially required for drug mechanism of action. ln addition, TRlM16 protein expression is increased by BRAF inhibitor treatment in vivo. Taken together, these data indicated that repression of TR|M16 expression enhanced replication and migration of melanoma cells, associated with poor prognosis and thus represents as a potential therapeutic target. Furthermore, we identified a novel small molecule, compound 012, which unexpectedly increased BRAF inhibitor potency in BRAF wildtype mutant melanoma cells, an effect partially dependent on TR|M16. This establishes a basis for investigation of the combination of efficacious small molecules synergistic with vemurafenib, and may open an important novel treatment option for BRAF wild-type patients.

Declaration relating to disposition of proiect

I hereby grant to the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or in part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act I 968. I retain all property rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part ofthis thesis or dissertation.

I also authorise University Microfilms to use the 350 word abstract of my thesis in Dissertation Abstracts lntemational (this is applicable to doctoral theses only).

The University recognises that there may be exceptional circumstances requiring restrictions on copying or conditions on use. Requests for restrictionforaperiodofupto2yearsmustbemadeinwriting. Requestsforalongerperiodofrestrictionmaybeconsideredinexceptional circumstances and require the approval of the Dean of Graduate Research.

FOR OFFICE Date of completion of requirements for Award:

THIS SHEET IS TO BE GLUED TO THE INSIDE FRONT COVER OF THE THESIS ORIGINALITY STATEMENT 'I hereby declare that this submission is my own work and to the best of my knowledge it contains no materials previously published or written by another person, or substantial proportions of material which have been accepted for the award of any other degree or diploma at UNSW or any other educational institution, except where due acknowledgement is made in the thesis. Any contribution made to the research by others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis. I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the project's design and conception or in style, presentation and linguistic expression is acknowledged.' COPYRIGHT STATEMENT

'I hereby grant the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act 1968. I retain all proprietary rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation.

I also authorise University Microfilms to use the 350 word abstract of my thesis in Dissertation Abstract Intemational (this is applicable to doctoral theses only). I have either used no substantial portions of copyright material in my thesis or I have obtained permission to use copynght material; where perrnission has not been granted I have applied/will apply for a partial

restriction of the digital copy of my thesis or dissertation.'

AUTHENTICITY STATEMENT 'I certifu that the Library deposit digital copy is a direct equivalent of the final officially approved version of my thesis. No emendation of content

has occurred and if there are any minor variations in formatting, they are the result of the conversion to digital format.' Signed%

111 Table of Contents Page

List of tables xiv List of figures xv Abstract xxii Abbreviations xxv Publications arising from this thesis xxxi Publications in in collaboration during this thesis xxxi Presentations arising from this thesis xxxi Acknowledgements xxxiv

Chapter 1. Introduction 1 1.1 Neural crest development and melanocyte formation 2 1.1.1 The neural crest 2

1.2 Melanoma 4 1.2.1 Melanoma incidence worldwide 4

1.2.1.1 Melanoma incidence in Australia 4 1.2.1.2 Known risk factors for melanoma development 5

1.2.1.3 Ultraviolet radiation exposure 6

1.2.1.4 Melanoma ABCDE 7

1.3 Diagnosis of melanoma 9 1.3.1 Melanoma diagnosis and early prognosis 9 1.3.2 Response Evaluation Criteria in Solid Tumours (RECIST) 10 1.3.3 Melanoma stage and 5 year survival rates 10 1.3.4 Sentinel lymph node biopsy 12 1.4 Melanoma pathology 12 1.4.1 The role MITF in melanocyte differentiation and melanoma development 14

1.4.2 The MAPK pathway in normal melanocytes 15

1.4.3 Mutations in the MAPK pathway 19 1.4.4 BRAF and NRAS mutations 19 1.4.4.1 BRAF mutation 20 1.4.4.2 NRAS mutation 24 1.4.5 Activation of the PTEN-PI3K pathway and AKT 27 1.4.6 Oncogene induced senescence in BRAFV600E melanomas 28 1.4.7 CDKN2A and p16 -CDK4-RB senescence barrier 29 1.4.8 Inactivation of ARF-p53 pathway senescence barrier 31 1.4.8.1 Defective apoptosis in melanoma 31 1.4.9 KIT mutations 32 1.4.10 Additional melanoma mutations 33 1.4.11 Melanoma migration and metastasis 33 1.5 Epigenetic modification in melanoma 35 1.6 Current melanoma therapies 37 1.6.1 Surgery, radiation therapy and chemotherapy 37 1.6.2 Immunotherapy 39 1.6.3 Key targeted therapies 41 1.6.3.1 Drug targeting of activating BRAF 41 1.6.3.2 Sorafenib 42 1.6.3.3 Development of Vemurafenib 42 1.6.3.4 NRAS targeting 45 1.6.3.5 PI3K/AKT pathway targeting 46 1.6.4 MEK inhibitors 47 1.6.4.1 Trametinib 47 1.6.4.2 Relevance of the use of MEK inhibitors in BRAF wild-type cell lines 48 1.6.5 Personalized medicine 48

1.7 Key clinical trials 49 1.7.1 BRIM-3 study 49 1.7.2 BREAK-3 study 51 1.7.3 METRIC study 52 1.7.4 FDA approval of treatments for metastatic melanoma 53 1.7.5 Combination therapies and recent clinical trials 55 1.7.5.1 Immunotherapy and BRAF inhibitor combination 56 1.8 Raf isoform switching 57 1.8.1 Mechanisms of RAF isoform switching 58 1.9 Drug resistance in melanoma 64 1.9.1 Mechanism of BRAF inhibitor resistance 64 1.9.1.1 Intrinsic resistance 67 1.9.1.2 Acquired resistance 67 1.9.1.3 ERK dependant mechanisms of resistance 68

1.9.1.4 ERK-independent mechanisms of resistance 70 1.9.2 Strategies to overcome vemurafenib resistance 71 1.9.2.1 Intermittent dosing and Vemurafenib resistance 72 1.9.3 MEK inhibitor resistance 73 1.9.4 Molecular biomarkers and prognostic markers in melanoma 73 1.10 Novel tumour suppressors in proteins melanoma 75 1.11 TRIM family proteins 76 1.11.1 TRIM family proteins and keratinocyte biology 77 1.11.2 TRIM family proteins in innate immunity 78 1.11.3 TRIM family proteins as E3 ubiquitin ligases 82 1.11.4 TRIM family proteins in cancer 82 1.11.5 TRIM family tumour suppressors and oncogenes 84 1.11.6 TRIM protein biomarkers and prognostic markers 86 1.12 TRIM16 chromosomal location and tissue expression 87 1.12.1 TRIM16 protein and post-translational modification 87

1.12.2 The role of TRIM16 in innate immunity 89 1.12.3 The role of TRIM’s in type 1 Interferon response 89 1.12.4 TRIM16 mutation in cancer 91 1.12.5 TRIM16 loss of heterozygosity in cancer 91 1.12.6 TRIM16 as a candidate tumour suppressor in neuroblastoma 94 1.12.7 TRIM16 as a candidate tumours suppressor in squamous cell carcinoma 94 1.12.8 TRIM16 and retinoid signalling 95 1.12.8.1 RARβ as a tumour suppressor gene in melanoma 97 1.12.8.2 TRIM16 regulation of retinoid signalling in neuroblastoma 97 1.13 Innate immunity in melanoma 97

1.13.1 INF-β and melanoma 97 1.13.2 Clinical use of Interferon treatment 99 1.13.2.1 IFN-α2b 99 1.13.2.2 IFN-β 99 1.13.2.3 IFN-β transcription via the enhanceosome 101

1.14 Perspectives and experimental directions 103

Chapter 2. Materials and Method 105

1. Materials 106

2.1 General techniques 107

2.1.1 Tissue culture conditions 107

2.1.2 Plasmid preparation 107

2.1.3 TRIM16 plasmids and siRNA transfection 108

2.1.4 Co-transfection of TRIM16 plasmid and siRNA’s, IFNβ1 and c-Jun 108

2.1.5 Tissue microarray construction 109

2.1.6 Cytoplasmic and nuclear protein fractionation 109

2.1.7 BCA Protein assay 109

vii

2.1.8 Western blotting 110

2.1.9 BrdU incorporation cell proliferation assay 111

2.1.10 Alamar Blue cell viability assay 111

2.1.11 ELISA apoptosis assay 111

2.1.12 Scratch wound migration assay 112

2.1.13 Transwell migration assay 112

2.1.14 PCR cancer pathway array 113

2.1.15 Quantitative real-time polymerase chain reaction (RT-qPCR) 113

2.1.16 Microarray and analysis 113

2.1.17 PCR of genomic DNA and sequencing 114

2.1.18 Chromatin immunoprecipitation assay 114

2.1.19 Cycloheximide and 5-aza-2’-deoxycytidine (5-Aza) treatment 115

2.1.20 MG132 proteosomal inhibitor treatment 115

2.1.21 Clonogenic assay 115

2.1.22 Immunofluorescence 116

2.1.23 Immunohistochemistry 116

2.1.24 Drug synergy assays 117

2.1.25 Microsomal stability assay 117

2.1.26 Statistical analysis 118

2.2 Generation of TRIM16 skin specific knock-out mice 118

2.2.1 Preparation of TRIM16 floxed mice 118

2.2.2 Development of skin lesions using a two-stage skin

carcinogenesis model 119

2.2.3 Design of the TRIM16 knock-out construct 119

2.2.3.1 Genotyping PCR design to determine TRIM16 knockout 121

2.2.4 Development of skin lesions using a two-stage skin

carcinogenesis model 124 viii

2.2.5 Culture of primary keratinocytes 124

Chapter 3. Identification and characterization of a candidate tumour suppressor protein, TRIM16 in melanoma 125 3.1 Introduction 126 3.2 Results 129 3.2.1 TRIM16 protein expression is reduced in metastatic melanoma and correlates with overall survival risk in Stage III disease 129 3.2.1.1 TRIM16 protein expression is reduced during melanomagenesis 130 3.2.1.2 High TRIM16 protein expression is associated with favourable patient outcome in a cohort of lymph node metastasis patients 133 3.2.1.3 TRIM16 protein expression does not correlate with primary tumour ulceration or thickness 136 3.2.2 Nuclear and total TRIM16 protein is lost in melanoma cells compared to normal melanocytes 141 3.2.2.1 TRIM16 is reduced by the proteosomal degradation pathway in melanoma cells 146 3.2.3 TRIM16 protein half-life is reduced in melanoma by an epigenetic mechanism 148 3.2.3.1 E3 ubiquitin ligase, NEDD4, is decreased with TRIM16 overexpression, but does not correlate with TRIM16 expression in melanoma 153 3.2.3.2 TRIM16 is mutated at low levels in melanoma cell lines 158 3.2.4 TRIM16 protein expression reduces melanoma cell proliferation and migration and TRIM16 loss in melanocytes results in increased migration capacity 161 3.2.4.1 TRIM16 gene silencing reduces melanocyte proliferation and increases migration 166 3.2.4.2 TRIM16 gene silencing reduces melanoma cell proliferation and induces EMT marker, Snail in A375 cells 169 3.2.5 TRIM16 directly binds to the IFNβ1 promoter and induces IFNβ1 transcription 172

3.2.5.1 TRIM16 over-expression induces transcriptional changes in gene involved in angiogenesis and invasion & metastasis 173 3.2.6 TRIM16 binds to the IFNβ1 promoter in the enhanceosome region 176 3.2.7 IFNβ1 expression is significantly decreased in primary human metastatic melanoma tissues and is required for TRIM16 mediated inhibition in cell proliferation 183 3.2.7.1 TRIM16 protein expression strongly correlates with IFNβ1 expression in distant metastasis melanoma 187 3.2.7.2 Gene silencing of IFNβ1 decreases melanoma cell proliferation 192 3.2.8 Exogenous recombinant human IFNβ1 protein does not increase TRIM16 protein in vitro 195 3.2.9 TRIM16 protein is increased and stabilized with vemurafenib treatment and is partially required for vemurafenib drug action 198 3.2.9.1 TRIM16 protein expression is increase in melanoma cell lines with vemurafenib in a dose dependent manner 198 3.10 TRIM16 is increased with vemurafenib treatment in human patient samples 203 3.11 Proposed model of TRIM16 action in response to vemurafenib treatment and modulation of IFNβ1 expression 206 3.3 Discussion 208 3.4 Conclusions 211

Chapter 4. Modeling TRIM16 as a novel tumour suppressor protein in melanoma in vivo 212 4.1 Introduction 213 4.2 Results 215 4.2.1 Characterization of skin-specific knockout mice 215 4.2.1.1 Analysis of hair, and keratinocytes in skin-specific knockout mice 219 4.2.2 Characterization of TRIM16 full-tissue knockout mice 222 4.2.3 Skin carcinogenesis study on skin-specific TRIM16 knockout mice 225 4.2.3.1 Two-stage skin carcinogenesis on TRIM16 conditional skin-specific knock-out mice 228

4.2.3.1.1 TRIM16 heterozygous skin-specific knockout mice have reduced latency for papilloma development 231 4.2.3.1.2 TRIM16 heterozygous keratinocyte mice have increased squamous cell carcinoma development 233 4.2.3.4 Development of melanocytic lesions with two-stage carcinogen treatment 236 4.2.3.4.1 Skin-specific TRIM16 heterozygous mice have reduced latency of melanocytic lesion development 240 4.2.3.4.2 TRIM16 homozygous knockout mice develop larger melanocytic lesions that heterozygous or wild-type mice 242 4.2.3.7 Genotype has no influence on inguinal lymph node size 246 4.2.3.7.1 Melanocytic lesions demonstrate metastatic potential by pigmentation of regional lymph nodes 249 4.2.4 Skin carcinogenesis study on TRIM16 full-tissue knock-out mice 252 4.2.4.1 TRIM16 full-tissue knockout mice do not have increase melanocytic lesions size with skin carcinogen challenge 253 4.2.4.2 TRIM16 full-tissue knockout mice do not have reduced tumour latency with skin carcinogen treatment 255 4.3.2.3 TRIM16 full-tissue knockout mice do not develop squamous cell carcinomas compared to wild-type and heterozygous mice 259 4.3 Discussion 261 4.4 Conclusions 264

Chapter 5. Identification, characterization, mechanism of action and in vivo modeling of compound 012 as an enhancer of vemurafenib 266 5.1 Introduction 267 5.2 Results 268 5.2.1 Identification and characterization of three anti-melanoma compounds 268 5.2.1.1 Determination of the IC50 of anti-melanoma compounds 271 5.2.1.2 Compounds demonstrate greater efficacy against melanoma compared to normal fibroblasts 273 5.2.1.3 Compounds decrease melanoma cell colony forming ability 275 5.2.1.4 Lipinski’s rule of 5 277

5.2.1.5 Anti-melanoma compounds demonstrate activity in combination with dacarbazine and vemurafenib 279 5.2.2 Characterization of compound 012 synergy with vemurafenib 281 5.2.2.1 Compound 012 demonstrates minimal synergy with vemurafenib in BRAF mutant cells 281 5.2.2.2 Compound 012 demonstrates strong synergy with vemurafenib in BRAF wild-type cells 285 5.2.2.3 Compound 012 demonstrates synergy with vemurafenib across a panel of BRAF wild-type melanoma cells 289 5.2.2.4 Combination of compound 012 and vemurafenib induces cell senescence 292 5.2.2.5 Combination of compound 012 and vemurafenib reduces clonogenicity in both BRAF mutant and wild-type melanoma cells 294 5.2.3 Development of the compound 012 substructure library 299 5.2.3.1 Similarity search 299 5.2.3.2 Compound 012 structure activity relationship 301 5.2.4 Combination compound 012/vemurafenib mechanism of action 308

5.2.4.1 Compound 012 decreases ERK phosphorylation after prolonged treatment 308 5.2.4.2 Compounds demonstrate an increase in TRIM16 protein expression 5.2.4.3 Compound 236 increases cytoplasmic TRIM16 protein, but decreases the level of nuclear TRIM16 312 5.2.4.4 TRIM16 is partially required for compound 012/vemurafenib mechanism of action 314 5.2.4.5 Microarray analysis reveals WNT and Beta3 Integrin signalling may be required for combination compound 012/vemurafenib mechanism of action 319 5.2.4.6 Combination of compound 012 with melanoma pathway inhibitors 326 5.2.5 Development of three melanoma cell lines resistant to vemurafenib 329 5.2.5.1 Compound 012 may demonstrate activity against resistant vemurafenib cell lines 335 5.2.6 Testing combination vemurafenib/compound 012 in vivo 337 5.2.6.1 Determination of compound 012 stability 338 xii

5.2.6.2 Maximum tolerated dose study for compound 012 340 5.3 Discussion 345 5.4 Conclusions 348

Chapter 6. Concluding remarks 349

6.1 General discussion 350

6.2 Future perspectives 355

6.3 Conclusions 356

References 358

xiii

List of Tables

Table Page

Table 1.1. Survival data based upon the American Joint Committee on Cancer (AJCC) five-year survival with proper treatment, seventh edition (2009) 11 Table 1.2. Percentage of DNA methylation of genes in melanoma 36 Table 1.3. TRIM gene alterations in cancer pathogenesis 85 Table 1.4. TRIM16 post-translational modification of human and corresponding mouse protein 88 Table 1.5. Amino acid post-translational modification and corresponding region of TRIM16 protein motif 88 Table 1.6. Incidence of TRIM16 loss of heterozygosity across human cancer samples 96 Table 2.1 Primer design for genotyping of TRIM16 knockout mice 122 Table 4.1. Summary of development of squamous cell carcinoma at 21 weeks in skin-specific knockout mice 235 Table 4.2. Summary of development of squamous cell carcinoma at 28 weeks in full-tissue knockout mice 260

Table 5.1 IC50 of compounds against a melanoma cell line panel and normal human embryonal fibroblast lines 272 Table 5.2. Summary of cell line and synergy between compound 012 and vemurafenib 287 Table 5.3. Specifics of the maximum tolerated dose study for compound 012 341 Table 5.4. Treatment groups of maximum tolerated dose study 341 Table 5.5. Appearance and behavior of mice with compound 012 treatment at days 10-13 342

xiv

List of Figures

Figure Page

Figure 1.1. Development of melanocytes from the neural crest 3

Figure 1.2. The ABCDE’s of melanoma 8

Figure 1.3. Melanoma pathogenesis and molecular aberrations 13

Figure 1.4. Signalling of the MAPK pathway in a normal melanocyte cells 16

Figure 1.5. Downstream targets of ERK signalling in a normal melanocyte cell 18 Figure 1.6. Melanoma cells that harbour BRAF mutations have increased cell proliferation through constitutive MAPK signalling and PI3K pathway disruption 23 Figure 1.7. Melanoma cells that harbour the NRAS mutation result in melanoma cell proliferation through CRAF mediated MAPK activation and AKT activation 26 Figure 1.8. CDKN2A pathway control in melanoma 30

Figure 1.9. Timeline of FDA approved treatment for advanced melanoma 54

Figure 1.10. Summary of key melanoma therapies 54

Figure 1.11. RAF isoform switching with BRAF inhibitor treatment 60

Figure 1.12. Vemurafenib treatment of BRAF wild-type/NRAS mutant melanoma cell 63 Figure 1.13. Mechanisms of Vemurafenib resistance by-passing BRAF 66

Figure 1.14. TRIM proteins are modulated by type I and II IFN’s and regulate innate immune response 81 Figure 1.15. Schematic of the TRIM16 protein containing two B-Box groups, a coiled-coil domain and a PRY/SPRY grouping 88 Figure 1.16. Summary of the known function of TRIM16 protein 96

Figure 1.17. Schematic representation of the components of the IFNβ1 enhanceosome 102 Figure 2.1. The FLOX construct and homology arms 120

Figure 2.2. Schematic of PCR to determine presence of the flox construct and hetero or homozygosity 123

Figure 3.1. TRIM16 protein expression is reduced in metastatic melanoma compared to nevi and primary melanoma 131 Figure 3.2. High TRIM16 is prognostic of favourable overall survival in lymph node metastasis patients 134 Figure 3.3. TRIM16 expression is not higher in lymph node metastasis from ulcerated melanoma compared to non-ulcerated melanoma 138 Figure 3.4. Schematic summary of TRIM16 loss and melanoma stage 140 Figure 3.5. Melanoma cell line panel and known mutations in key melanoma pathogenesis genes 143 Figure 3.6. TRIM16 protein expression is decreased in melanoma cell lines compared to normal human epidermal melanocytes 144 Figure 3.7. Nuclear TRIM16 localization is lost in melanoma cells compared to normal human melanocytes 145 Figure 3.8. TRIM16 is reduced by the proteosomal degradation pathway in melanoma cells 147 Figure 3.9. TRIM16 half-life is shorter in melanoma cells compared to normal human melanocyte and fibroblast cells 150 Figure 3.10. TRIM16 protein expression and half-life is decreased in melanoma cell lines compared to normal melanocytes and is increased following treatment with the demethylating agent, 5-aza-cytidine 152 Figure 3.11. Knockdown of NEDD4 increases TRIM16 protein and decreases cell proliferation 156 Figure 3.12. TRIM16 harbours a mutation in B-Box-1 in 11% of melanoma cell lines 159 Figure 3.13. TRIM16 over-expression reduces melanoma cell proliferation and induces apoptosis 163

xvi

Figure 3.14. TRIM16 overexpression reduces melanoma cell migration 165 Figure 3.15. TRIM16 gene silencing reduces melanoma cell proliferation and increases migration of NHEM cells 168 Figure 3.16. TRIM16 gene silencing reduces melanoma cell proliferation and results in a phenotype change 170 Figure 3.17. TRIM16 modulates epithelial to mesenchymal marker, Snail, in A375 cells 171 Figure 3.18. IFNβ1 mRNA expression is increased with TRIM16 overexpression 175 Figure 3.19. TRIM16 directly binds the IFNβ1 gene promoter and induces IFNβ1 transcription 180 Figure 3.20. TRIM16 binds to the IFNβ1 promoter in the same region as c-JUN, an integral part of the enhanceosome complex essential for IFNβ1 transcription 182 Figure 3.21. IFNβ1 expression is significantly decreased in primary human metastatic melanoma tissue 185 Figure 3.22. IFNβ1 is decreased in some melanoma cell lines compared to normal melanocyte NHEM cells, but does not correlate with TRIM16 expression 186 Figure 3.23. IFNβ1 expression correlates with TRIM16 expression in melanoma distant metastasis and is required for TRIM16-mediated inhibition in cell proliferation 189 Figure 3.24. c-JUN is required for TRIM16-mediated inhibition of cell proliferation 191 Figure 3.25. Knockdown of IFNβ1 reduces cell viability and TRIM16 protein expression 194 Figure 3.26. Recombinant human IFNβ1 treatment reduces melanoma cell proliferation but does not increase TRIM16 protein 197 Figure 3.27. Vemurafenib treatment increases TRIM16 protein in melanoma cells in vitro 200 Figure 3.28. TRIM16 protein stability is increased by PLX4032 and is required for PLX4032-induced loss of melanoma cell viability 202 Figure 3.29. TRIM16 protein expression is increased with vemurafenib treatment in patient clinical samples 205 Figure 3.30. A proposed model for the role of TRIM16 in melanoma 207

xvii

Figure 4.1 Confirmation of TRIM16 knockout mice 216 Figure 4.2. TRIM16 skin-specific knockout mice do not have difference in food consumption 218 Figure 4.3. TRIM16 skin-specific knockout mice do not have different morphology in the Awl hair 220 Figure 4.4. Primary keratinocytes from TRIM16 knockout mice do not have different morphology 221 Figure 4.5. TRIM16 skin-specific knockout mice do not have difference in food consumption 223 Figure 4.6. TRIM16 skin-specific mice do not show differences in body weight or length 224 Figure 4.7. Two mouse models are used to understand the role of TRIM16 in skin carcinogenesis 227 Figure 4.8. Schematic of the progression of squamous cell carcinoma with the two-stage skin carcinogenesis protocol 229 Figure 4.9. Representative examples of normal mouse skin, papilloma development and squamous cell carcinoma development 230 Figure 4.10. TRIM16 heterozygous skin knockout mice have reduced latency in papilloma development 232 Figure 4.11. Squamous cell carcinoma develop in TRIM16 heterozygous mice with two-stage skin carcinogenesis treatment 234 Figure 4.12. Evidence of blood supply to the squamous cell carcinoma tumour 235 Figure 4.13. Schematic of melanocytic lesion development with two-stage skin carcinogen treatment 238 Figure 4.14. Model of the development of murine melanoma by two-stage skin carcinogenesis 239 Figure 4.15. TRIM16 heterozygous skin knockout mice have reduced latency in melanocytic lesion development 241 Figure 4.16. TRIM16 homozygous mice develop larger melanocytic lesions than wild-type and heterozygous mice 244 Figure 4.17. TRIM16 heterozygous skin-specific mice display larger papilloma with carcinogen challenge 245

xviii

Figure 4.18. No significant difference is found in lymph node size between wild-type and heterozygous TRIM16 mice 248 Figure 4.19. Melanocytic lesions demonstrate metastatic potential by pigmentation of regional lymph node 250 Figure 4.20. TRIM16 wild-type and homozygous mice develop larger melanocytic lesions in full-tissue knockout mice 253 Figure 4.21. TRIM16 homozygous mice develop smaller papilloma compared to wild-type and heterozygous mice 254 Figure 4.22. TRIM16 homozygous full-tissue mice do not have reduced latency for melanocytic lesion development 256 Figure 4.23. TRIM16 homozygous full-tissue mice do not have reduced latency for papilloma development 257

Figure 5.1. Identification of three anti-melanoma compounds from compound SAHA enhancer screening project 270 Figure 5.2. Anti-melanoma compounds have activity against melanoma cells, but not normal fibroblasts 274 Figure 5.3. Colony forming ability is reduced in the presence of all three anti-melanoma compounds at IC50 doses 276 Figure 5.4. Structure of the three anti-melanoma compounds and comment on the suitability as a lead compound for drug development 278 Figure 5.5. Anti-melanoma compounds demonstrate increased potency with vemurafenib over dacarbazine 280 Figure 5.6. Compound 012 demonstrates synergy with vemurafenib in BRAFV600E mutant cells 284 Figure 5.7. Compound 012 demonstrates synergy with vemurafenib in BRAFWT mutant cells 287 Figure 5.8. Combination 012/vemurafenib is effective against a BRAFWT melanoma cell line panel with minimal toxicity to normal fibroblasts 290 Figure 5.9. Combination 012/vemurafenib effectively induces cell death in BRAFWT cell lines 291 Figure 5.10. Compound 012 induces senescence in BRAFWT cells and combination compound 012/vemurafenib induces senescence in both BRAFWT and

xix

BRAFV600E cells 293 Figure 5.11. Combination compound 012/vemurafenib ablates colony formation in BRAFWT melanoma cells 296 Figure 5.12. Combination compound 012/vemurafenib reduced colony forming ability in 2/3 of both BRAFV600E, and BRAFWT cells 298

Figure 5.13. IC50 of focus library from parental compound 012 300 Figure 5.14. Compound 012 structure activity relationship 302 Figure 5.15. The twenty-two compound focus library demonstrates variable cell viability to compound 012/vemurafenib combination 303 Figure 5.16. Enhanced anti-melanoma activity in Mel-JD cells but not MM200 cells 305 Figure 5.17. Enhanced activity in Mel-JD and MM200 cells and loss of activity 306 Figure 5.18. Compound 012 decreases ERK phosphorylation with long term treatment 309

Figure 5.19. Anti-melanoma compounds increase TRIM16 at IC50 concentrations 311 Figure 5.20. TRIM16 protein is increased in the cytoplasm and is decreased in the nucleus in a dose dependent manner with compound 236 313 Figure 5.21. Combination compound 012/vemurafenib increases TRIM16 in BRAFWT cells 315 Figure 5.22. TRIM16 is partially required for compound 012/VEM mechanism of action 317 Figure 5.23. Twenty-one genes differentially expressed in the combination of 012/vemurafenib treatment of Mel-JD cells 321 Figure 5.24. Key components of cholesterol biosynthesis are up-regulated by compound 012 322 Figure 5.25. Significantly regulated genes are involved in WNT signaling and Beta-3 integrin signaling 323 Figure 5.26. Anti-melanoma compounds and their primary targets 327 Figure 5.27. Melanoma inhibitors display an additive effect in combination with compound 012 328 Figure 5.28. Development of three melanoma cell lines resistant to vemurafenib 330 Figure 5.29. Cell morphology and western blotting confirms ERK re-activation in vemurafenib resistant cells compared to parental cells 332

Figure 5.30. Summary of resistance mechanism of three melanoma cell lines 333 Figure 5.31. Vemurafenib resistant cells show increased sensitivity to compound 012 compared to parental cells 336 Figure 5.32. Compound 012 has a half-life if 13.5 minutes by the microsomal stability assay 339 Figure 5.33. Schematic of melanoma xenograft and treatment 344

xxi

Abstract

The incidence of cutaneous malignant melanoma is rapidly increasing with an American study estimating an 8-fold increase in women, and 4-fold increase in men between 1970 and 2009. Melanoma is an extremely aggressive malignancy making up only 4% of total skin cancers, but it is responsible for 80% of skin cancer deaths. While early detection and removal by surgical excision results in an almost complete cure, metastatic melanoma patients have a poor prognosis, with median survival estimated between 6-9 months. Metastatic melanoma is highly chemo-resistant with traditional chemotherapies such as dacarbazine achieving only a 5% response rate. New targeted melanoma treatments, notably vemurafenib (PLX4032), offer a much higher response rate due to specific BRAFV600E targeting in patients. However, disease progression is invariably observed at 5-7 months, further highlighting the need for a more comprehensive understanding of disease pathology. Clearly, efficacious combination therapy is required to target such a highly heterogeneous cancer for melanoma.

The tripartite motif (TRIM) family members have been implicated in the pathogenesis of multiple cancers functioning as both oncogenes and tumour suppressors. TRIM16 is a member of the tripartite motif family of proteins comprising over 100 members which are characterized by a RING B-box-Coiled-coil protein domain architecture. TRIM16 has been identified as having a role in innate immune function as part of the NALP1 inflammasome. TRIM16 was secreted by keratinocytes in a caspase-1 dependent manner, a process that is enhanced by interleukin-1β (IL-1β). In addition, TRIM16 increased IL-1β secretion. TRIM16 was highly expressed in epidermal growth factor-responsive basal keratinocytes but TRIM16 protein down- regulation was observed in the hyper-thickened epithelium of skin wounds. TRIM16 has been shown to increase differentiation markers keratins 6, 10 and involucrin in keratinocytes. Intriguingly, in the histological progression of squamous cell carcinoma (SCC) from normal skin, TRIM16 was significantly reduced in vivo and TRIM16 expression reduced SCC cell migration in vitro. In addition to the role in keratinocyte biology, TRIM16 has demonstrated some of the features of a tumour suppressor protein in neuroblastoma through down-regulation of protein binding partners: cytoplasmic vimentin and nuclear E2F1. Taken together, these data suggest that TRIM16 acts to

xxii repress cancer cell replication and migration. However, the role of TRIM16 in melanoma is presently unknown.

We developed a keratinocyte-specific TRIM16 knockout mouse model for the evaluation of TRIM16 in the development of SCC. Surprisingly, these mice demonstrated large cutaneous melanocytic lesions, indicating TRIM16 may have a role in melanoma. We then used human melanoma cell lines and primary tissues to show that low TRIM16 expression associates with enhanced cell migration in vitro, metastatic disease in vivo, and poor prognosis in a large cohort of melanoma patients with lymph node metastasis. Overexpression of TRIM16 in melanoma cells markedly up-regulated interferon β1 (IFNβ1) and c-Jun expression levels. The expression pattern of IFNβ1 in human melanoma primary tissues closely mirrored TRIM16, while the anti-proliferative TRIM16 effect was dependent on IFNβ1 expression. TRIM16 protein expression is increased with vemurafenib treatment in vitro and is partially required for drug mechanism of action. In addition, TRIM16 protein expression is increased by BRAF inhibitor treatment in vivo. Taken together, these data indicated that repression of TRIM16 expression enhanced replication and migration of melanoma cells and thus represents a novel therapeutic target.

Furthermore, among 24 ‘hit’ compounds from our recent drug screen for small molecules that enhanced the cytopathic effects of histone deacetylase inhibitors, we identified compound 012, which unexpectedly had significant single agent activity on melanoma cell viability and increased BRAF inhibitor potency in wild-type BRAF and NRAS mutant melanoma cells. We found that compound 012 had significant single agent activity on melanoma cell viability at 10µM, and limited toxicity against normal WI38 fibroblast cells. Most importantly, when combined with the mutant BRAF inhibitor, vemurafenib, compound 012 markedly increased vemurafenib potency in 4 BRAF wild-type melanoma cell lines. Vemurafenib and compound 012 exhibited synergy at a range of concentrations, and, dramatically reduced anchorage-independent colony formation. Vemurafenib and compound 012 had no effect on the level of phospho-ERK, CRAF, Cot1 or phospho-MEK levels, which increased as expected in BRAF wild-type cells treated with vemurafenib alone or with the combination, but instead, the vemurafenib and compound 012 combination markedly increased a growth

xxiii suppressor, TRIM 16 protein level, and, knockdown of TRIM16 in melanoma cells significantly reduced vemurafenib and compound 012 induced growth inhibition by Alamar Blue assay, suggest that vemurafenib and compound 012 exerted synergistic anti-cancer effects by inducing TRIM16 expression, resulting in consequent growth arrest. Taken together, we have identified a novel compound which works synergistically with BRAF inhibitor and specifically target BRAF wild-type/ NRAS mutant melanoma cells.

For the first time, this research establishes the functional role of TRIM16 in melanoma pathogenesis as an inhibitor of metastasis and proliferation. TRIM16 binds directly and induce the transactivation of IFNβ1. This research also determines a new mechanism of anti-cancer signaling that is deregulated in melanomagenesis. Our results indicate that low TRIM16 levels may be used to predict the patients with localized melanoma who would benefit from BRAF inhibitor therapy. Identification of small molecule, compound 012, working in synergy with vemurafenib in BRAFWT patients establishes a basis for investigation of the combination of efficacious small molecules synergistic with vemurafenib, and may open a novel treatment option for BRAF wild- type patients.

xxiv

Abbreviations

4HT 4-hydroxy-tamoxifen

AMPK AMP activated protein kinase

Abi2 Abl-interactor 2

ACP5 Acid phosphatase 5, tartrate resistant

ARF6 ADP-ribosylation factor 6

ATF4 Activating transcription factor 4 cAMP Cyclic adenosine monophosphate

AJCC American joint committee on cancer

AKT Murine thymoma viral oncogene homolog 1

APAF Apoptosis associated factor

ARAF v-Raf murine sarcoma 3611 viral oncogene homolog

ASK-1 Mitogen-activated protein kinase kinase kinase 5

AXIN1 Axis inhibition protein 1

BAD Bcl-2 associated agonist of cell death

Bak Bcl-2 antagonist/killer

Bax Bcl-2 associated x protein

Bcl-2 B-cell CLL/lymphoma 2

BCSC Breast cancer suppressor candidate-1

Bim Bcl-2-like 11 apoptosis facilitator

BMP Bone morphogenic protein

BRAF v-Raf murine viral sarcoma oncogene homolog B1

BRAFi BRAF inhibitor

BREAK BRAF E and K Mutations (study)

BRIM BRAF Inhibitor in Melanoma (study)

BRN2 POU class 3 homeobox 2

xxv

CARD Caspase activation and recruitment domains

CCD Coiled-coil domain

CCND1 Cyclin D1

CDK4 Cyclin-dependent kinase 4

CDKN2A Cyclin-dependent kinase inhibitor 2A

ChIP Chromatin immunoprecipitation

CI Confidence interval or combination index

CONAN Copy number analysis (database)

COSMIC Catalogue of somatic mutations (database)

COT Cancer Osaka thyroid kinase/MAP3K8

CRAF Raf proto-oncogene serine/threonine-protein kinase

CSK C-src tyrosine kinase

CTLA-4 Cytotoxic T-lymphocyte antigen number 4

DMBA 7,12-Dimethylbenz(a)anthracene

DMFS Distant metastasis free survival dsDNA double stranded deoxyribonucleic acid

DTIC Dacarbazine

DUSP Dual specificity phosphatase

E2 Estradiol

E2F1 E2F transcription factor 1

EGFR Epidermal growth factor receptor

ER Estrogen receptor

ERBB3 v-erb-b2 erythoblastic leukaemia viral oncogene homolog

ERK Extracellular signal regulated kinase

ETV Ets variant

FAK Focal adhesion kinase

FBXW7 F-box/WD repeat-containing protein 7 xxvi

FDA Food and Drug Administration

FLT3 Fms-related tyrosine kinase 3

FOSL1 FOS-like antigen 1

FOXD3 Forkhead box D3

FOXO Forkhead box O

FTI Farnesyl transferase inhibitors

GAPDH Glyceraldehyde 3-phosphate dehydrogenase

GFP Green fluorescent protein cGMP Cyclic guanosine monophosphate

GLI2 GLI family zinc finger 2

GNAQ Guanine nucleotide binding protein (G protein), q polypeptide

Gp100 Glycoprotein 100

GRP78 Glucose-regulated protein 78

GTP Guanosine triphosphate

HCC Hepatocellular carcinoma

HEK Human embryonic kidney

HGF Hepatocyte growth factor

HIV Human immunodeficiency virus

HMB-45 Human melanoma black-45

HMEC Human mammary epithelial cell

HRAS Harvey rat sarcoma viral oncogene

IF Immunofluorescence

IFN Interferon

IFNB1 Interferon beta 1

IGF Insulin-like growth factor

IHC Immunohistochemistry

IMVS Institute of Medical and Veterinary Science xxvii

IR Ionizing radiation

IRF Interferon regulator factor

ISRE Interferon stimulated response element

JAK Janus Kinase

JARID1B Lysine (K)-specific demethylase 5B

JNK Jun N-terminal kinase cKit v-kit Hardy-Zuckerman 4 feline sarcoma viral oncogene homolog

KRAS Kirsten rat sarcoma viral oncogene

LDH Lactate dehydrogenase

LKB1 Serine/threonine kinase 11

LOH Loss of heterozygosity

MAP2K1 Dual specificity mitogen-activated protein kinase kinase 1

MAPK Mitogen-activated protein kinase

MC1R Melanocortin-1 receptor

MDM2 MDM2 oncogene, E3 ubiquitin protein ligase

MDMX Mdm4 p53 binding protein homolog (mouse)

MEK Mitogen-activated protein kinase kinase

MEKK1 Mitogen-activated protein kinase kinase kinase 1

MEN1 Multiple endocrine neoplasia 1

METRIC Members of the MEK versus DTIC or Taxol in Metastatic Melanoma (study)

MITF Microphtalmia-associated transcription factor

MLCK Myosin light chain kinase

MMP Matrix metalloproteinase

MRI Magnetic resonance imaging

MST-2 Serine/threonine kinase 2

MT Mutant

xxviii

MYC v-myc myelocytomatosis viral oncogene homolog (avian)

MYH9 Nonmuscle myosin IIa heavy chain

NALP3 NACHT, LRR and PYD domains-containing protein 3

NCBI National Centre for Biotechnology Information

NEDD Neural precursor cell expressed, developmentally down-regulated E3 ubiquitin protein ligase

NER Nucleotide excision repair

NK Natural killer

NRAS Neuroblastoma RAS viral oncogene homolog

OS Overall survival

OPN Osteopontin

PABA para-aminobenzoic acid

PAX3 Paired box 3

PBMC Peripheral blood mononuclear cell

PCNA Proliferating cell nuclear antigen

PD-1 Programmed death-1

PDE5A Phosphodiesterase 5A, cGMP specific

PDGFR Platelet derived growth factor receptor

PDH Pyruvate dehydrogenase

PD-1 Programmed death-1

PD-L1 Programmed death ligand-1

PDSS2 Prenyl diphosphate synthase subunit 2

PET Positron emission tomography

PI3K phosphatidylinositol-4,5-bisphosphate 3-kinase, catalytic subunit alpha

PKA Cyclic AMP dependent protein kinase A

PPP6C Serine/threonine-protein phosphatase 6 catalytic subunit

PREX1 Phosphatidylinositol 3,4,5-trisphosphate-dependent Rac exchanger 1 protein xxix

PRR Pattern recognition receptors

PTEN Phosphatase and tensin homolog

RAC1 Ras-related C3 botulinum toxin substrate 1

RAR Retinoic acid receptor

RARE Retinoic acid response element

Rb Retinoblastoma

RECIST Response evaluation criteria in solid tumours

RFP Red fluorescent protein

RFP Ret-finger protein (TRIM27)

RFS Recurrence free survival

RGP Radial growth phase

RIG1 Retinoic acid inducible gene 1

ROS Reactive oxygen species

RTK Receptor tyrosine kinase

SAR Structure activity relationship

SCC Squamous cell carcinoma

SEER Surveillance, Epidemiology, and End Result

SLUG Snail family zinc finger 2

SNAIL Snail family zinc finger 1

SPF Sun protection factor

SPRY SPIa/Ryanodine receptor

STAT Signal transducer and activator of transcription

STING Stimulator of interferon gene

SUMO Small ubiquitin-like modifier

TGFβ Transforming growth factor, beta

TLR Toll-like receptor

TMZ Temozolimide xxx

TNF Tumour necrosis factor mTOR mechanistic target of rapamycin (serine/threonine kinase)

TPA 12-O-tetradecanoylphorbol-13-acetate

TRAF TNF receptor associated family

TRIF TIR domain-containing adaptor inducing IFNβ

TRIM Tripartite motif

TTC4 Tetratricopeptide repeat protein 4

TYRP1 Tyrosinase-related protein 1

UVA Ultraviolet radiation A

UVB Ultraviolet radiation B

UVR Ultraviolet radiation

VEGF Vascular endothelial growth factor

VGP Vertical growth phase

WNT5A Wingless-type MMTV integration site family, member 5A

WT Wild-type

XBP1 X-box binding protein 1

xxxi

Publications in preparation from this thesis

1. Sutton S, Koach, J., Tan, O., Liu, B., Yosufi, B., Liu, T., McArthur, G., Zhang, XD., Scolyer, RA., Cheung, BB., Marshall, GM. TRIM16 suppresses melanoma cell migration and predicts improved prognosis in advanced melanoma The manuscript is under review by Oncotarget

2. Selina Sutton, Owen Tan, Richard A Scolyer, Jonathan B. Baell, Naresh Kumar, Grant McArthur, Xu Dong Zhang, Glenn M Marshall and Belamy B Cheung. Identification of a novel compound that synergistically inhibits cell growth with PLX4032 in BRAF wild-type melanoma cells. Manuscript in preparation.

3. Belamy B Cheung, Owen Tan, Jessica Koach, Sela Pouha, Selina Sutton, Louis Chesler, Michelle Haber, Murray D Norris, Tao Liu, Maria Kavallaris, Geraldine O’Neill, Glenn M Marshall. Targeting the Retinoic Acid Receptor α and Thymosin-β4 with a Combination of Fenretinide and Suberoylanilide Hydroxamic Acid Impairs Cell Survival and Migration in Neuroblastoma Cells. Manuscript in preparation

4. Murat Bingula, Glenn M. Marshall, Selina Sutton, Greg Arndt, Belamy B. Cheung, Naresh Kumara, David StC. Black. Synthesis, Characterization and Anti-cancer activity of New Hydrazide Derivatives incorporating different heterocyclic moieties. The manuscript was submitted to Bioorganic & Medicinal Chemistry.

Publications in collaboration during this thesis

5. Cheung B, Koach, J, Tan, O, Kim, P, Bell, JL, D'Andreti, C, Sutton, S, Malyukova, A, Sekeyre, E, Norris, M, Haber, M, Kavallaris, M, Cunningham, AM, Proby, C, Leigh, I, Wilmott, JS, Cooper, CL, Halliday, GM, Scolyer, RA, Marshall, GM. The retinoid signalling molecule, TRIM16, is repressed during squamous cell carcinoma skin carcinogenesis in vivo and reduces skin cancer cell migration in vitro. The Journal of Pathology 2012;226(3):451-462.

Presentations arising from this thesis

1. Selina Sutton, Owen Tan, Jessica Koach, Bing Liu, Benafsha Yosufi, Tao Liu, Grant McArthur, Xu Dong Zhang, Richard A Scolyer, Glenn M Marshall and Belamy B Cheung. High-level TRIM16 expression predicts improved prognosis in advanced melanoma. Oral presentation. TOW Awards UNSW 2013-joint first place winner

xxxii

2. Selina Sutton, Owen Tan, Jessica Koach, Bing Liu, Benafsha Yosufi, Tao Liu, Grant McArthur, Xu Dong Zhang, Richard A Scolyer, Glenn M Marshall and Belamy B Cheung. High-level TRIM16 expression predicts improved prognosis in advanced melanoma. Oral presentation. Australian Associated of Chinese Biomedical Scientists (AACBS) Scientific meeting 2013-winner

3. BB Cheung, SK Sutton, O Tan, Bing Liu, Xu Dong Zhang, RA Scolyer and GM Marshall. High TRIM16 expression predicts favourable outcome in patients with metastatic melanoma. The Lorne Cancer Conference, Lorne, Australia, 2013.

4. SK Sutton, O Tan, G Ardnt, XD Zhang, GM Marshall and BB Cheung, Identification of a novel compound that synergistically inhibits cell growth with PLX4032 in BRAF wild-type melanoma cells. Society for Melanoma Research International Conference (SMR), California, USA 2012

5. SK Sutton, O Tan, Xu Dong Zhang, RA Scolyer, GM Marshall and BB Cheung. Characterisation of a novel tumour suppressor TRIM16 in melanoma. Society for Melanoma Research (SMR) 2012 Congress. California, USA. 2012.

6. SK Sutton, O Tan, XD Zhang, RA Scolyer, GM Marshall and BB Cheung. Characterisation of a novel tumour suppressor TRIM16 in melanoma. Lorne conference, 2012

7. Belamy B Cheung, Jessica Koach, Jessica L Bell, Patrick Kim, Selina Sutton, Owen Tan, Eric O Sekyere, Alena Malyukova, Anne M Cunningham, Murray D Norris, Michelle Haber, Maria Kavallaris, Gary Halliday, Richard Scolyer and Glenn M Marshall. TRIM16: A novel activator of the retinoid anticancer signal and therapeutic target. The Lorne Cancer Conference, Lorne, Australia, 2011.

8. SK Sutton, GM Marshall, RA Scolyer and Belamy B Cheung. Characterization of TRIM16 as a novel tumour suppressor in melanoma. Tow Research meeting, Prince of Wales Hospital, Sydney, 2010. (Poster prize awarded)

9. SK Sutton, GM Marshall, RA Scolyer and Belamy B Cheung. Characterization of TRIM16 as a novel tumour suppressor in melanoma. 7th International Congress of the Society for Melanoma Research, Sydney Convention & Exhibition Centre, 2010

xxxiii

Acknowledgements

Thank you to my supervisors, Dr Belamy Cheung and Professor Glenn Marshall for their guidance and helpful suggestions in this work. Thanks to Dr Dan Carter and Dr Sonya Diakiw for friendly input and critical reading of this thesis. Thanks to Christopher Gardner for assistance in understanding compound structure and function. Thanks to Owen Tan and Jessica Koach for experimental assistance. Thank you also to the molecular carcinogenesis group for wonderful teamwork and sharing the highs and lows of medical research.

Thank you to my collaborators Professor Xu Dong Zhang and Professor Grant MacArthur for thoughtful input and support in this work and supplying cell lines. Thank you to Professors, Richard Scolyer and Georgina Long for supplying patient samples for analysis.

Lastly, thank you to my mum, Cherryl, dad, Malcolme and brothers, Kelly and Dale Sutton, and to numerous friends, for their encouragement and interest in my research and for letting me talk about the importance of sun safety at all times.

xxxiv

Chapter 1

Introduction

1.1 Neural crest development and melanocyte formation

1.1.1 The neural crest

The developing neural crest harbours a multi-potent cell population giving rise to neurons, glial cells, smooth muscle cells, medullary secretory cells, bone and cartilage cells and, melanocytes, and the neural crest is a transient structure (1-3). This wide array of cell types share the common origin of the neural crest, but develop into specialised cell types and are organised into specific lineages. Neural crest derived cells leave their origin through a delamination process and settle in their target organ to differentiate (4). The first cells emigrating from the neural crest are the neuronal and glial cells, followed by the melanocytes (5). Melanoblasts marked by the specific expression of tyrosinase-related protein (Tyrp1) precede the development of mature melanocytes (3) and require expression of the phosphatidylinositol 3,4,5-trisphosphate- dependent Rac exchanger 1 protein (P-Rex1) gene for migration (6). Melanoblasts express Microphtalmia-associated transcription factor (MITF) and Sox10 and proceed down the dorsolateral route of the neural crest becoming mature melanocytes via the ectoderm (Figure 1). MITF regulates melanocyte development and acts as an oncogene in melanoma (7). Melanocytes can also be derived from detachment of nerves innervating the skin (2). The fate of melanocyte development resides in the expression of MITF, which is specific to melanoblasts. The transcriptional repressor of MITF and FOXD3 is expressed exclusively in neural/glial precursors (1, 5). The over-all developmental process is governed by Wnt and bone morphogenic protein (BMP) signalling (1, 8). The unique migratory behaviour and role in epithelial to mesenchymal (EMT) transition of neural crest lineage cells highlights the propensity for invasion of these cell types in the context of cancer metastasis (4, 9, 10).

Melanocytes produce melanin by a process known as melanogenesis. Melanin is required to give tissues pigmentation. In the epidermis, Melanin acts as a shield for keratinocyte DNA against ultraviolet (UV) radiation from the sun (7). Consequently, the load of mutation present in melanomas is high compared to other cancers due to significant UV induced damage (11) mostly resulting in C>T somatic mutations (12). Melanocytes are intricately regulated by keratinocytes in the skin (13).

Figure 1.1. Development of melanocytes from the neural crest (Figure adapted from Thomas et al, 2008 (1) and Sommer et al, 2011 (2))

1.2 Melanoma

1.2.1 Melanoma incidence worldwide

The prevalence of melanoma is increasing worldwide (14-16). In one study statistics showed that the melanoma mortality increased in the United States at a rate of 1.5% annually from 1950 to 2005. This reaches approximately 8000 deaths per year from the estimated 76,000 new melanoma cases and 56,000 melanoma in situ cases in 2012 (17). Another study estimated that the incidence of melanoma in the United States has increased 8-fold in women, and 4-fold in men between 1970 and 2009 (18).

Epidemiological data from Americas National Cancer Institute’s Surveillance, Epidemiology, and End Result (SEER) registry and other studies highlights a gender difference in mortality rates in melanoma (19-21). Men comprise 40% of diagnosed melanoma cases, but a disproportionate 60% of melanoma deaths (21). In addition, newly diagnosed melanomas are thicker in men than women. Disease prognosis was equally poor for men as for women in the distant metastatic disease, though men with earlier stage melanoma deeper than 4mm had a poorer prognosis than women (21).

1.2.1.1 Melanoma incidence in Australia

Australia currently has the highest incidence of melanoma in the world (22). A comparative study recorded incidence rates in Australia is 40-60 cases per 100,000 inhabitants per year, United States is 10-20 cases per 100,000 inhabitants per year and Germany is 10-12 cases per 100,000 inhabitants per year in the 1990’s, which demonstrated Australia has the highest incidence by a significant margin (14, 23). The incidence of melanoma in Australia and New Zealand are around four times as high as those found in Canada, the US and the UK (AIHW). Worldwide, Queensland Australia has the highest incidence of invasive cutaneous melanoma at 55.8/105/year and 41.1/105/year for males and females, respectively (24). Melanoma is the fourth most common cancer in Australia with more than 10,300 cases diagnosed annually, of which 1250 result in death (Australian Cancer incidence and Mortality-Melanoma). The risk of developing a melanoma in women to age 85 was 1/24, but in men was 1/15 (Melanoma research institute). Furthermore, rates of melanoma have doubled in Australia during the 4

20 year period from 1986-2006. In Australia, over 40% of melanomas are associated with chronically sun-damaged skin (25).

In a large population based study in Queensland Australia, 26,736 patients diagnosed with thin melanomas (<1.00mm) had a 20 year survival of 96% (22). The greatest influences on prognosis were melanomas greater than 0.75mm and patients over 65 years of age compared to less than 25 years (22).

The lifetime risk of developing a melanoma will be 1/50 in the United States in 2015 as estimated (26). Currently the lifetime risk of developing melanoma in American men is 1/58, but is as high as 1/25 for Australian men (27).

1.2.1.2 Known risk factors for melanoma development

Melanoma is predominantly a cancer type associated with Ultraviolet radiation (UVR) exposure, but there are additional risk factors that influence an individual developing melanoma outlined below;

A large number (more than 50) of atypical nevi and a family history of melanoma Fair complexion and/or blue eyes, and red or fair hair High exposure to intermittent solar UV, particularly a single sunburn event under the age of 18 Countries in lower latitudes e.g. Australia have a 10-20 x greater risk for melanoma than European counterparts The incidence of malignant melanoma is higher in people with non-melanoma skin cancers (basal and squamous cell carcinomas), which is indicative of cumulative UV exposure Use of tanning beds delivering 5 x the UV exposure of solar sun exposure (Source: World Health Organisation (WHO) - Health effects of UV radiation)

It has recently been shown that obesity increases the risk of developing many cancers, including melanoma. This is thought to be due to the activation of melanoma related signalling pathways, MAPK, PI3K/AKT and STAT3 associated with an obese state (28). 5

This, combined with additional factors, is thought to contribute towards the increase in melanoma observed.

1.2.1.3 Ultraviolet radiation exposure

Ultraviolet radiation from the sun is made up of ultraviolet A and ultraviolet B (UVA and UVB) light at 320-400 nm and 280-320 nm respectively (17). UVB is the main culprit of DNA damage by inducing erythema and pyrimidine dimer formation. UVA has been implicated in the generation of reactive oxygen species causing an indirect form of DNA damage. UVA is implicated in photo aging and UVA2 is implicated in the increase of pigment darkening or ‘age spots’. UVA penetrates deeper into the dermis than UVB and produces DNA damage by generation of reactive oxygen species (ROS) (17). UVA comprises 95% of the UVR rays to reach the earth and UVB comprises the remaining 5%. The sun protection factor (SPF) is a measurement of the level of UVB protection unless a ‘broad spectrum’ is stated, which blocks both UVA and UVB radiation. The Cancer Council of Australia recommends an SPF 30 sunscreen which is broad spectrum and water resistant which blocks approximately 97% of the suns UVB rays (29). The deregulated multiple molecular pathways (MAPK and PTEN/AKT) in melanoma are associated with UV exposure (17). Additionally, sunburning of skin activates inflammasomes which are constitutively activated in melanomas (30-32). Inflammasomes are the first line of immune response to cell stress (33) and mediate secretion of interleukin-1β (30, 31, 34), leading to chronic inflammation and tissue damage when uncontrolled (34). UV exposure is one of the major risk factors for melanoma development (35).

Melanoma prevention measures include;

Regular skin examinations assessing the ABCDE of melanoma (self-examination or a skin cancer pathologist) Avoiding UV exposure during peak times (10am-4pm) Use of chemical or physical barriers to UV exposure in the form of sunscreens (Cancer Council of Australia)

1.2.1.4 Melanoma ABCDE

Cutaneous melanomas are typically identified by visible characteristics that separate a normal nevus from a malignant melanoma (36) (Figure 2).

A= Asymmetry within a mole

B= Border that is uneven or ragged in appearance

C= Colour variegation

D= Diameter that exceeds 6mm

E= A mole that is Evolving over time

Figure 1.2. The ABCDE’s of melanoma. Asymetry, Border, Colour, Diameter, Evolution

However, nodular melanoma, which is the most aggressive form of melanoma, is not governed by these observations. It is identified by being elevated above the skin surface, firm to the touch and growing. Nodular melanomas undergo the vertical growth phase without appreciable radial growth (10, 36). The majority of melanomas are de novo and do not arise from the oncogenic transformation of established nevi (37). The plasticity of melanoma cells enables them to adapt to the selective pressures in the microenvironment and to establish tumours (38).

1.3 Diagnosis of melanoma

1.3.1 Melanoma diagnosis and early prognosis

Analysis of a suspect mole or tissue from the skin or lymph node (or distant metastasis site) by a histopathologist is the primary means of melanoma diagnosis. In addition, staining of tissue specimens with antibodies S100, HMB-45 or Melan-A to confirm melanocytic tissue, coupled with observed malignant morphologic features (39). Cutaneous primary tumours are measured for Breslow thickness which is an indicator of the tumour depth measured in mm (40). Additional information is the assessment of Clark level, which is the level of melanoma cell invasion, though it is in contention whether Clark level retains prognostic significance in addition to Breslow thickness (41). To address this, an analysis was performed on 919 patients with stage I or II melanoma with an average follow up of 10.9 years. Melanoma thickness was classified into <0.75, 0.76-1.50, 1.51-4.00, and >4.00mm. Cross classification of Clark level II-V was given and the relevance of Clark level for each Breslow thickness was assessed by Kaplan-Meier survival curves. Additional prognostic factors were controlled for. The authors concluded that level of invasion (Clark level) was prognostic for melanoma survival and both Breslow and Clark levels provided useful prognostic information in the prognosis of stage I and II melanomas (41). Despite this, ulceration and mitotic rate remain the dominant prognostic factors for stage I and II melanomas (39, 42).

1.3.2 Response Evaluation Criteria in Solid Tumours (RECIST)

To evaluate therapeutic efficacy, Response Evaluation Criteria in Solid Tumours (RECIST) criteria is used as a clinical standard. RECIST describes criteria for evaluating lesions, lesion frequency, duration of overall response, duration of stable disease, progression free survival, time to progression and all standardized criteria required to assess a patient (43). A RECIST criterion is routinely applied in assessment in clinical trials.

1.3.3 Melanoma stage and 5 year survival rates:

Melanoma invades local tissue, and metastasizes to local lymph nodes, and the lung, liver and brain are common metastasis sites in the advanced disease. Widespread metastasis is the main cause of death in melanoma patients and staging is based on primary tumour thickness, ulceration, lymph node metastasis and distant metastasis (44) (Table 1). Extravasation of melanoma cells into organs is regulated by matrix metalloproteinases, adhesion molecules, growth factors and chemokine’s (44). Median overall survival of patients with melanoma metastasis is less than 1 year (45).

Table 1.1. Survival data based upon the American Joint Committee on Cancer (AJCC) five-year survival with proper treatment, seventh edition (2009).

Stage 0 Melanoma In situ Survival 99.9%

Clark level 1 Stage I

Survival 85-99%

<1.00 mm without ulceration; no lymph node involvement, no distant 1A Tumour metastases.

<1.00 mm with ulceration or Clark level IV or V tumour 1.01 – 2.0 mm 1B Tumour without ulceration; no lymph node involvement; no distant metastases.

Stage II

Survival 40-85%

1.01 – 2.0 mm with ulceration; tumour 2.01 – 4.0 mm without ulceration; IIA Tumour no lymph node involvement; no distant metastases.

IIB Tumour 2.01 – 4 mm with ulceration.

> 4.0 mm without ulceration; no lymph node involvement; no distant IIB Tumour metastases. IIC Tumour > 4.0 mm with ulceration; no nodal involvement; no distant metastases. Stage III

Survival 25-60%

Tumour of any thickness without ulceration with 1 positive lymph node IIIA and micrometastasis or macrometastasis.

IIIB Tumour of any thickness without ulceration with 2-3 positive lymph nodes and micrometastasis or macrometastasis.

Tumour of any thickness and macrometastasis OR in-transit IIIC met(s)/satellite(s) without metastatic lymph nodes, OR 4 or more metastatic lymph nodes, matted nodes or combinations of in-transit met(s)/satellite(s), OR ulcerated melanoma and metastatic lymph node(s). Stage IV

Survival 9-15%

Tumour of any thickness with any nodes and any metastases IV

1.3.4 Sentinel lymph node biopsy

The strategy of sentinel lymph node biopsy for melanoma staging was developed in the early 1990’s (46), by selectively removing the draining lymph nodes and surrounding nodes in the immediate vicinity of the tumour (47). Sentinel nodes are able to be identified 95% of the time and are subsequently stained for melanoma markers S-100, HMB-45 and Melan-A to determine the presence of melanocytic tissues.

1.4 Melanoma pathology

Melanomas in the skin undergo a radial growth phase before a vertical growth phase and invasion into the basement epidermis (10, 36, 48) (Figure 3). An exception to this is nodal melanomas that experience very little radial growth, but are highly aggressive and have a poor prognosis (SEER). Mutations in serine/threonine protein kinase B-Raf (BRAF) and neuroblastoma RAS viral oncogene homolog (NRAS) are an early event in melanoma development (49, 50), but alone are not enough to develop melanoma as many melanocytic proliferations (nevi) also harbour BRAF and NRAS mutation (51, 52). Additional mutations in CDKN2A, mutation or inactivation of PTEN with subsequent AKT pathway activation, AKT mutations and loss of the tumour suppressor activity of p53 are additional steps in the development of melanomagenesis (48) (Figure 3). In the skin, keratinocytes control melanocyte growth through paracrine growth factors and cell-cell adhesion molecules (53). Melanocytes escape keratinocyte regulated growth control by down-regulation of adhesion molecules such as E-cadherin. Down-regulation of E-cadherin increases melanoma-melanoma and melanoma- fibroblast adhesion molecule, N-cadherin, enhancing trans-endothelial migration. In addition, loss of cell anchorage also occurs due to changes in expression of integrin family members (53).

Figure 1.3. Melanoma pathogenesis and molecular aberrations. The progression of melanoma from the normal melanocytes in the skin, the development of the nevus commonly due to UV exposure, radial and vertical growth phases and melanoma invasion into the blood stream resulting in distant metastasis

1.4.1 The role MITF in melanocyte differentiation and melanoma development

MITF functions as a key regulator of melanocyte development by controlling genes encompassing differentiation and cell-cycle progression (7). It has also been identified as an oncogene by amplification in a subset of human melanomas. In addition, MITF and several MITF-regulated factors have been associated with the development of chemoresistance to dacarbazine and temozolomide in melanoma tumours and cell lines (54). In general, melanomas are a highly chemoresistance cancer type (51, 55, 56) thought to be due to the melanocyte lineage which is highly resistant to UV and reactive oxygen species (44). This dual nature of MITF towards promoting terminal differentiation and pigmentation in melanocyte development and the oncogenic potential in melanoma assign MITF the role as a “master regulator” of melanocyte development (7). Activating BRAF mutations down-regulate the expression of MITF resulting in a de-differentiated cell state (57). In a temperature sensitive MITF zebrafish mutant, low levels of MITF activity was demonstrated to be oncogenic with BRAFV600E and promoted melanomas reflective of human pathology (58). Abrogation of MITF activity in the mitfBRAFV600E model led to tumour regression, melanophage infiltration and apoptosis (58). This study illustrates that the dual nature of MITF is a potential melanoma therapeutic target, but that low levels of wild-type MITF activity is oncogenic in the cells.

A recurrent mutation has been identified in the MITF gene that is associated with both familial and sporadic melanoma (59). MITFE318K variant has been identified as an intermediate risk variant in melanoma families. This was identified in two large case- controlled cohorts from Australia and the UK (59). The E318K variant produces a MITF protein that has impaired sumoylation and differentially regulated target genes, and the mutant MITF protein is associated with a large population based study showed increased in risk of melanoma (59). MITFE318K is also associated with increased nevus count and non-blue eye colour (51).

In addition, ectopic expression of FOXD3 represses MITF expression in B16-F10 melanoma cells by inhibiting PAX3 binding to the MITF promoter, which can be

14 rescued by PAX3 over-expression (5), indicating FOXD3 as a controller of lineage direction in the neural crest (1, 5).

1.4.2 The MAPK pathway in normal melanocytes

The MAPK pathway is highly conserved between species and is responsible for relaying extracellular signals to modulate major cellular decisions such as proliferation, differentiation, migration and apoptosis (60). There are six MAPK groups presently identified in mammals; ERK1/2, ERK3/4, ERK5, ERK7/8, Jun N-terminal kinase (JNK)1/2/3 and the p38 isoforms α/β/γ(ERK6)/δ (60). For cancers to develop, the cell must enable itself to be independent from proliferation and anti-growth signals, evade apoptosis, gain unlimited replicative capability and gain the ability for invasion and metastasis with angiogenesis to ensure tumour growth (50). Additionally, melanomas that successfully overcome these barriers also develop distinct mechanisms to acquire drug resistance and evade oncogene induced senescence. Indeed, high phosphor-ERK levels in melanoma contribute markedly to chemoresistance (55). The MAPK pathway involving ERK1/2 is the predominantly affected pathway in melanoma pathogenesis (48, 61).

Figure 1.4. Signalling of the MAPK pathway in a normal melanocyte cells

The RAF-MEK-ERK pathway is a three tiered kinase cascade commonly referred to as the MAPK pathway and is important in regulating cell proliferation and survival (61). RAS proteins activate the RAF kinases (ARAF, BRAF and CRAF) and facilitate dimerization (61-63).

MAPK signalling in melanocytes is mediated through BRAF as opposed to CRAF. This is determined by accumulated cyclic AMP in normal cells with subsequent activation of protein kinase A (PKA) thereby inhibiting CRAF activity by phosphorylation of sites S43 and S233 (64, 65) (Figure 4). In normal melanocytes, activation of the MAPK pathway occurs through stimulation of growth factors, stem cell, fibroblast and hepatocyte growth factor (65). For a melanoma to switch to CRAF controlled signalling of the MAPK pathway, de-regulation of PKA and inactivation of BRAF must occur. This event is termed RAF isoform switching where the preferential isoform for signalling (BRAF) is switched in favour of CRAF (66). In addition, the MITF and AP-1 gene targets are regulated by ERK1/2 signalling (Figure 5) (67).

Figure 1.5. Downstream targets of ERK signalling in a normal melanocyte cell

1.4.3 Mutations in the MAPK pathway

The RAS-RAF-MEK-ERK pathway, commonly referred to as the MAPK pathway, is hyper-activated in a number of human cancers, typically due to mutation in KRAS, NRAS or BRAF (68).

Constitutive MAPK activation is essential for melanoma invasion and progression. Mutations in BRAF and NRAS are the primary mechanisms of MAPK pathway activation (68). However, these mutations alone are insufficient to initiate tumorigenesis as the majority of benign melanocytic nevi also harbour BRAF mutation (69, 70), but have undergone senescence as opposed to malignant tumour formation (52). NRASQ61K harbouring melanocytes act primarily via the pRb pathway confirmed by the loss of pRb displaying delayed onset of NRAS-induced cell cycle senescence (71). The mechanisms that allow for uncontrolled MAPK activation and melanoma development are varied and required additional pathway de-regulation for melanomagenesis to proceed. Interestingly, melanomas that are wild-type for BRAF/NRAS carry a higher mutation load than those harbouring BRAF or NRAS mutation (11).

It is important to note that mutation in the MAPK pathway is not the only mechanism of pathway hyperactivation. The pathway may be activated in some cancers (notably breast cancer) by oncogenic HER2 over-expression activating the MAPK pathway (72). However, HER2 has recently been proposed to serve as a target for cytotoxic T cells, showing favourable results in melanoma xenograft models (73).

1.4.4 BRAF and NRAS mutations

Mutation in BRAF and NRAS are mutually exclusive as both are responsible for MAPK pathway activation (74-76). Activating mutations in BRAF are directly responsible for the constitutive phosphorylation of MEK, while activating mutations in NRAS signal via CRAF to achieve MEK phosphorylation (77, 78). MEK is the only identified substrate of BRAF whereas activation of CRAF (also known as RAF1) has additional molecular targets (76). In addition to CRAF activation, mutations in NRAS activate the PI3K signalling pathway (76). Additional CRAF targets include MST-2, ASK-1 and

NFкB, and CRAF also regulates apoptosis through control of phosphor-BAD and Bcl-2 (79). The molecular signalling in BRAF and NRAS mutant melanomas is markedly different. An example of this is the ability of BRAF mutant melanoma to down-regulate cGMP-specific phosphodiesterase (PDE5A) via MEK and the transcription factor BRN2 (80, 81). This results in an increase of intracellular calcium, stimulating increased contractility and cellular invasion. This is not observed in NRAS mutant melanomas and is unique to melanoma harbouring the BRAF mutation (81). This highlights the difference between melanomas that utilise a common signalling pathway, but have different mutational backgrounds and is important to consider when applying a treatment strategy.

1.4.4.1 BRAF mutation

Melanomas arising from skin that is not chronically sun exposed harbour a 70% likelihood of BRAF mutation (16). Approximately 40-50% of uncultured cutaneous melanomas harbour the BRAFV600E mutation resulting in a constitutively activated BRAF protein due to the missense mutation at 1799T->A and protein change p.Val600Glu (77, 82-86). This results in a Valine to Glutamic acid substitution causing the BRAF protein to assume an active conformation and function as a phosphomimetic thereby constitutively phosphorylating downstream protein, MEK (48, 77, 83) (Figure 6). This mutation is found in the CR3 kinase domain, a region that is attributed >90% reported BRAF melanoma mutations (76, 87, 88). Other mutations including V600K and V600D/V600R mutations account for 16 and 3% of all BRAF mutations respectively (25). Mutations in BRAFV600K are associated with older age, male sex, and a shorter overall survival from diagnosis of stage IV disease compared to BRAFV600E (89). A small subset of patients harbours mutations in a location other than the 600 position. Approximately 1% of melanomas harbour either the D594G or G469E mutations (79). These variants demonstrate high levels of ERK phosphorylation, low MEK phosphorylation activity and MEK inhibitor resistance. These “low-activating” BRAF mutants require the presence of CRAF to transactivate MAPK signalling (25). Consequently, when these “low-activating” mutants are transfected into COS-1 cells, they are able to activate the MAPK pathway by binding to CRAF and causing CRAF

20 transactivation (90). BRAF mutation causes constitutive activation of the MAPK pathway by BRAF actively phosphorylating MEK in an unregulated manner. Activation of ERK in BRAFV600 mutant cells is RAS independent and does not require RAF dimer formation (61). Additional RAF proteins, ARAF and CRAF are not mutated in melanoma indicating their mechanism of action in promoting MAPK pathway activation is significantly different to that of BRAF (77). The MEK/ERK signalling cascade mediates effects on cell cycle entry by control of cyclin D1 and p27KIP1 (25). Additionally, the MAPK pathway has control over cell invasion and survival. Melanoma development is dependent on deregulation of the MAPK pathway and additional activation of the phosphatidylinositol-4,5-bisphosphate 3-kinase/murine thymoma viral oncogene homolog 1 (PI3K/AKT) pathway. Commonly, BRAF mutations are accompanied by activation of AKT3 from loss or mutation of PTEN (91, 92). Further, AKT3 activation has been associated with melanoma cell survival. Ectopic expression of constitutive AKT3 or endogenous expression of the mutant form of AKT3 protected melanoma cells from apoptosis induced by knockdown of BRAF or BRAF inhibitor (BRAFi) treatment with PLX4720 (93). In addition, melanoma cells that were intrinsically resistant to BRAF knockdown and BRAFi treatment demonstrated higher AKT phosphorylation in a 3-D collagen model, an effect that was reverse with AKT gene silencing (93).

Interestingly, the MAPK pathway can be over-stimulated in a manner that is dis- advantageous to melanoma cell proliferation (94, 95). An in vitro and in vivo model demonstrated that hyperactivation of ERK by BRAF displayed a senescent-like phenotype and induced autophagy in a manner specific to oncogenic BRAF melanoma cells that was not observed in NRAS mutant cells. In vivo, there was a positive correlation between autophagy marker light chain 3 and BRAF protein levels, and tumour growth was impaired with oncogenic BRAF over-expression (95).

Clinicopathologic features in primary melanomas reported to be associated with BRAFV600E/K mutations include fewer markers of chronic sun damage in surrounding skin, and higher total body nevus counts (96). Epidemiologically, BRAF mutations are most frequently identified in younger melanoma patients averaging between 44.7 and 55.3 years diagnosis (69, 89, 96) and most commonly located on the trunk (89). Childhood sun exposure is also a significant risk factor for BRAF mutant melanoma

21 development identified from a cohort of 251 melanoma patients (97). There is also a significant correlation between BRAF mutation and melanocortin-1 receptor (MC1R) variant in head and neck melanomas (98). Additionally, MC1R genotype is a predictor of early-onset melanoma (99).

Figure 1.6. Melanoma cells that harbour BRAF mutations have increased cell proliferation through constitutive MAPK signalling and PI3K pathway disruption

1.4.4.2 NRAS mutation

Mutations in NRAS are the second most common aberration in melanoma and are present in approximately 15-20% of cutaneous melanomas (7, 65, 85, 100). NRAS mutations were first identified in 1984 by screening for genes that indicated transforming properties in a melanoma cell line and identified 4/30 samples harbouring NRAS mutation (101). The patients with NRAS mutant melanoma tend to be older (>55 years), present with thicker melanomas and have often experienced chronic UV exposure (65, 98). This is in contrast to the intermittent UV exposure and younger age observed in the majority of BRAF mutation positive patients (102). Mutations in HRAS (Harvey Rat sarcoma Virus) and KRAS (Kristen rat sarcoma virus) occur at a much reduced frequency of 2 and 1% respectively (25, 103). In a murine model of oncogenic KRAS mice developed invasive tumours bearing histopathologic features similar to human melanoma progression established the ability of KRAS to be a founder event in melanomagenesis (103). RAS proteins exist in the inactive (GDP-bound) state or active (GTP-bound) state determined by a family of guanine nucleotide exchange factors (65). Mutations are most commonly located in NRAS exon2 (codon 61; Q61R, Q61K and Q61L, 82%) with the remainder located in exon1 (codon 12&13, 18%) (89). Activating NRAS mutations are typically located at p.Gln61 or p.Gly12 substituting to leucine similar to other RAS protein members (76, 88) with mutations located at p.Gln13 to leucine occurring with less frequency (65). These substitutions lock NRAS into the active (GTP-associated) conformation thereby causing a constitutively active protein. The picture is slightly more complex however, as the level of ERK phosphorylation in patients is variable and does not correspond with BRAF/NRAS mutational status (104). This suggests additional mechanisms impinging on this critical pathway. Importantly, RAS mutations can provide an oncogenic signal through both MAPK and PI3K/AKT pathways in contrast to BRAF mutations, which directly increase MAPK signalling but require additional mutations or loss of function of the PI3K/AKT pathway for melanoma development (105).

Mutational activation of RAS results in the dimerization and membrane localization of RAF family members, BRAF and CRAF (68). These mutations in NRAS confer a RAF isotype switch in which BRAF is no longer responsible for MAPK pathway activation, but CRAF becomes the primary kinase that induces ERK phosphorylation (77). CRAF 24 is the key mediator for mutant NRAS signalling (88) (Figure 7). The difference between the NRAS and BRAF mutational status of a melanoma may also be indicative of response to inhibitor therapy. BRAF mutant patients are more sensitive to MEK inhibitors than those harbouring the NRAS mutation (74).

NRAS mutations are also commonly observed in conjunction with p16INK4A promoter methylation (106). Patients harbouring the NRAS mutation typically have a shorter overall survival compared to patients that harbour wild-type NRAS (107).

Patients with BRAF mutation positive melanomas have a poorer prognosis than those with NRAS mutations overall, but this is partially offset by BRAF mutation positive patients having more treatment options available than NRAS melanoma patients (100). Despite this consideration, NRAS mutations status serves as an independent prognostic marker in advanced melanoma (108). In a cohort of 677 patients (47% BRAF mutant, 23% NRAS mutant, 32% wild-type for both oncogenes), patients harbouring NRAS mutation had a median survival of 8.2 months from stage IV diagnosis compared to the median survival of 15.1 months for NRASWT BRAFWT patients. NRAS mutation was also associated with decreased overall survival compared to wild-type patients (p=0.005) (108). An additional study of 79 melanoma stage III patients confirmed that absence of the BRAF mutation or NRAS mutations were favourable prognostic factors (109).

Figure 1.7. Melanoma cells that harbour the NRAS mutation result in melanoma cell proliferation through CRAF mediated MAPK activation and AKT activation

1.4.5 Activation of the PTEN-PI3K pathway and AKT

The PI3K/AKT/mTOR pathway is stimulated by growth factors and regulators and is involved in numerous cellular processes including cell metabolism, proliferation, migration, survival and angiogenesis (110-113). Numerous genetic and epigenetic aberrations that activate this pathway have been identified de novo (114). Co-operation of the PI3K pathway by pathway mutation or inactivation of PTEN with oncogenic drivers is frequently observed in cancer (111, 113). Melanomas seldom harbour mutations in the AKT genes, but regulatory protein, PTEN loss has been identified in 30% of cultures cell lines and approximately 10% of clinical melanomas examined (115). Increased phosphor-AKT is observed in 77% of melanoma metastasis compared to 17% of normal nevi in a cohort of 292 patient samples (110). Strong phosphor-AKT expression is also inversely correlated with overall 5-year survival in primary melanoma patients (110). De-regulation of PI3K/AKT signalling is almost exclusively associated with BRAF mutant melanomas with few NRAS mutant melanomas having PTEN mutation or silencing of negative regulators of this pathway (65). This is thought to be due to the ability of NRAS to activate PI3K/AKT pathway directly, though the activation of the PI3K/AKT pathway in NRAS melanomas is at a lower level than that pathway activation associated with BRAF mutant melanomas (65). Activation of the PI3K/AKT pathway is thought to be important for overcoming oncogene induced senescence seen in benign nevi that frequently harbour BRAF and NRAS mutations (70). In addition, AKT is able to transform melanocytes under hypoxic conditions. Around 50% of melanomas have constitutively active AKT3 with evidence of co- operation between AKT3 and BRAFV600E mutations. This is due to a gain of function mutation predominantly in AKT3 but mutations in AKT1 have also been observed in melanoma (113).

Activation of the AKT pathway has a wide range of cellular functions including activation of genes involved in immune activation, cell proliferation and survival, and apoptosis (92). AKT has been implicated in promoting melanoma cell survival by the direct phosphorylation of pro-apoptotic BAD (rendering the protein unable to bind Bcl- 2/Bcl-xL and blocking initiation of Bax/Bak mediated apoptosis) and negatively regulating Bim expression through phosphorylation of FOXO3a (93). This is one mechanism that melanoma cells evade apoptosis. 27

PTEN has been identified as a tumour suppressor protein in numerous cancers and functions as a negative regulator of the oncogenic PI3K/AKT signalling pathway which promotes cell survival and growth. PTEN deletion is a common event in the pathogenesis of melanoma (113). PTEN was initially identified as a potential tumour suppressor by the observation of frequent deletion of chromosome 10q23-24 identified in gliomas and melanomas (116). Mutations in PTEN are a less frequent event but are present in 12% of melanomas (117). PTEN mutations are a late event in melanomagenesis and are seldom found in primary melanomas. De-regulation of the PI3K/AKT pathway may be involved in the process of invasion and metastasis. Around 50% of melanoma metastasis has high PI3K-AKT signalling activity (85). Additional mutations in NRAS, c-KIT and PIK3CA impact this pathway inducing its activation (117). To strengthen this finding, mutations in the NRAS gene and PTEN gene seldom occur in tandem and are likely to contribute to the same functional outcome (102). Conversely, mutations in BRAF are frequently accompanied by PTEN mutations (116).

1.4.6 Oncogene induced senescence in BRAFV600E melanomas

It is well established that oncogenic BRAF induces melanocyte senescence in vitro and in vivo (118). Studies looking at metabolic profiling in melanoma cells have determined the mitochondrial gatekeeper, pyruvate dehydrogenase (PDH) is induced by oncogene BRAFV600E (119, 120). This induction was in parallel to the suppression of the PDH- inhibitory enzyme pyruvate dehydrogenase kinase 1 (PDK1) and the induction of the PDH-activating enzyme pyruvate dehydrogenase phosphatase 2 (PDP2) (120). Enforced normalization of either PDK1 or PDP2 expression abrogated the oncogene induced senescence effect and resulted in BRAFV600E mediated melanoma development (120). This mechanism is in addition to the observed depletion of PTEN abrogating BRAFV600E mediated senescence in human melanocytes and fibroblasts (70).

1.4.7 CDKN2A and p16 -CDK4-RB senescence barrier

Senescence due to telomere erosion, oncogene activity or reduced tumour suppressor activity in addition to apoptosis, are the two failsafe mechanisms a cell employs to evade tumourigenesis (121). The CDKN2A locus encodes two tumour suppressor proteins, p16INK4A and p19ARF. Alternative splicing of the exons in the CDKN2A locus produces tumour suppressor protein p16INK4A and ARF (p14ARF) (Figure 8) (48, 51). p16INK4A has found to be inactive in 25-40% of familial melanomas (122) due to mutation (19%), deletion or hypermethylation of the p16 promoter (48, 76, 117). Furthermore, the p16-CDK4-phospho-Retinoblastoma (Rb) pathway (Figure 8) is almost always deregulated in melanoma (116) with estimates at 90% of melanomas (123). De-regulation in this pathway by CDK4 up-regulation, results in phosphorylation, hence inhibition, of the Rb protein leading to G1-S cell cycle transition (123). Tumour suppressor protein, p16 is responsible for regulating the cell cycle at the G1-S checkpoint. Protein p16 inhibits CDK4 and suppresses the cell cycle progression of cells that have damaged DNA or oncogene activation (48). In animal studies mutations in either p16 or PTEN alone are inadequate to produce melanoma, but the two mutations together, or a combination with mutations in other melanoma related genes, NRAS or BRAF, results in melanoma development (48). Further genetic alteration in cyclin D1 (CCND1) causing amplification and cyclin dependant kinase 4 (CDK4) mutations occur exclusively from loss of p16 (76). Mutations in the genes of CDK4, p16 and Rb occur exclusively from each other, suggesting that each component of this pathway is integral for maintaining cell senescence and blocking melanomagenesis. Loss of CDKN2A is an early event and is typically found in the dysplastic nevus stage of melanomagenesis (48). Alongside MITF, both CDK4 and CDKN2A have been identified in familial melanomas (59, 122, 124). In addition, melanoma prone families with CDK4 mutations are associated with MC1R variants (124). In vivo, CDK-4 targeted mice develop highly invasive melanomas after carcinogen challenge (125).

Transcription

Figure 1.8. CDKN2A pathway control in melanoma. The CDKN2A gene gives rise to p16 and p14 which control cell cycle progression and apoptosis respectively.

1.4.8 Inactivation of ARF-p53 pathway senescence barrier

Tumour suppressor protein, p53 mediates a damage response in melanocytes by arresting the cell cycle at the G2M phase in response to DNA damage allowing for repair or apoptotic induction (48). p53 is strongly up-regulated in p16 deficient melanocytes that have reached senescence. This suggests that a secondary mechanism of cell senescence involving p53 must be overcome before melanocytes can proceed to melanoma. Though the mutations in p53 in melanoma are present in only 5-25% of melanomas (56), there is a frequent loss of p14ARF that positively regulates p53, which may be enough to negate the activity of p53 as a tumour suppressor (48, 56). In vivo, loss of ARF sensitized mice to UV induced melanoma development (126).

The lack of p53 mutation in melanoma can be understood by the fact that CDKN2A encodes both p16 and p14. Protein p14 interacts with MDM-2 which binds to p53 protein and ubiquinates p53 for targeted proteosomal degradation (116). Hence, while p53 mutation rate remains low, deregulation of p14-MDM2 ensures p53 protein is effectively targeted for degradation and is consequently unable to induce apoptosis (56).

1.4.8.1 Defective apoptosis in melanoma

The key regulator in apoptosis, p53 is mutated in approximately 19% of melanomas (117). Despite the majority of melanoma harbouring wild-type p53, the induction of apoptosis is de-regulated. Pro-apoptotic p53 target transcripts are under-expressed, while positive cell cycle regulators are over-expressed in human melanoma patient samples and cell lines (127).

Inhibition of apoptosis by p53 negative regulation is mediated by increased MDM2 or loss of p16 activity due to epigenetic repression via promoter methylation. In addition, MDMX is responsible for negatively regulating MDM2 and is frequently lost in melanoma allowing for un-regulated MDM2 activity and constitutive ubiquitination and degradation of p53. Negative regulator of and apoptosis antagonist of p53, MDM4, is frequently increased in primary and metastatic melanomas (128). Aside from induction of apoptosis, p53 is also responsible for the prevention of nevi progressing to melanoma

31 by is effects on the cell cycle (48). Melanoma anti-apoptotic mechanisms include activating mutation of the BRAF gene, sustained expression of Bcl-2 and methylation and loss of heterozygosity of the apoptosis activating factor-1 (APAF-1) gene which is a down-stream target of p53 (56, 129). p53 target transcripts are commonly involved in apoptosis are under-expressed as determined by whole genome bead array examination of 82 melanoma metastasis (56 differentially expressed genes) and 6 melanoma cell lines (34 differentially expressed genes) as compared to normal human melanocytes and fibroblasts (127). Conversely, genes involved in positive regulation of the cell cycle were over-expressed and this was independent of p53 mutational status (127).

Interestingly, tripartite motif 24 (TRIM24) has also been identified to bind to and ubiquitinate p53 after DNA damage. TRIM24 gene silencing by shRNA resulted in an increase of endogenous p53 while overexpression of TRIM24 promotes p53 ubiquitination and degradation (130).

1.4.9 KIT mutations

Melanomas that arise from acral, mucosal or skin exposed to chronic sun-damage demonstrate different clinical, histopathology and molecular pathology than cutaneous melanomas (131). c-Kit is found to be frequently mutated in these types of melanomas at rates of 39% mucosal (88, 132), 36% acral, (88, 131) and 28% chronic sun damage (76, 131). KIT mutations only contribute to 1% overall of melanoma mutations (76). Mutations occur in KIT exons 11, 13 and 17 (132). The c-kit receptor is responsible for signalling via the NRAS pathway. Kit inhibitor, imatinib mesylate induces response in patients with the KITK642E or KITL576P variants (88) but not all melanomas that harbour KIT mutations are responsive to KIT inhibitors. This suggests that in these melanomas, KIT may not always be the driver mutation influencing melanoma progression and survival (133). KIT mutation positive melanomas are typically treated with imatinib (84), though sorafenib has demonstrated some efficacy against this type of melanoma (132). KIT mutations harbouring the variant D816H in the kinase domain may be resistant to imatinib therapy (84).

1.4.10 Additional melanoma mutations

Uveal melanomas experience a high degree of GNAQ mutation located in the α-subunit of the Gq family. These types of melanomas rarely harbour c-Kit, NRAS or BRAF mutations, though GNAQ mutant melanomas are sensitive to inhibition of MEK (131). Recently identified mutations in Ras-related C3 botulinum toxin substrate 1 (RAC1) were determined by exome sequencing at a rate of 12% in sun exposed melanomas (12) and regulates the NF-кB pathway by PIP3 regulation (134). The Serine/threonine- protein phosphatase 6 catalytic subunit (PPP6C) gene is newly identified with a mutation rate at 9% (12, 117), Dual specificity mitogen-activated protein kinase kinase 1 (MAP2K1) at 5% and F-box/WD repeat-containing protein 7 (FBXW7) at 3% among other genes (117).

1.4.11 Melanoma migration and metastasis

Melanomas undergo metastasis at a greater rate than other cancers is due to the innate lineage-specific biology of melanocytes (6, 10, 135) and, interestingly, require P-rex1 for migration; P-rex1 is also required for melanoblast migration from the neural crest (6). Primary melanomas that are proliferative at the dermal-epidermal junction must gain the ability to migrate into surrounding tissues for disease progression (36). Typically, dermal invasive melanoma is observed in a pagetoid spread or ‘buck- shotting’ into the epidermis (36, 136). Concurrently, the melanoma becomes invasive to the dermal layers. Metastasis occurs in approximately 10-15% of patients with cutaneous melanoma (83). At the molecular level, changes in transcription factors pertaining to epithelial-mesenchymal transition (EMT), and expression of fibronectin, vimentin, metalloproteinases (137) and snail (138) are involved. There is cross-talk between signaling pathways NFкB, MAPK and PI3K/AKT which are all de-regulated in BRAF mutant melanomas (137).

The β-catenin/WNT pathway is frequently activated in a constituitive manner in melanoma (80, 139, 140) and is associated with chemoresistance (139). Furthermore, WNT signalling potentiates nevogenesis and suppression of WNT signalling is a factor in melanocyte senescence (141). β-catenin has a dual activity in melanoma cells both 33 strengthening cadherin adhesion at the plasma membrane and activating transcription of various genes in the nucleus (142). WNT/β catenin activation from the inhibition of oncogenic BRAF is able to induce apoptosis, a process mediated by AXIN1 (140). Beta-catenin induced by canonical Wnt signaling is activated in melanoma and induces metastasis (143, 144) but inhibits melanocyte migration (143). This observation was confirmed in vivo, in which β-catenin signaling in melanoblasts reduced cell migration in a manner dependant on MITF and src inhibitor, CSK. In addition, β-catenin promoted lung metastasis in an NRAS driven melanoma model (143). The researchers propose β- catenin may have dual roles of repression of melanoma cell migration and promotion of metastasis (143). Additionally, β-catenin induces immortalization of melanocytes by suppression of p16INK4A and works with NRAS in melanoma development (145).

Wnt signaling in melanoma involved multiple proteins and is complex with the ability to signal via canonical and non-canonical pathways (142). WNT5A and ARF6 have been identified as having a role in driving β-catenin expression and by so doing, melanoma metastasis. ARF6 has been shown to stimulate β-catenin transcriptionally during WNT5A mediated melanoma invasion (142, 146). WNT5A is a direct effector of melanoma migration and metastasis (147).

The activation of the transforming growth factor-β (TGFβ) pathway is frequently constituitively activated in melanoma and is another mechanism by which melanoma cells promote invasion and metastasis. TGFβ transcriptional target, GLI2 is an effector protein of the hedgehog pathway. Increased GLI2 expression is associated with loss of E-cadherin expression and melanoma cell invasiveness by matrigel invasion assays as well as observed bone metastasis in vivo (148). Furthermore, reduction of GLI2 by shRNA inhibited basal and TGFβ-induced cell invasion, matrix metalloproteinase (MMP) secretion associated with invasion, and bone metastasis (148). Recently identified gene, ACP5 has been shown to confer spontaneous metastasis in vivo and also be prognostic of primary human melanomas (149).

Melanoma migration is mediated by EMT in which there is a transcriptional repression of E-cadherin by SLUG transcriptional regulation of ZEB1 and induction of factors positively regulating migration (150-152). The induction of ZEB1 is specific to SLUG as other EMT regulators, Snail and Twist did not induce ZEB1 transcription (151). This

34 suggests a heirarchy of organization of transcription factors involved in the EMT process (151).

Importantly, stromal factors also influence melanoma progression and metastasis. Osteopontin (OPN) is a secreated chemokine-like phosphoglycoprotein that facilitates cell-matrix adhesion and tumour progression (153). In an OPN knockout mouse model, reduced melanoma growth, angiogenesis and metastasis are observed. OPN induces VEGF, and ERK1/2 expression in B16 mouse melanoma cells and stromal derived OPN has been shown to regulate melanoma side populations phenotype through ERK2 activation (153). OPN has been proposed as a marker of poor prognosis in melanoma and is associated with increased tumour thickness, clark level and mitotic index in a cohort of 345 patients of primary cutaneous melanoma (154). Additionally, RAS-ERK and PI3K/AKT activator, Gab2, is found to be overexpressed in approximately 50% of melanomas and amplified in 11% and correlates with metastatic potential in melanomas (155).

1.5 Epigenetic modification in melanoma

Epigenetic gene repression by methylation is one mechanism by which melanoma cells abrogate the expression of genes that regulate cell cycle, DNA repair, and pathway regulation, resulting in de-regulation of key signalling pathways (156). Melanomas experience methylation in tumour suppressor genes that have been identified in other cancer types namely, RARβ, MGMT, PTEN and cell cycle regulators CDKN2A and 1B (157, 158). In addition, notable methylation is melanoma is found on genes CDH1 (158), COL1A2, RASSF1 (156), DAPK1 and ER-α (42-86%) (157, 159, 160) (Table 2). DNA modification in the form of loss of 5-hydroxymethylcytosine is a diagnostic and prognostic indicator in melanoma. This is thought to occur by the down-regulation of IDH2 and enzyme members of the TET family that regulate the conversion of 5- methylcytosine (5-mC) to 5-hydroxymethylcytosine (5-hmC) (161). Use of epigenetic drugs in melanoma can modulate the immunophenotype of melanoma cells (162) and may represent an as yet under-utilised anti-melanoma therapy.

Table 1.2. Percentage of DNA methylation of genes in melanoma Data source: Ongenaert, M. et al, 2008 (163)

Malignant cutaneous melanoma primary patient data Methylation Gene % Scale

RARβ 70% 0

PYCARD 63% 0-20 %

PTEN 62% 20-40 %

RASSF1 57% 40-60 %

ESR1 51% 60-80 %

MGMT 34% 80-100 % IGFBP3 32% CDKN2A 30% DAPK1 19% CDKN1B 9%

1.6 Current melanoma therapies

1.6.1 Surgery, radiation therapy and chemotherapy

Surgical excision of melanoma is a highly effective treatment for localized melanomas (84). Early surgical treatment of melanoma leads to a 90% cure rate, with unresectable advanced melanomas being very difficult to treat with chemotherapies (51, 56, 164). Once the melanoma has undergone the VGP and invaded the basement membrane and dermis, the effectiveness of surgical intervention significantly decreases (93) but may be used for palliative care (165). For early stage melanomas, surgical excision of a suspect mole is the first line of treatment. A wide excision margin is taken to ensure any potential migrating cancer cells are also removed. Surgery is also a first line of treatment for metastasized melanoma in the excision of regional lymph nodes often as a precaution to prevent further spread of disease, or due to the positive identification of melanoma cells in PET or MRI scans. Melanoma is typically resistant to conventional chemotherapy and requires a combinatorial targeted therapy approach for effective treatment (55, 84).

Radiotherapy has a useful place beneath surgery and systemic therapies. Radiation therapy may be used for metastatic melanoma that has formed tumours on the brain (166, 167). This involved using a localized dose of stereotactic radiation to target the tumour directly while minimising the side effects for surrounding tissues. It is presently used in melanoma in adjuvant settings and palliative care situations (165, 168). Recent data has shown benefit with combination radiotherapy and oncogenic BRAF inhibitor treatment with minimal acute toxicity (169).

Until recently, the standard of care for metastatic melanoma was the alkylating agent Dacarbazine (DTIC). Dacarbazine and its structural analogue, Temozolomide (TMZ) are pro-drugs producing the metabolite, 5-(3-methyltriazen-1-yl)imidazole-4- carboximide (MTIC) which is the active compound facilitating DNA alkylation (165, 170, 171). DTIC was routinely used as a single agent treatment in advanced melanoma, functioning by the addition of an alkyl group (CnH2n+1) to DNA and inducing cell death and is generally well tolerated (165). The DNA alkylation occurs at the N-7 or O-6 of guanine residues. DTIC was approved for the treatment of metastatic melanoma by the Food and Drug Administration of America (FDA) in 1975 (165). TMZ is not yet 37 approved for the use in the treatment of metastatic melanoma despite proven efficacy (165, 171), but is used as an off-label treatment (172). In a phase III clinical trial of 305 metastatic melanoma patients, TMZ demonstrated a longer progression free survival (1.9 months) compared to DTIC (1.5 months) and increased median survival of 7.7 months compared to DTIC at 6.4 months (171). In addition, TMZ has the added advantage of penetration of the central nervous system and can be administered orally (47).

Interestingly, a single institution study found that low BRAF and NRAS expression levels exhibited a clinical and prognostic benefit from DTIC therapy identified by objective response or stable disease at 3 months (107). The efficacy of DTIC and TMZ chemotherapies depends on low O-6-methylguanine-DNA methyltransferase repair and on high mismatch repair in melanoma cells (17). Nucleotide excision repair (NER) genes correlate with poor patient prognosis due to the ability for aggressive melanomas to repair DNA damage induced by chemotherapies (DTIC and TMZ) and maintain enough genomic integrity for the tumour to progress (17).

Until 2011, DTIC was the most effective anti-melanoma agent producing a result in 15- 20% of patients with a median duration of response of four months and was the standard-of-care for advanced melanoma patients (47, 173).

Current frontline treatment for BRAFV600E melanoma includes the recently approved BRAF inhibitor, PLX4032 (Vemurafenib). This compound was approved the FDA in 2011 in adults in the absence of brain metastasis (174). Vemurafenib has not been widely studied in patients with brain metastasis, but case studies have demonstrated efficacy of vemurafenib against lesions in the brain (174). It has a response rate of 81% for BRAFV600E positive melanomas. It has also extended median survival by an average of 7 months (range 2-18 months). A severe side effect not uncommonly observed is the development of squamous cell carcinomas. The median progression free survival for patients receiving vemurafenib (960mg orally twice daily) vs. Dacarbazine (1,000mg per m2 intravenously once every 3 weeks) is 5.3 and 1.6 months respectively (175). As clinical trials using vemurafenib have typically excluded patients with brain metastasis, a study conducted by the Mayo clinic in Minnesota report of three patients treated with vemurafenib that experience disease regression in extra cranial tumours, but disease

38 progression in brain metastasis. As yet, there is no clear consensus on the efficacy of vemurafenib for melanoma brain metastasis (174).

1.6.2 Immunotherapy

Melanoma is a highly immunogenic tumour type and some cases of sporadic disease regression due to host immune response have been recorded. However, the molecular pathology of melanoma is such that in the majority of cases it evades immune surveillance and the disease progression occurs. The aim of immune based therapy is to trigger the immune system to infiltrate melanoma tumour tissue with activated T-cells with cancer specific cytotoxicity (133).

Despite many melanomas displaying tumour-associated antigens able to be recognised by the immune system, the majority of patients become tolerant to these antigens at an early stage of disease (47). Immunotherapy can induce serologic and clinical manifestations of autoimmunity (176), providing long-lasting immune responses as the immune system memory is corrected to targeting melanoma (177). In a clinical trial of 200 patients treated with high-dose intravenous interferon alfa-2b adjuvant treatment, 26% of patients (52/200) demonstrated expression of autoantibodies. The median relapse free survival was 16 months for patients without autoimmunity (108/148 experienced relapse) and was not reached at a median follow up of 45.6 months among patients with evidence of autoimmunity (7/52 experienced relapse) (176). This indicated that auto immunity is an independent prognostic marker of relapse free and overall survival (176). Examination of a 79 melanoma stage III patient cohort identified a 46 gene expression signature bearing a strong representation of immune response genes was associated with favourable survival at >100 months for those patients having the signature and only 10 months in the absence of the gene signature (109).

High dose Interleukin-2 (IL-2) is also used in the treatment of metastatic melanoma and is typically administered as a bolus intravenous dose (178, 179). High dose bolus IL-2 was approved by the FDA for treatment of metastatic melanoma in 1998 (165, 180) and is able to mediate an objective response associated with survival beyond 4 years in approximately 13% of patients (178). Given that treatment of IL-2 has the potential for

39 a small subset of patients to achieve complete, durable response this provides ‘proof of concept’ for the role of immunotherapies in melanoma treatment (165). Interestingly, patients harbouring the NRAS mutation are more likely to respond to IL-2 treatment than those with BRAF mutant/NRASWT disposition (181).

Ipilimumab (anti-CTLA-4) is a frontline melanoma therapy, known clinically as Yervoy. Ipilimumab was approved for the treatment of metastatic melanoma by the FDA in 2011 and modulates the immune system against melanoma (182, 183). Its target, CTLA4 is required for the development of regulatory T-cells and when active, expresses an inhibitory signal resulting in a decrease in immune responsiveness (165, 182, 184). Anti-CTLA-4 blocks the inhibitory signal between T-cell and dendritic cell, resulting in activation of the T-cell (184-186). Ipilimumab works by monoclonal antibody blockade of CTLA-4 and subsequently increases patient immune response to melanoma tumours (183, 184). Animal studies confirmed that mice receiving CTLA-4 antibody gained the ability to reject melanoma tumours in xenograft models (187). It has been demonstrated to cause tumour shrinkage, though autoimmune based toxicity is problematic for the majority of patients receiving treatment (76). It has a low response rate of approximately 10% (65), with responding patients demonstrating long-lasting benefits (of several years) after a 3 month treatment course (175, 183). The 2 and 3 year survival rates are twice as high as with the glycoprotein 100 (gp100) peptide vaccine alone giving significant benefit to a minority of patients (183).

In phase III clinical trial, improved median overall survival of 10.0 months was noted in patients receiving ipilimumab compared to 6.4 months in patients receiving the gp100 peptide vaccine alone (188). The study consisted of 676 patients with unresectable stage III or IV melanoma with progressive disease after prior therapy. Study arms included ipilumamib alone (137 patients), ipilimumab + gp100 (403 patients) and gp100 alone (136 patients). There was no improvement in overall survival with ipilimumab (188).

A similar CTLA-4 blocking antibody, human IgG2 tremelimumab, is also in phase III clinical development. In this trial, 655 patients with treatment naïve stage III or IV metastatic melanoma were enrolled (189). Median overall survival was 12.6 months (95% CI, 10.8 to 14.3) for tremelimumab treated patients and 10.7 months (95% CI, 9.36 to 11.96) for the chemotherapy (single agent DITC or TMZ) arm (189).

Importantly, the duration of response for the tremelimumab was significantly longer than the chemotherapy arm (35.8 v 13.7 months; P = .0011).

Anti-PD-1, clinically known as nivolumab, is similar to ipilimumab and tremelimumab as it is also a negative regulator of T cells (133, 190). Nivolumab is a human IgG4 antibody blocking PD-1 and PD-L1. The mechanism of action blocks cancer cell PD-1 and PD-L1 ligands from interacting with T-cell surface receptors (190). This blocks the negative feedback from receptor signalling in the T-cell and promotes immune responsiveness to the cancer tissue (185, 186). Treatment with nivolumab achieved a durable (greater than 1 year) objective response in 30% of patients in phase I clinical trial (133). Immune blocking therapies targeting both PD-1 and CTLA-4 (nivolumab and ipilumimab) are currently in phase I/II clinical trial (191). Preliminary data suggests that expression of PD-L1 may select for patients with an improved response to PD-1 inhibitors (177). Initial findings from phase I clinical trial indicates an objective response of 53% for patients that received lambrolizumab (anti-PD-1) at 10mg/kg every two weeks with prior treatment of ipilumamab (192). All nine of the responding patients had more than 80% tumour shrinkage and three complete responders were observed. Single agent alone does not produce a high number of complete responses (193). Responses were durable in that majority of patients with 81% of those that had a response (42/52 patients) still receiving treatment at the median follow-up of 11 months (192). Anti-PD-1 produces an immune response in one in four to one in five patients (194).

1.6.3 Key targeted therapies

1.6.3.1 Drug targeting of BRAF

The discovery of the presence of unique mutations in BRAF has led to the development of specific agents that preferentially target mutant BRAF creating a therapeutic window for the MAPK pathway to be inhibited in mutant melanoma (195). This is due to the specific targeting of the mutant BRAF kinase that does not significantly affect MAPK signalling in normal tissues (63, 195). Selective inhibitors of BRAF are presently a standard of care for BRAFV600 mutant patients (196). These inhibitors have achieved remarkable clinical success, inducing tumour regression in the majority of patients 41 treated. Similar screenings to identify agents that can target mutant NRAS have proved technically difficult and a highly successful agent specific to this type of melanoma has yet to be discovered (65).

1.6.3.2 Sorafenib

Sorafenib is an orally available multi-kinase inhibitor with a preference towards CRAF and a degree of activity towards both BRAFV600E mutant and BRAFWT forms at an IC50 of 38 and 25 nM respectively (197). The in vitro efficacy against CRAF is an IC50 of 6 nM. Additional kinase targets include the inhibition of vascular endothelial growth factor receptors (VEGF1, VEGF2 and VEGF3) which impairs angiogenesis and platelet derived growth factor receptors (PDGFRs) (25). Tumour progression is further targeted by inhibition of FLT3 and c-kit (197). Sorafenib induces apoptosis by modulating Bcl-2 expression that is independent of MEK (79).

In a phase II clinical trial of Sorafenib as a monotherapy, a dose of 400mg twice daily was given to 36 treatment naïve patients continuously. One patient experienced a RECIST partial response lasting 175 days, three patient’s experienced stable disease with an average duration of 37 weeks (198). Sorafenib is not considered an effective single agent in the treatment of melanoma due to its off-target effects and the complexity of RAF dimer interactions that require BRAF inhibitors to be specific to the mutant form, or risk transactivation via CRAF dimerization (87).

1.6.3.3 Development of Vemurafenib

The difluorophenylsulfonamine, PLX4720, was the first generation of specific BRAF inhibitor to be highly selective for the mutant (V600E) form of the BRAF protein at 13 nM (199, 200). This was developed in the search for a specific and potent inhibitor of BRAFV600E in a structure guided discovery approach (199, 201). PLX4720 preferentially inhibits BRAFV600E over other kinases and induces cytotoxic effects by way of cell cycle arrest and apoptosis, specific to mutant cells and also demonstrates a strong inhibition of phosphor-ERK (199). PLX4720 preferentially targets BRAFV600E

42 selectively (IC50 13 nM mutant vs. IC50 160 nM wild type BRAF) and was found to crystallize with BRAFV600E (201). The analogue of the compound PLX4032 V600E (vermurafenib) displays a higher degree of selectivity for BRAF with an IC50 of 31 nM, CRAF IC50 48 nM and wild type BRAF IC50 100 nM (88). Doses of vemurafenib that achieved a >90% reduction of phosphorylated ERK were required to ensure a relevant clinical response. It is specific to the BRAFV600E form of melanoma and does not have efficacy against BRAFWT melanomas (202).

PLX4720 has an IC50 in the nM range, high specificity for mutant BRAF and was found to crystallize with BRAFV600E (201). PLX4720 has been shown to increase mitochondrial density in BRAF mutant cells, but not in MeWo wild-type BRAF melanoma cells (203). Furthermore, PLX4720 increases the production of lactate in mutant, but not wild-type BRAF cells. This increase in mitochondrial biogenesis is also associated with the increase in the mitochondrial master regulator, PGC1α which these researchers demonstrate to be regulated by BRAF target, MITF (203). Treatment of the BRAF mutant melanomas with BRAF inhibitors resulted in their dependence on oxidative phosphorylation. This suggests that the BRAF inhibitor has effects on cell metabolism in addition to the specific role of inhibiting mutant BRAF. Mitochondrial uncouplers also enhance the efficacy of PLX4720 in BRAF mutant tumours (203). PLX4720 demonstrates induction of apoptosis, reduced cell proliferation and decreased ERK phosphorylation consistent with an inhibitor targeting the ATP pocket binding domain of mutant BRAF melanoma cells (87). Further studies confirmed the in vivo significance showing tumour regression and apoptotic induction in xenograft models (87). Screening of a small molecule panel resulted in PLX4032 (vemurafenib) being identified as efficacious in like-manner to PLX4720 and vemurafenib was developed through to clinical use due to a more favourable pharmacokinetic profile (87). Importantly, vemurafenib retains the active sulphonamide moiety of PLX4720 (88).

Mutations in BRAFV600E involve a Valine (V) to Glutamic acids (E) substitution causing mutant BRAF to act as a phosphomimetic and induce constitutive MAPK signalling (75). BRAFV600E is present in 50-60% of cutaneous melanomas and accounts for approximately 90% of all BRAF mutations (76, 200, 204). Substitutions in V600K/D account for a 10-20% of remaining observed BRAF mutations and, in like-manner, lead 43 to constitutive MAPK signalling (75, 96, 205). Additional mutations have also been described in the glycine rich loop and the activation segment in the kinase domain, though these are rare (200). Inhibition of BRAFV600E results in melanoma cells death by apoptosis (75). Vemurafenib is effective against the mutant form of BRAF (V600E/K/D), and is ineffective against the BRAFWT form (206). In fact, a paradoxical signalling occurs when NRAS mutant, BRAFWT melanoma cells are treated with vemurafenib in which ERK signalling is activated due to raf isoform switching (206). In general, phosphorylation of ERK is regulated by negative feedback loops including the up-regulation of the dual specific phosphatases (DUSPs) which promote ERK dephosphorylation (206). Because of this, assessment of melanoma proliferative state based on degrees of ERK phosphorylation is a poor indicator of ERK signalling output and it is suggested that measurement of downstream effectors (cyclin D1, CCDN1), transformation mediators (ETV1, ETV5, FOSL1 and MYC) and feedback regulators (SPRY2 and DUSP6) may more accurately reflect cellular signalling (206).

Vemurafenib induces apoptosis and cell cycle arrest in sensitive melanoma cell lines, but only cell cycle arrest in resistant and less sensitive lines (25). It was recently found that vemurafenib induces endoplasmic reticulum (ER) stress-mediated apoptosis in cells harbouring the BRAFV600E mutation (207). On exposure of mutant cells to vemurafenib, a decrease in anti-apoptotic proteins was observed alongside a decrease in ER chaperone protein glucose-regulated protein 78 (GRP78) and an increase in the ER stress activator, X-box binding protein-1 (XBP1). Subsequent increase in translation initiation factor, eIF2α and ER stress-related genes was observed. Gene silencing of ER stress response protein, activating transcription factor 4 (ATF4) abrogated vemurafenib induced apoptosis (207). Tumour regression induced clinically by vemurafenib requires almost complete suppression of ERK signalling to be achieved for clinical efficacy (208).

In addition to the induction of apoptosis, senescence is also induced by vemurafenib. This is characterized by heterochromatin formation, an increased in β-galactosidase activity and changes in cell shape (209). This may contribute towards the observation that some patients do not experience a complete or durable response to vemurafenib treatment and only a portion of cells undergo apoptosis it is hypothesized that the

44 remaining cells are employing drug resistant mechanisms and a portion may be in a senescent state (209).

Inhibition of BRAFV600E by PLX4720 results in the up-regulation of pro-apoptotic Bim protein isoforms; BimEL, BimL and Bims (210). Jiang and colleagues identified Bims preferential splicing is achieved with BRAF inhibition by vemurafenib (210). To strengthen this finding, they also determined that enforced expression of BRAFV600E in wild-type BRAF melanoma cells and melanocytes resulted in inhibition of Bims expression (210).

1.6.3.4 NRAS targeting

Farnesyl transferase is responsible for adding a farnesyl group to RAS proteins ensuring full activation (88). Employment of farnesyl transferase inhibitors (FTI’s) to target oncogenic RAS are able to be circumvented by geranylgeranyltransferases within cells, preserving RAS function despite inhibitor presence (56, 88). This lack of progress in development of farnesyl-transferase inhibitors is due to the fact that multiple crucial cellular proteins are farnesylated (giving off-target effects) and additionally, RAS is able to be membrane targeted in the absence of farnesyl-transferases (65). Directly targeting RAS proto-oncogene for inhibition in melanoma has been unsuccessful so far (211).

In the absence of directly targeting NRAS, a focus on the downstream effectors of RAS activity (MEK inhibitions and AKT/PI3K inhibitors) has been the focus of development of clinically relevant therapeutics (133, 212). Presently, these are primarily agents that target the MAPK pathway, a key effector of which is MEK. This strategy is problematic as these pathways control many cellular functions and specificity of the agent towards melanoma with minimal impact on normal tissues is difficult (57). Searching for effective agents that widen the therapeutic window is currently the focus for targeting NRAS melanoma therapy.

The first molecularly targeted therapy to show activity in NRAS mutant melanomas is the MEK162 small molecule MEK1/2 inhibitor clinical trial (213, 214). This trial involved both BRAF and NRAS mutant melanoma patients and three study arms; 45mg 45

MEK162 twice daily for BRAF mutant patients, 45mg twice daily for NRAS mutant patients and 60mg twice daily for BRAF mutant patients. Six (20%) of 30 patients harbouring the NRAS mutation achieved a partial response alongside 8 of 41 BRAF mutant patients (20%). No patients achieved a complete response but this study demonstrated a targeted therapy that had clinical efficacy in both BRAF and NRAS melanoma patients and is the first to show targeted efficacy for NRAS patients (213, 214).

1.6.3.5 PI3K/AKT pathway targeting

In NRAS mutant melanomas, the NRAS mutation results in the activation of both the MAPK pathway and the PI3K/AKT pathway (65, 215). The direct activation of PI3K occurs by RAS directly binding catalytic subunit, p110, establishing a link between the two proteins (113). Furthermore, activation of the PI3K/AKT pathway is required for melanoma tumour progression and resistance to anti-neoplastic drugs (111, 112) and a gain of metastatic competence (111). Inhibitors of the PI3K pathway include the development of rapalogues (mTOR allosteric inhibitors) which are rapamycin analogues (111, 112).

Some inhibitors such as LY294002 and wortmannin target the whole PI3K family and related proteins (mTOR, PI4K and DNA-PK) resulting in high toxicity and for LY294002, low potency in pre-clinical models (215, 216). One candidate drug to progress to clinical trials in melanoma is NVP-BEZ235 (BEZ235), a dual PI3K/mTOR inhibitor (216). This compound results in G1 cell cycle arrest, reduction of cyclinD1 and increased p27KIP1, but no marked increase in apoptosis (216). BEZ235 is presently in stage II clinical trial combination study with MEK inhibitor, MEK162. Additionally, GSK2126458 is a potent AKT competitive inhibitor of dual PI3K/mTOR and has entered clinical trial for melanoma combination studies with MEK inhibitor, GSK1120212 (217).

1.6.4 MEK inhibitors

As BRAF and MEK inhibitors achieve the same functional outcome (inhibition of the MAPK pathway) it was thought that they would exhibit similar efficacy clinically. This has not been the case due to the mechanistic nature of the two agents and the therapeutic index that can be gained by selective targeting of mutated kinase, BRAFV600 by vemurafenib. MEK inhibitors demonstrate a non-specific nature of inhibition of the MAPK pathway not just in melanomas but in all tissues (57, 88). While very high doses of vemurafenib are able to be given to patients with some side effects, the required dose of MEK inhibitor to effectively reduce MAPK signalling to that required to induce tumour regression is not viable due to limiting treatment toxicities (57).

Cells with BRAF mutations also are more sensitive to MEK inhibitors than cells harbouring NRAS mutations (88). This is likely to be due the NRAS mutation controlling the activation of both the MAPK and PI3K/AKT signalling pathways, with BRAF mutations controlling the MAPK signalling pathway alone. Given the MEK inhibitor action on normal tissues as well as melanomas, the therapeutic window in targeting MEK is reduced and it may not be possible to adequately inhibit MEK before drug toxicity in normal tissues is reached (88).

1.6.4.1 Trametinib

As the MAPK pathway is activated by either BRAF mutation or NRAS mutation, an additional approach to attenuate signalling is the development of inhibitors specific to the downstream effector molecule, MEK. One such agent, trametinib, is highly specific in targeting MEK1/2 (84) and has been approved in May 2013 for the treatment of unresectable or metastatic melanoma (218). This phase II trial demonstrated a 6 month overall survival of 81% in the trametinib arm vs. 67% in the chemotherapy arm (hazard ratio for death, 0.54; 95% CI, 0.32 to 0.92; P=0.01). Median progression free survival was also 4.8 months in the trametinib group vs. 1.5 months in the chemotherapy group) hazard ratio for disease progression or death in trametinib group, 0.45; 95% CI, 0.33 to 0.63; P<0.0001) (218).

1.6.4.2 Relevance of the use of MEK inhibitors in BRAFWT cell lines

Oncogenic signalling by mutant NRAS via the PI3K pathway may have greater significance than signalling via MAPK in a subset of melanoma tumours (74). In a study comparing the use of MEK inhibitors in BRAFV600E and BRAFWT cells it was found that despite a decrease in phosphorylation in ERK in all lines treated with the MEK inhibitor, expression of cyclin D1 was maintained in the BRAFWT cells, both those harbouring the NRAS mutation and those that had undergone mutations conferring alternative PI3K pathway induction. All lines harbouring the BRAFV600E mutation demonstrated a decrease in ERK phosphorylation and a following decrease in the ERK target, cyclin D1 (74).

1.6.5 Personalized medicine

Cancer treatment is rapidly shifting from generalised chemotherapies to a targeted tumour biology approach as understanding of protein targets e.g. mutant kinases develops and leads to rational design of specific inhibitors (212, 219, 220). Targeting of a particular oncogene that is mutated in a given melanoma has yielded the development of vemurafenib (BRAFV600E target) which is highly specific and effective at inhibiting the activated MAPK pathway (72, 221). Further identification of targetable mutations in melanoma may provide a means to personalise cancer therapy based on an individual patients molecular subtype to maximise the effectiveness of treatment and minimise side effects by abstaining from treatments that are unlikely to provide benefit (72, 221, 222). For the development of personalized medicine, it is important to identify patient groups by way of biomarkers or presence of mutation that indicate these patients as candidates for treatment strategies (219). The mechanism of the MEK inhibitor determines the efficacy in RAS or BRAF oncogene driven melanomas (223). Researchers determined that MEK inhibitors, GDC-0623 and G-573 were more efficacious in RAS driven tumours than BRAF by forming a hydrogen-bond interaction with S212 in MEK that blocks MEK feedback phosphorylation by the wild-type RAF. BRAF mutant tumours required a strong MEK inhibitor, GDC-0973 for effective inhibition of the MAPK pathway (223). These data highlight a melanoma subtype that 48 experiences greater efficacy by a MEK inhibitor due to the underlying molecular background. The melanoma field is shifting towards a molecular classification of individual melanomas to determine individual patient therapy likely to be most efficacious (222, 224, 225) with the view towards long-term disease control (65). Of great interest is the development of biomarkers for patients that are likely to be complete responders to immune based therapy (84, 190).

1.7 Key clinical trials

1.7.1 BRIM-3 study Clinical trial number NCT01006980

The BRAF Inhibitor in Melanoma 3 study (BRIM3, ClinicalTrials.gov number NCT01006980) was an assessment of patient survival when treated with vemurafenib vs. dacarbazine in previously untreated BRAFV600 mutant melanomas (226). The BRIM3 study involved the enrolment of 657 patients with previously untreated, unresectable, metastatic melanoma. Patients were randomly assigned a dose regime of vemurafenib (960mg orally twice daily) or the standard dacarbazine treatment of 1000mg per square metre of body surface area intravenously every 3 weeks. The median progression free survival in the vemurafenib group was 5.3 months and 1.6 months in the dacarbazine group. The objective response rates were 48% vemurafenib compared to 5% dacarbazine and the 6 month overall survival was 84% in the vemurafenib group compared to 64% in the dacarbazine group at the time of interim analysis (226). This resulted in patients being given the opportunity to switch into the vemurafenib group if they desired. The results of this trial represent a turning point in the treatment of metastatic melanoma that resulted in FDA approval of vemurafenib as a frontline treatment for BRAFV600E positive melanomas (88). This also signifies a split in the treatment of melanoma patients as vemurafenib is effective solely for BRAFV600 mutant patients and there remains a critical need for effective agents that target patients harbouring NRAS mutation. In addition to this, despite an encouraging initial response rate in BRAFV600 mutation positive patients, drug resistance invariably develops at approximately 7-9 months further highlighting the need for effective combination 49 agents (63, 221, 227). Vemurafenib has the same efficacy for BRAFV600K mutation positive patients as it does for BRAFV600E patients (84).

The development of novel BRAF inhibitors has had a major impact on melanoma patient treatment. The early onset of response and palliation of disease represent a highly attractive therapy for melanoma patients (57). Off target effects of vemurafenib have been noted in vitro and include the inhibition of ACK1, SRMS and MAP4K5 protein kinases at concentrations of around IC50<50 nM (88). Development of keratoacanthomas and squamous cell carcinomas occurs in 15-30% of patients receiving BRAF inhibitor treatment such as vemurafenib (87, 204). Thought this side effect is significant, development of these lesions has been effectively dealt with by surgical excision and lesions have been observed to grow locally, without metastatic spread (87). An additional study has noted the advent of melanocytic lesions developing or dramatically changing in morphology after BRAFi treatment. 22 lesions were identified in 19 patients on inhibitor therapy (52). Melanocytic lesions were analysed for BRAF and NRAS mutation and assessed for p-ERK, p-AKT status and key signalling molecules for melanoma. Twelve primary melanomas were detected in 11 patients with 27 weeks of BRAFi treatment. 10 nevi developed of which 9 were dysplastic and all lesions were wild-type for BRAF, but interestingly did not harbour NRAS mutation. Common nevi from patients were used as a control which harboured frequent BRAF mutations (52). This study suggests that secondary malignancies (particularly melanocytic lesions and primary melanomas) are potentially increased in patients on BRAFi therapy and frequent skin checks are advised (52, 84, 221). The concern is also raised for tumours below the skin harbouring RAS mutation that may also be accelerated by BRAFi treatment (221). Long term BRAFi therapy may promote the growth of transformed sub-populations of cells with malignant potential and are a growing concern for patients receiving BRAFi therapy (228).

Follow-on from the BRIM-3 trial was a phase II clinical trial of 132 BRAF mutation positive melanoma patients sought to determine the efficacy of vemurafenib treatment on patients that had received prior BRAF-mutation treatment (229). The overall response was 53% and median OS, 16 months (229). This suggests that patients that have already received prior therapy may benefit from vemurafenib treatment.

1.7.2 BREAK-3 study Clinical trial number NCT01227889

Dabrafenib (also known as GSK2118436) is the second agent in the class of BRAF inhibitors selective for mutant BRAFV600E (175, 230). Dabrafenib is active against both the BRAFV600E and BRAFV600K mutant melanomas with a preference towards the BRAFV600E mutation though the phase I trial included both mutations. This was in contrast to the phase I trials for vemurafenib which focussed exclusively on the BRAFV600E mutant melanomas. The dose escalation trial determined a maximum tolerated dose of 150mg twice daily giving and intra-tumoral phosphor-ERK inhibition of >90% (173). BRAF E and K mutations (BREAK) -1 and BREAK-2 clinical trials followed a similar trajectory of the efficacy demonstrated by vemurafenib. Both vemurafenib and dabrafenib BRAF inhibitor agents demonstrate substantial improvement within 1-2 weeks of treatment. Patients that do not harbour the BRAFV600E mutation or have active brain metastases were excluded from the study. However, dabrafenib has been shown to have efficacy against melanoma patients that have brain metastasis (231).

In the BREAK-3 study (Clinical trial number NCT01227889) Stage III and IV melanoma BRAFV600E patients were randomly assigned in a 3:1 ratio either dabrafenib (150mg twice daily, orally n=187) or dacarbazine (1000mg every three weeks, injection n=63). The study primary endpoint was progression free survival. In the dabrafenib group PFS was 5.1 months compared to 2.7 months in the dacarbazine group. Adverse events related to treatment (grade 2 or higher) were noted in 53% (n=100) of the dabrafenib patients and 44% (n=26) of the dacarbazine patients, though grade 3 and 4 events were rare in both groups. In a similar manner to the vemurafenib study, researchers determined that dabrafenib significantly increased PFS compared to standard of care, dacarbazine (205). Dabrafenib has been approved for the treatment of metastatic melanoma patients carrying the BRAFV600 mutation by the FDA in 2013 (180).

Side effects from dabrafenib differ from vemurafenib in that the incidence of SCC or keratoacanthomas is reduced to 6-19% compared to approximately 20% with vemurafenib treatment (84). This could be explained by a greater affinity of dabrafenib

51 for mutant BRAF and reduced cross-reactivity with CRAF thereby inhibiting the MAPK pathway more effectively. In contrast, incidence of fever and chill increased in dabrafenib treated patients (84).

1.7.3 METRIC study Clinical trial number NCT01245062

The phase I dose escalation study for trametinib, 97 patients were enrolled (36 BRAF mutant, 39 BRAF wild-type, 6 BRAF status unknown, and 16 uveal melanoma). From the BRAF mutant patients, 30 were BRAFi naïve. There were 2 complete responses and 10 partial responses in this subgroup giving a response rate of 33% (232). The median PFS for this subgroup was 5.7 months. For the patients that had received prior BRAFi treatment, one partial response was observed. For the patients with BRAF wild-type melanomas, four partial responses were observed with a response rate of 10% (232).

Members of the MEK versus DTIC or Taxol in Metastatic Melanoma (METRIC) study involved clinical trials to test the efficacy of trametinib against DTIC (233) and DTIC and paclitaxel (218). The phase II METRIC trial sought to determine the responsiveness of BRAF mutant melanoma patients to MEK1/2 inhibitor, trametinib against DTIC (233). The phase II trial involved two study arms; in arm 1, patients previously treated with BRAF inhibitors (40 patients), in arm 2, patients had been previously treated with chemotherapy and/or immunotherapy but not BRAFi (57 patients). The dosage for both groups was 2mg trametinib, twice daily which was well tolerated (233). In the first arm, no objective responses were observed and 28% of patients achieved stable disease. The median PFS was 1.8 months. In the second arm, one complete response was observed (2%), 13 partial responses (23%) and 29 patients (51%) with stable disease. The median PFS was 4.0 months (233). This study indicates that trametinib may be a useful therapeutic option for BRAF mutant patients that are naïve to BRAFi therapy, but does not demonstrate clinical efficacy for patients previously treated with BRAFi (233).

The phase III METRIC trial evaluated the efficacy of trametinib against DTIC or paclitaxel in BRAFV600E/K mutant positive patients in an open label study (218). 322 patients were randomly assigned trametinib (2mg daily), or chemotherapy (DTIC 1000mg/m2 body surface area or paclitaxel 175mg/m2 every three weeks) in a 2:1 ratio.

PFS was the primary endpoint and was 4.8 months in the trametinib group vs. 1.5 months in the chemotherapy group (218). Patients who experienced disease progression in the chemotherapy group were permitted to cross over to the trametinib group. OS was the secondary endpoint and was 81% at 6 months for the trametinib group and 67% for the chemotherapy group. This study confirms that patients with BRAFV600E/K melanomas have improved PFS and OS compared to chemotherapy (218). These trials did not include NRAS mutant patients and therapeutic efficacy of trametinib remains unclear for this molecular subtype.

1.7.4 FDA approval of treatments for metastatic melanoma

Given the recent success and ability to target mutant BRAF in harbouring melanomas using selective BRAF inhibitors, attention has now turned to BRAF wild-type tumours in an effort to identify drugable targets that may demonstrate similar efficacy to their BRAF counterparts (65). In particular, NRAS mutant melanoma largely remains without effective therapy (211) (Figure 9). Key targeted therapies in their respective categories are represented in Figure 10.

Figure 1.9. Timeline of FDA approved treatment for advanced melanoma. Timeline of advancement of melanoma treatment involves the approval of dacarbazine, interleukin-2, vemurafenib, ipilumamib and trametinib. Targeted treatment for NRAS melanomas is currently lacking

Figure 1.10. Summary of key melanoma therapies Treatment of advanced melanoma has an increasing arsenal of targeted therapies

1.7.5 Combination therapies and recent clinical trials

Combination therapies are presently being investigated for the treatment of metastatic melanoma (114, 193, 217, 234, 235). Given the heterogeneity of the melanoma tumour and complexity of signalling deregulation of individual cells a combination approach is required to effectively target metastatic melanoma (83, 215, 235, 236). Enhancers of clinically used BRAF inhibitors that target the MAPK pathway or another de-regulated pathway in addition to the clinically efficacious immune therapies may be an appropriate avenue for treatment of patients with metastatic melanoma (234). Inhibition of mutant BRAF is associated with enhanced melanoma antigen expression providing a rationale for BRAFi and immune combination therapy (237).

Clinical studies using combinations of BRAF and MEK inhibitors demonstrate the advantage of this pairing by reduction in single agent toxicities (238). This is possible due to the reduction of paradoxical activation downstream consequence by the presence of MEKi and on the flip side the paradoxical activation that does occur, attenuates the effect of MEK inhibition (133). This data is encouraging as the combination of MEK and BRAF inhibitors may also negate some of the effects of BRAFi alone such as the development of RAS driven secondary cutaneous SCC’s developing (133). Combinations that target the MAPK and AKT/PI3K pathway are an attractive option due to the de-regulation of both of these pathways in melanoma progression and the lack of agents to inhibit mutant NRAS directly. This may be problematic however, due to the importance of these pathways in multiple cellular functions (133).

A recent study by Posch and colleagues has demonstrated that combined targeting of MEK and PI3K/mTOR pathways in is necessary for effective tumour regression in vitro and in vivo in NRAS melanomas (239). The combination of the two inhibitors results in induction of apoptosis. In previous studies binding of RAS to PI3Kp100α is required for

RAS driven tumorigenesis and mTORC2 is required for melanoma transformation and favourable melanoma growth (239-241). Targeting if the PI3K pathway is however problematic and optimization of patient selection, better understanding of pathway crosstalk and feedback looks is required to avoid development of drug resistance (114).

Vemurafenib has demonstrated synergy with nutlin-3, depleting survivin and suppressing melanoma cell viability and tumour growth in vitro (37 melanoma lines) 55 and tumour growth in vivo (242). Gene depletion of survivin recapitulated the cytotoxic effect induced by nutlin-3/vemurafenib combination and both agents were found to independently attenuate survivin levels. Nutlin-3 may serve as an effective anti- melanoma agent as the majority of melanomas harbour wild-type p53 protein and may serve as an effective therapy alongside BRAFi (242).

Combination of immune therapy with classical chemotherapies (biochemotherapy) has not to date demonstrated clinical success for melanoma patients (10, 165). Biochemotherapy using IL-2 requires a dose reduction (due to combination toxicity) which may compromise the induction of complete response experienced by a small number of patients (165). Potential combination of radiation therapy combined with immunotherapy may induce an abscopal effect in some patients leading to tumour regression (133). Given that melanoma is a heterogeneous cancer type (45), it is clear that identification of subsets with particular molecular signatures that are responsive to anti-melanoma agents is important in gaining greater therapeutic response (45, 193, 215, 217).

1.7.5.1 Immunotherapy and BRAF inhibitor combination

Possibly the most promising combination is that of BRAFV600E inhibitors with CTLA-4 or PD-1 blocking antibodies. This combination offers the advantage of two mechanisms, first, vemurafenib acts rapidly to reduce metastatic burden in a two week period (62) combined with the slow, durable response associated with responders to ipilimumab (188, 243). In clinical trial however, an unexpected degree of hepatotoxicity was observed in at phase I of combination vemurafenib and ipilimumab (234). This was unexpected as the only significant clinical overlap of side effects between the two agents was skin and liver toxicities, which generally do not become limiting in patient treatment. All participants of the trial harboured the BRAFV600 mutation and had not received prior therapy of BRAF or MEK inhibitors, neither CTLA-4 nor PD-1 blocking antibodies (234). Both agents were given at the maximum doses, vemurafenib at 960mg twice daily run in for 1 month, followed by four treatments of ipilimumab (3mg/kg every 3 weeks). The second round in the study lowered the dose of vemurafenib to 720mg twice daily with all other conditions kept the same. In both studies, grade 3 56 levels of aminotransferase (signifying hepatotoxicity) were noted in 4/6 and 2/6 patients respectively (234). This study highlights that although agents are safely used at high doses as single agents, or in combination with MEK inhibitors, agents with differing mechanism of action may exhibit unforseen adverse events in combination (234).

There is also the question of the order of melanoma treatment required to maximally benefit the patient. Is it more favourable to treat a patient with BRAFi to rapidly reduce tumour burden and follow by the more durable response obtained by immunotherapies? Inhibiting the MAPK pathway with BRAFi has been shown to increase the expression of melanoma derived antigen thereby increasing the melanoma cell recognition by T cells (244). Is prognosis more favourable if both BRAFi and immune therapies are administered together? What is the role of adjuvant treatments in this setting? These questions are presently being addressed by clinical trials (227).

1.8 Raf isoform switching

Curiously, BRAF inhibitor treatment of BRAFWT melanoma cells that harbour the NRAS mutation increases MAPK pathway activation (64, 66, 72). As NRAS mutant cells activate ERK via enhanced CRAF signalling it has been noted that treatment of these cells with the BRAFi, vemurafenib increases ERK phosphorylation. Further studies indicate that addition of the inhibitor induces BRAF binding to CRAF and is thought to act as a molecular scaffold thereby enhancing CRAF signalling (76, 200, 245). The isotype switch is confirmed by treatment of RAS mutant (BRAFWT) melanoma cells with BRAFi and observation of phosphor-MEK and phosphor-ERK activation. Subsequent knockdown of CRAF with concomitant BRAFi treatment ameliorated this effect (246). This RAF isoform switch is thought to explain the incidence of keratoacanthoma-type SCC in patients treated with vemurafenib as the prior presence of the NRAS mutation in squamous cells may predispose them to greater activation of CRAF by the binding of BRAF on inhibitor treatment and pathway activation (72, 76). In clinical trials of BRAFV600E patients receiving vemurafenib in dose escalation studies, and phase II and III trials received 720 or 960mg twice daily (204). On vemurafenib treatment, 21 of 35 patients developed squamous cell tumours (60%). The result of vemurafenib treatment was not only a reduction in BRAFV600E tumour burden but an initiation of squamous cell tumour growth in squamous cells that 57 harboured RAS (particularly HRAS but also NRAS) mutations. Side effects at this dosage in addition to squamous cell carcinoma include fatigue, rash and arthralgia (173). The most prevalent mutation was the HRASQ61L and this was also seen in HRASQ61L mutant cell lines that displayed increased cell proliferation by MAPK pathway activation and increased ERK transcription on treatment with vemurafenib (204).

Binding of BRAF to CRAF in the presence of BRAF specific inhibitors, sorafenib, 885- A, PLX4720 (200) and vemurafenib (245) have been observed in the presence of an NRAS mutation, however, this binding is not observed in the mutations absence suggesting that inhibition of BRAF alone does not cause CRAF binding. Researchers have proposed that in the presence of activated NRAS, BRAF may be recruited to the plasma membrane to bind CRAF where the protein predominantly resides (200). Additionally, NRAS mutant cells experience resistance to apoptosis in the presence of BRAFi and hyperactivation of the MAPK pathway (62).

RAF isoform switching can be overcome by co-targeting the MAPK pathway and IGF- 1R/PI3K in resistant cells that utilise PI3K signalling. In this study, researchers developed BRAFi resistant cells by chronic treatment of BRAF mutant melanoma cell with selective BRAFi, SB-590885. They further determined that these cells had developed cross-resistance with other BRAF inhibitors. They observed an increase in IGF-1R/PI3K in these melanomas and sought to block MAPK signalling using a combination of MEKi and IGF-1R/PI3K inhibitors with ensuing cell death (164). Cross- resistance to MEKi has also been demonstrated in BRAFi resistant cells that have undergone resistance by NRAS mutation and gain of COT expression (247).

1.8.1 Mechanisms of RAF isoform switching

Melanocytes use BRAF to activate the MAPK pathway whereas NRAS melanomas utilise CRAF to elicit MAPK signalling (64). In melanomas that have a MAPK pathway activation mediated by RAS mutation and induction of BRAF phosphorylation on Serine 151 near the RAS binding domain is observed, the binding of BRAF to RAS is prevented (64, 66, 248). In addition, BRAF T401, S750, and T753 phosphorylation by

58 active ERK interferes with the ability of BRAF to dimerize CRAF and provides a negative feedback regulation of BRAF protein. To by-pass the negative feedback of BRAF, NRAS mutant melanoma utilise CRAF to activate the MEK/ERK signalling cascade. This is also accompanied by a deactivation of cAMP by an increase in the negative regulator by degradation of cAMP, phosphodiesterase 4 (64, 248) (PDE4) (Figure 11).

Figure 1.11. RAF isoform switching with BRAF inhibitor treatment

Treatment of BRAF+/+ and BRAF-/- cell lines with BRAFi resulted in the induction of CRAF specific activity and downstream phosphor-MEK and phosphor-ERK activation indicating the RAF isoform switch can occur in the presence or absence of BRAF (246). To test the necessity for CRAF in phosphor-MEK activation, a potent inhibitor against CRAF, AZ-628, was used. Treated cells showed no phosphor-MEK activation or hyper- proliferation with BRAFi treatment (246).

ATP competitive RAF inhibitors do not inhibit ERK signalling in cells that co-express mutant NRAS and BRAFV600E (61). Vemurafenib treatment of wild-type and Braf-/- murine embryonal fibroblasts (MEF) cells demonstrated a significant increase in phosphorylated MEK and ERK. This response was decreased in Craf-/- MEF’s indicating its importance in MEK/ERK activation with vemurafenib treatment (61, 246). Treatment of non-BRAFV600E melanoma lines with PLX4720 resulted in a translocation of BRAF and CRAF from the cytoplasm to the membrane fraction and a concomitant increase in CRAF phosphorylation. This process was dependant on RAS- GTP (246). The significance of RAS-GTP is confirmed after transfection of RAS mutant A375 cells with BRAFV600E, but not wild-type RAS, gives rise to CRAF-BRAF heterodimerization and CRAF activation after BRAFi treatment (246). A general increase in BRAFWT melanoma cell proliferation is noted by these authors with BRAFi treatment. This contrasts with other literature that indicates a reduction in proliferation of both BRAFV600E and BRAFWT melanomas (Mel-RM) with increasing doses of PLX4720 though the reduction in proliferation in BRAFWT cells is appreciably less than the mutant, Mel-RMu (210). In addition, treatment of ex-vivo tumour tissue with vemurafenib showed a decrease in tumour kinase activity in BRAFV600E melanomas and also a subset of BRAFWT melanomas indicating BRAF mutational status is not the only determining factor for vemurafenib inhibition (249). The effect of BRAF inhibitor on the proliferative status of BRAFWT/NRASMT cells may be somewhat dose dependant and/or dependent on individual cell line response.

Interestingly, the BRAF inhibitor used is important in determining the dimerization and mechanism of transactivation of RAF signalling. Treatment of mutant NRAS, wild-type BRAF cells with BRAFi, PLX4720 resulted in the homodimerization of CRAF-CRAF (61, 246) and the formation of BRAF-CRAF heterodimerization was prevented (246) (Figure 12). It is proposed that the use of pan-RAF inhibitors that potentially inhibit

CRAF at clinically achievable levels should be considered for the treatment of NRAS mutant melanomas to avoid stimulation of the MAPK pathway (72).

Figure 1.12. Vemurafenib treatment of BRAF wild-type/NRAS mutant melanoma cell

1.9 Drug resistance in melanoma

Melanoma is a defined as a molecularly heterogeneous cancer that either harbours drug resistant sub-clones within the tumour, and present before treatment, or develops resistant sub-populations as a result of treatment intervention (250). Resistant tumours do not become independent of their initial driver mutation deregulated pathway/s; rather, they develop mechanisms that re-activate the pathway despite the presence of the inhibiting agent (251).

1.9.1 Mechanism of BRAF inhibitor resistance

The current standard of vemurafenib treatment is to continue until there is clear evidence of disease progression (57). Though clinically there is no one predominant mechanism of resistance to vemurafenib, a number of contributing mechanisms have been identified. These fall into four main categories; intrinsic and acquired resistance (57) and either ERK or non-ERK dependant mechanisms (252). While the mechanisms of intrinsic BRAFi resistance are not well defined, a number of molecular mechanisms of tumour escape have been identified in acquired resistance, predominantly defined in vemurafenib treated patients (57). As a predictive biomarker, one study identified high expression of nuclear p27 preceding BRAFi treatment was the strongest predictor of poor OS and predicted unfavourable response to BRAFi (196).

Despite abrogation of major signalling pathways with clinical agents, most notably the BRAF inhibitor, vemurafenib, melanomas typically circumvent inhibitors and resume proliferation (236, 253). Heterogeneous mechanisms of resistance to BRAF inhibitors fall into two categories. The first is a redundancy of the MAPK pathway and an alternative dependence on the AKT signalling pathway. The second mechanism is a reactivation of MEK-ERK signalling by by-pass of BRAF and direct phosphorylation of the downstream proteins (236, 251, 253).

Treatment with the most effective BRAFV600 anti-melanoma agent, vemurafenib carries some significant side effects. The treatment period with vemurafenib and observed tumour regression is on average 6-7 months (175, 250). After this period, resistance to

64 the inhibitor develops due to by-pass of BRAF and re-activation of the MAPK signalling cascade (253). This resistance is de novo or acquired and is not attributed to additional mutation in the BRAF gene (68, 75, 173). Resistance mechanisms to vemurafenib also include alternative splicing of the BRAF mutant protein such that it retains functional activity, but is no longer effectively targeted by vemurafenib (253). These splice variants lack the middle portion of the BRAF protein and are no longer effectively bound by the inhibitor. Initial studies have showed the presence of these splice variants in six of nineteen tumours indicating it may be a common mechanism of resistance. In general, most resistance mechanisms present in BRAFi treated melanomas rest on the re-activation of ERK phosphorylation (236). These can be via intrinsic or extrinsic mechanisms. Figure 13 provides an overview of the elucidated mechanisms currently known in vemurafenib resistance.

Figure 1.13. Mechanisms of Vemurafenib resistance by-passing BRAF. 1. Secondary mutations in Ras. 2. Up-regulation of COT capable of phosphorylating MEK. 3. Increased signalling through receptor tyrosine kinases. 4. Raf isoform switching from BRAF to CRAF. 5. Secondary mutation of MEK.

1.9.1.1 Intrinsic resistance

Intrinsic resistance involves a mechanism by which the melanoma cell mutational background ensures it is innately resistant to anti-melanoma treatment (170, 254). Examples of this are BRAF mutant cells with a concomitant amplification of cyclin D1 (CCND1) which are more resistant to BRAFi treatment (252). Given that cell cycle disruption is observed in up to 75% of melanoma patients and CCND1 amplifications are seen in up to 20% of patients, this is a clinically relevant finding (252). In addition, PTEN mutational status correlated with poor BRAFi treatment response. Low tissue expression of PTEN (a negative regulator of the PI3K pathway) is associated with shorter progression free survival in patients treated with dabrafenib as a single agent (255). Another intrinsic mechanism of resistance is the maintenance of slow-cycling JARID1Bhigh melanoma subpopulations, maintaining the tumour burden after a reduction in tumour bulk cells that are sensitive to the BRAFi (254). This enrichment of tumour maintaining cells have increased oxidative phosphorylation and researchers determined that blocking mitochondrial respiration halted the development of the JARID1Bhigh subpopulation and sensitized the cells to BRAFi therapy (254). This research lends support to a therapeutic approach to decrease tumour bulk with BRAFi’s and target remaining resistant sub-populations with agents tailored to blocking oxidative phosphorylation as a specific approach to address tumour heterogeneity (254).

1.9.1.2 Acquired resistance

Acquired resistance has been observed in patients treated with vemurafenib and typically emerges 7-9 months after treatment commences (87). Interestingly, no ‘gatekeeper’ mutations have been observed negating the possibility of a second selective inhibitor against a new mutation in BRAF (252, 253). Splice variants of BRAF as opposed to observation of gatekeeper mutation have been observed as a resistance mechanism in BRAFi treated melanomas (236). NRAS mutation is one of the most common mechanisms of circumventing BRAF inhibitors alongside an increase in the amount of BRAF protein expression, that is able to overwhelm levels of inhibitor present and maintain ERK phosphorylation (68, 173, 236, 256). Additional mechanisms of acquired resistance are varied and involve amplification of RTK’s (68, 257, 258), 67 amplification of BRAF hereby overcoming inhibition by drug, or upregulation of COT which by-passes BRAF and directly phosphorylates MEK (68, 87, 173). These acquired resistance mechanisms are a combination of both ERK dependant and independent ways of overcoming vemurafenib treatment and allowing uncontrolled tumour growth to continue (63, 259).

Drug resistance is a complex matter and is influenced by the tumour stroma in addition to internal cellular signalling. A study published by Straussman et al (260) performed studies on a comprehensive panel of melanoma cell lines and human patient samples to confirm that stromal cell secretion of hepatocyte growth factor (HGF) results in the activation of the HGF receptor, MET, reactivation of the MAPK and PI3K/AKT pathways and resistance to BRAF inhibition by PLX4720 (260). These findings were validated in the confirmation that melanoma patients resistant to BRAFi therapy exhibited a high stromal expression of HGF. The HGF that is secreted by surrounding fibroblasts is also known to block gefitinib, and EGF receptor inhibitor used in the treatment of non-small cell lung cancer (260).

1.9.1.3 ERK dependant mechanisms of resistance

ERK-dependant mechanisms of acquired resistance involve a re-activation of the MAPK pathway in the presence of the BRAFi. Mutations in the NRAS gene following resistance to vemurafenib result in the reactivation of the ERK pathway despite BRAF inhibition (88). A case report of a progressing BRAFV600E positive melanoma patient treated with vemurafenib indicates the presence of two distinct sub-clones both harbouring BRAFV600E and one gaining the G13R NRAS mutation (250). This highlights the need for individualized therapies to target newly emerging resistance mechanisms.

Cancer Osaka thyroid kinase (COT, also known as MAP3K8) drives resistance to RAF inhibition through MAP kinase pathway reactivation and has been observed as a resistance mechanism in laboratory experiments as well as clinical samples (88, 261). COT is capable of directly phosphorylating MEK and driving ERK activation in a manner independent of BRAF or CRAF (68). Mutant BRAF activation down-regulates

COT expression and aberrant expression of COT causes resistance to RAF inhibitors (68). COT levels increase in cell lines treated with BRAFi highlighting the potential for COT in a feedback regulation relationship with MEK (63).

Secondary mutations in BRAF to date do not feature as a resistance mechanism (251), but point mutation in MEK is noted to be a factor in over-coming BRAF inhibition (76, 183, 251, 261). These mutations were noted to be located in or near the allosteric drug- binding pocket and hypothesized to interfere with drug binding (63). Additional point mutations identified in MEK are thought to affect intrinsic MEK kinase activity or conformation of the protein (63). Clinically, a MEK1P124L mutation in was identified in a random mutagenesis screen resulting in cellular resistance to the MEK inhibitor, AZD6244 (75). In addition, a vemurafenib resistant patient (after an initial strong response to treatment) was identified that had gained a mutation in MEK1C121S which increased kinase activity and showed resistance to RAF and MEK inhibition in vitro (262). It was also identified in a patient that had acquired resistance to AZD6244. Combination of PLX4720 and AZD6244 was sufficient to overcome resistance due to MEK1P124L (75). Mutations in MEK2Q60P with BRAF amplification also confer resistance to both BRAF and MEK inhibitors (263). Copy number gain of BRAF itself has been implicated as a resistance mechanism in vitro, despite no secondary mutation of BRAF itself being observed (75, 251, 259). In one study, BRAF copy number gain has been observed in 4/20 (20%) of resistant patient samples analysed by whole-exome sequencing (251). However, comprehensive understanding of the clinically relevant mechanisms is restrained due to the limiting numbers of MEKi resistant tumours available for analysis (63). BRAFi and MEKi often results in hyperactivation of phospho-MEK compared to parental cells and dose response requiring a 50% reduction of phosphor-MEK in resistant cells before any effect is seen in downstream phosphor- ERK (63). In comparison, a 50% decrease in phosphor-MEK resulted in a 50% decrease in phosphor-ERK in parental cells, suggesting resistant cells harboured levels of MEK activation vastly in excess to what is required for ERK activation. This affect was attenuated with BRAF inhibitors in resistant cells, reducing MEK levels to parental and restoring the dose response relationship between MEK and ERK (63).

1.9.1.4 ERK-independent mechanisms of resistance

Two proteins upstream of PI3K/AKT pathway are platelet derived growth factor receptor beta (PDGFRβ) and insulin growth factor 1-receptor (IGF-1R). These activate the AKT pathway and subsequent downstream mTOR signalling. Over-expression or increased signalling of these RTK’s can confer BRAF inhibitor resistance (75, 76, 183, 257). Furthermore, AKT activation is biochemically associated with increased PDGFRβ expression and additional receptor tyrosine kinase IGF-1 is also associated with BRAF inhibitor resistance (88, 183). In one study of vemurafenib resistant patient samples, it was observed that 4/11 tumours displayed up-regulation of PDGFRβ. To confirm this observation, induction of PDGFRβ in BRAFV600E melanomas has been shown to demonstrate resistance and the proliferation of resistant cells was abrogated by knockdown of the receptor (68). Combined inhibition of BRAF, PI3K and mTORC1/2 resulted in a synergistic growth inhibitory response in melanoma cells (82). In addition, the inhibition of MEK1/2, PI3K and mTORC1/2 triggered apoptosis efficiently (82). These data suggest combination inhibitors targeting major pathways may effectively overcome PDGFRβ mediated vemurafenib resistance.

Further, elevated AKT signalling has been implicated in cell lines with acquired BRAFi resistance (75). Cross-feedback activation of the MAPK and AKT pathways has been implicated as resistance mechanism of inhibitors targeting either RAF/MEK or PI3K/AKT/mTOR pathways. This is due in part to BRAFV600E being a negative regulator of the AKT pathway (264). BRAFV600E interacts with rictor complex, mTORC2 and regulates AKT. Gene silencing by siRNA’s define this process as a rictor complex dependant, MEK/ERK and BRAF kinase independent process (264).

An additional mechanism melanoma cells overcome RAF inhibitors is by FOXD3- mediated up-regulation of ERBB3 in melanoma cell lines and xenograft models (265). Co-targeting resistant xenograft tumours with PLX4720 and EGFR targeting lapatinib results in reduced tumour burden and increased tumour re-growth latency vs. vemurafenib alone (265).

STAT3 has recently been identified as contributing to chemoresistance in BRAFV600E melanoma cells (266). The mechanism of action involves oncogenic BRAF mediated stabilizing phosphorylation of Mcl-1 which is anti-apoptotic in melanoma. Activation of 70

STAT3 by phosphorylation at serine-727 and tyrosine-705 promoted by BRAFV600E signalling and the Mcl-1 promoter is dependent on a STAT consensus site for BRAF driven activation. Abrogation of STAT3 activity disrupted BRAFV600E transcriptional induction of Mcl-1 and reduced melanoma cell viability (266).

1.9.2 Strategies to overcome vemurafenib resistance

The chosen treatment strategy to target BRAFi resistant tumours will depend on whether the resistance mechanism is of an ERK-dependant or independent nature. This is due to the retained sensitivity of ERK dependant resistance mechanisms to other BRAF or MEK inhibitors (63). For the treatment of ERK-independent melanomas a combination of PI3K/MEK or PI3K/BRAF inhibitors target the dual nodes of the MAPK and AKT pathway and have been proven to be efficacious (63).

As ERK reactivation is required for the development of resistance in vemurafenib treated tumours, agents that inhibit MEK were put forward as a rational approach to further control MAPK pathway activation. BRAF inhibitors in combination with inhibitors of the PI3K/AKT pathway are also being trialled clinically to target two major signalling nodes simultaneously with the hypothesis of inducing greater tumour regression and targeting resistant cells.

A study performed by Liu et al discovered that activation of STAT3 or over-expression of PAX3 induced vemurafenib resistance in melanoma cells (267). This was based on the observation that the STAT3-PAX3 signalling pathway is up-regulated due to FGF2 secretion or increased BRAF kinase activity with the BRAFV600E mutation. To confirm this, STAT3 or PAX3 gene silencing in vemurafenib cells with acquired resistance demonstrated reduced growth and use of the STAT3 inhibitor, WP1066 resulted in growth inhibition in sensitive and resistant BRAFV600E melanoma cells. This work suggests that STAT3-PAX3 may be a viable therapeutic target in vemurafenib resistant melanoma cells (267).

1.9.2.1 Intermittent dosing and vemurafenib resistance

Emerging data suggests that a discontinuous dosing of vemurafenib is advantageous in avoiding lethal drug resistant occurrence (253, 268, 269). In a mouse model of parental human HMEX1906 tumour xenografts, it was found that a dosing regimen of 15mg/kg vemurafenib in a discontinuous manner of 4 weeks on 2 weeks off (twice daily) vs. continuous dosing, drug resistance did not develop (n=9) after 200 days. In contrast, the continuously dosed mice developed lethal drug resistant tumours evidenced by significant uncontrolled growth in tumour volume (n=7) (253). This observation was corroborated clinically a short time later indicating 14/19 patients with vemurafenib resistant tumours showed tumour regression with cessation of treatment (270). Furthermore, in studies using human patient tumour xenografts to mice that were either treated intermittently (4 weeks on 2 weeks off) or continuously, it was shown that the mice receiving intermittent dosing were protected from tumoural drug resistance as opposed to the continuously dosed mice which demonstrated resistant tumours. It is thought that this is due to the maintenance of the selective pressure favouring lethal drug resistance (270).

Patients can be successfully re-challenged with vemurafenib after an initial response (268, 269) and may experience a reduction in treatment related side effects (268). In a case study, one patient achieved a near-complete response to vemurafenib before undergoing progression. The patient then received chemo and immunotherapy before a successful re-challenge of vemurafenib. However, the patient exhibited multiple mechanisms of vemurafenib resistance including two NRAS activating somatic mutations (Q61K and Q61R) and a BRAF alternative splicing (271). These alterations were not present in pre-treatment tumour samples. Additionally a PIK3 mutation was detected in a tumour sub-population illustrating the heterogeneity of resistance mechanisms, though this mutation did not contribute significantly to vemurafenib resistance (271).

1.9.3 MEK inhibitor resistance

Resistance to MEK inhibitors has been identified in the form of increased BRAF or KRAS as oncogenic drivers of MAPK signalling (272). Co-targeting MAPK signalling with a BRAF inhibitor (for BRAFV600 positive melanomas) and MEK inhibitor may still provide a valuable means of delaying or overcoming drug resistance. This provides a mechanistic rationale for the investigation of BRAFi + MEKi therapy (272).

As ERBB3 is up-regulated by FOXD3 in PLX4720 resistant melanomas and identified as a resistance mechanism, ERBB3 may also serve as a resistance mechanism in MEK inhibitor resistant melanoma and be a viable therapeutic target though this is yet to be established (265).

It has been demonstrated that there is a high degree of cross-resistance between the BRAFi, vemurafenib and MEKi, AZD6244. The exclusion to this is when the vemurafenib resistance is a result of a gain of NRAS mutation (105). Researchers further demonstrate that acquired resistance to MEKi, AZD6244 was accompanied with persistence or increase of AKT pathway activity. Gene silencing mediated by siRNA and combination of AKT inhibitor or rapamycin reverse the resistance of MEKi (105).

The relationship of cross-resistance between MEK and BRAF appears reciprocal. Mutations, MEK1P124L and MEK1Q56P when treated with MEKi, AZD6244 confer cross- resistance to BRAFi, PLX4720, in BRAF mutant melanoma. Dual treatment of AZD6244 and PLX4720 prevented the development of resistant clones (273). These findings offer valuable information into the strategy of use of both BRAFi and MEKi and the potential benefit of co-targeting the MAPK and PI3K/AKT pathways. MEKQ56P and MEKE203K are two of the most prevalent secondary mutations after NRASQ61 in vemurafenib resistant patient samples (256).

1.9.4 Molecular biomarkers and prognostic markers in melanoma

Biomarkers are factors that correlate with tumour behaviour and patient prognosis (274). A number of molecular biomarkers have been proposed as predictors of poor or 73 favourable survival in melanoma. Hsp90 has been identified as having increased expression in metastatic melanoma compared to both primary melanoma and nevi. It also correlates with tumour thickness and Clark level however, no correlation has been found with high level of Hsp90 and survival between primary and melanoma metastasis (274).

LKB1 (serine/threonine kinase 11) has been identified as a tumour suppressor protein in sporadic lung and cervical cancers (275). Somatic loss of LKB1 has been identified as a causal event in melanomagenesis. LKB1 is a serine-threonine kinase responsible for phosphorylating AMP activated protein kinase (AMPK) and phosphorylation correlates in vivo with LKB1 loss (276). Conditional knock-out mouse modelling has shown that LKB1 results in a highly-metastatic melanoma phenotype. Interestingly, LKB1 loss can confer sensitivity to some classes of anti-cancer drugs (276).

Markers matrix metalloproteinase-2 (MMP2), Ki-67, proliferating cell nuclear antigen (PCNA) were associated with unfavourable melanoma prognosis (277). Microtubule associated protein-2 (MAP-2) is significantly reduced in metastatic melanomas compared to primary cutaneous melanoma. In a 5-year follow-up of 37 melanoma patients, high MAP2 expression was determined to be an independent favourable prognostic marker (278). High nuclear expression of β-catenin was associated with poor prognosis in advanced melanoma due to the regulation of WNT signalling and interaction with cell-cell adhesion protein, E-cadherin (278). Melanoma cell adhesion molecule (MCAM) is more highly expressed in melanomas than primary tumours. With a 7-10 year follow up of 76 patients, MCAM expression (low, moderate, high) was an independent prognostic marker with low expression being favourable (278). Tumour suppressor, Tip60, has been shown to be an independent prognostic marker for primary and metastatic melanoma (279). In a cohort of 448 melanoma cases and 105 control nevi, reduced Tip60 was associated with a poor 5-year overall survival. Functional studies further identified Tip60 expression to reduce cell migration and increase chemosensitivity, suggesting Tip60 as a potential therapeutic target in melanoma (279).

One study yielded a five-marker approach to determine patient prognosis from 38 candidates with stage II melanoma. This study sought to determine which patients would undergo melanoma metastasis from those with non-progressive disease (280). These researchers define an automated quantitative analysis (AQUA) using 74 immunofluorescence and immunohistochemistry methods. Key proteins were ATF2, p16INK4A, p21WAF1, β-catenin and fibronectin assessing nuclear and cytoplasmic staining alongside protein concentrations. Researcher determined that four of five conditions being met according to specified parameters determined a low risk group; three or fewer conditions determined a high risk group (280). This multi-marker approach was effective in determining high or low risk in the examined cohort (280).

1.10 Novel tumour suppressor proteins in melanoma

Identification of novel tumour suppressor proteins in melanoma can occur by numerous means. MEN1 is a gene that has been identified as required for oncogene induced senescence in melanocytes and has subsequently been investigated as a tumour suppressor in melanoma (281). It was further found that MEN1 expression is significantly reduced in melanoma cell lines and patient melanomas. MEN1 levels were found to increase after DNA damage and stimulated gene transcription of BRCA1, RAD51, and RAD51AP1 that are all involved in homologous recombination directed DNA repair (281).

Breast cancer suppressor candidate-1 (BCSC-1) has recently been identified as a novel tumour suppressor protein in melanoma by down-regulating MITF (282). BCSC-1 was found to be significantly decreased in metastatic melanoma and was found to be ectopically expressed in vivo (282). BCSC-1 binds to sox10, which then down-regulates MITF resulting in the melanoma cells switching from a proliferative state to a migratory phenotype (282). This switch to an invasive phenotype will need to be carefully considered in selecting BCSC-1 as a molecular target for melanoma therapy.

Tetratricopeptide repeat protein 4 (TTC4) is proposed as a novel candidate tumour suppressor protein in melanoma. Investigation of aberrations in chromosome 1p31 has revealed seven distinct point mutations in the TTC4 gene, of which six result in an amino acid change (283). No mutations were detected in control normal or dysplastic nevi. Six of 25 metastases harboured the point mutations. In cell line studies, two cell lines demonstrated whole exon loss (283). These indicate that TTC4 may contribute to melanoma development and is a candidate as a novel tumour suppressor protein. 75

Investigation of the chromosomal rearrangement between 6q and 17p present in the melanoma cell line, UACC-930 has yielded the novel tumour suppressor gene at 6q21, prenyl diphosphate synthase subunit 2 (PDSS2). PDSS2 is down-regulated in 59/87 (67.8%) of primary melanomas and only 7/66 (10.6%) of benign nevi (284). To determine the tumour suppressive potential of PDSS2, it was stably transfected into UACC-903, which is highly tumourigenic. PDSS2 transfection resulted in reduced proliferation, colony forming ability, and tumourigenecity was completely ablated in a mouse xenograft model (284).

1.11 TRIM family proteins

The TRIM protein super-family of tripartite motif-containing (TRIM) proteins is highly conserved between species (285, 286) with approximately 100 known members, and a number of new TRIM family proteins are frequently uncovered (287, 288). It is one of the largest protein families presently identified ((289)). It comprises family members that typically contain a RING-B-box-Coiled Coil motif with variable C-terminal domains (290-292). There are variations to this structural format, however, there is a sub-set of TRIM proteins, which lack of the RING domain while retaining the B-boxes. For example, TRIM16 protein has a two B-boxes and a Coiled Coil domain with PRY- SPRY (B30.2) in the C-terminal, but does not have the RING domain (292). The function of the B30.2 domain is not known, though it is thought to contribute to its antiviral TRIM activities (291). Approximately 45% of the SPRY containing proteins are E3 ubiquitin ligases and mutations in this region are associated with congenital disorders (293). A number of TRIM proteins including pyrin and midline1 (MID1) were discovered with harbouring the highly conserved B30.2 sequence (293). Emerging evidence shows that the TRIM family may also be closely related to the TNF receptor associated family (TRAF) due to structural similarity (285).

The TRIM family of proteins is functionally diverse and involved in cellular processes including cell cycle regulation, proliferation, differentiation and ubiquitination (290), apoptosis and tumour suppress functions and oncogenesis (294). TRIM proteins have been shown to oligomerize and interact with other proteins through their B-box, Coiled-

76 coil and SPRY domains, though additional functions of these domains remain to be determined (295). TRIM family members frequently homo and heterodimerize, increasing their diversity of function (289, 296). They may also form trimers to carry out particular functional activity (297). Additionally, TRIM family members have been identified as SUMO E3 ligases with substrates including p53 and MDM2 (298). The TRIM family of proteins can be classified into subfamilies I to XI dependant on their domain structures.

1.11.1 TRIM family proteins and keratinocyte biology

TRIM family proteins have been identified as playing a role in the biology of keratinocytes. One of the most extensively studied is TRIM29. TRIM29 is highly expressed keratinocytes and in reconstructed epidermis (299, 300). TRIM29 was determined to be differentially expressed alongside other known factors involved in keratinocyte differentiation and survival upon sub-lethal UV exposure (299). Knockdown of TRIM29 mediated by short hairpin RNA (shRNA) resulted in reduced keratinocyte viability after UV exposure. This UV induced increase of TRIM29 was dependent on a PKC signalling pathway (299). These data suggest TRIM29 is a survival factor in UV exposed keratinocytes. TRIM29 is highly expressed in SCC compared to normal skin (301). TRIM29 does not have high expression in metastatic melanoma from a study of 6 melanoma cell lines, though this expression was compared to normal skin and not normal melanocytes (301).

TRIM32 has been shown to be highly expressed in transformed and tumourigenic keratinocytes in a mouse skin carcinogenesis model (302). Transduced TRIM32 increased colony number in vitro in an epidermal transformation assay and skin-grafted to athymic nu/nu mice experienced epidermal thickening (302). TRIM32 was found to inhibit tumour necrosis factor α (TNF-α) UVB induced apoptosis of keratinocytes in vitro and in vivo due to its E3 ubiquitin ligase activity (302).

TRIM21 (Ro52) null mice develop severe dermatitis and systemic autoimmune disease, lupus. TRIM21 has been identified as an E3 ubiquitin ligase and is known to target multiple members of the interferon regulatory factor (IRF) family. In line with this TRIM21 -/- mice have increased pro-inflammatory expression of cytokines regulated by 77

IRF proteins including Th17 the pathway (303). Genetic deletion of IL-23/p19 in TRIM21-/- mice resulted in normal skin and protection from systemic autoimmune disease (303).

TRIM16 has been demonstrated to form part of the NALP1 inflammasome complex in UV stimulated keratinocytes and is secreted along with pro-inflammatory markers IL- 1β and caspase-1 (304). TRIM16 has been shown to interact via the RFP domain with caspase-1 and its activation protein, NACTH, LRR and pyrin domain containing protein (NALP) 1. These proteins are both inflammasome components and required for caspase-1 activation. TRIM16 is secreted by keratinocyte cells in a caspase-1 dependant manner, as process that is enhanced by interleukin-1β (IL-1β). Further, TRIM16 has been shown to enhance IL-1β secretion of which TRIM16 is also a binding partner. TRIM16 mediated IL-1β secretion occurs via an alternative pathway suggesting TRIM16 may play a unique role in innate immunity (304). These researchers also found that TRIM16 is highly expressed in epidermal growth factor responsive basal keratinocytes but observed TRIM16 protein down-regulation in the hyper-thickened epithelium of skin wounds (305). To confirm this they found that the stable overexpression of TRIM16 in HaCaT keratinocyte cells promoted early differentiation by the increase in early differentiation markers keratins 6 & 10 and involucrin (305).

1.11.2 TRIM family proteins in innate immunity

A sizeable number of TRIM proteins have been identified as having anti-retroviral activity including TRIM5α, TRIM19, TRIM22, TRIM79α and others at different stages of the virus life cycle (294, 306, 307). The most well documented of these are TRIM19 (promyelocytic leukaemia protein, PML) and TRIM5α (294, 308). In addition, numerous TRIM proteins possessing E3 ubiquitin ligase activity have been identified having anti-viral function (288, 295, 309, 310). TRIM19 (PML) has also been identified as possessing anti-viral activity (311) and the number of TRIM proteins that are characterized as having anti-viral activity is increasing. TRIM proteins mediate their antiviral activity using two distinct mechanisms. Some TRIM proteins function as antiviral restriction factors, directly targeting key proteins to interrupt viral replication. The

78 second branch of TRIM proteins target signalling pathways and modulating the production of cytokines and/or interferons (295).

The most well studied TRIM protein in immune function to date is TRIM5α, which can restrict retroviral infection, including HIV (286, 295, 312). TRIM5α binds to the viral capsid, inducing disassembly soon after the virus enters the host cell (312, 313). Additionally, TRIM5α also functions to activate NF-кB and AP-1, suggesting an innate immune sensory role linked to the detection of retroviruses (314). TRIM25 has been identified as an E3 ubiquitin ligase that is essential for the retinoic acid inducible gene 1 (RIG-I) mediated anti-viral activity (315). In this interaction, the TRIM25 carboxy- terminal SPRY domain interacts with the N-terminal caspase activation and recruitment domains (CARD) of RIG-I thereby delivering an ubiquitin moiety resulting in an increase in downstream RIG-I mediated signalling (315). With the emergence of close to half of TRIM family proteins being identified as positive anti-viral response regulators, it appears that this family is intimately involved with innate immune development (288). TRIM family members have been recognised as regulating immune response through the modulation of pattern recognition receptors (PRR) (316).

TRIM28 (KAP1) has been identified as a STAT binding partner and regulator of STAT3-mediated transactivation. KAP1 has further been identified as a repressor of IFN/STAT1-mediated signalling (317). Further, siRNA-mediated gene silencing of KAP1 enhanced IFN-induced STAT1 dependent IRF-1 gene expression suggesting KAP1 to be a regulator of the IFN/STAT1 signalling pathway (317). KAP1 has also been identified as an inhibitor of HIV integration into the host genome by binding acetylated integrase and inducing deacetylation via a protein complex containing HDAC1 (318).

In a broad study of the TRIM family and viral inhibition, it was shown that multiple TRIM proteins are effective at reducing the expression of hepatitis B virus (HBV) mRNA (319). Furthermore, 8 TRIM proteins were found to inhibit HBV enhancer I and II activity (319). The inhibition of HBV enhancer II by TRIM41 was dependant on TRIM41 E3 ubiquitin ligase activity and C-terminal domain in human hepatoma cells (319). This work also demonstrates that 8 of 38 TRIM family members tested were effective in blocking HBV replication (319).

TRIM genes have also been identified as part of the interferon signature observed in patients with Sjogren’s syndrome (320). This study examines gene expression from RNA isolated from peripheral blood mononuclear cells (PBMC) of 14 females with primary Sjogren’s syndrome and healthy matched controls. A significant up-regulation of IFN-signature genes was observed in primary Sjogren’s syndrome patients compared to controls including a differential gene expression of TRIM5, 6, 14, 22 and TRIM25 (320). Figure 14 provides an overview of the extensive role TRIM proteins play in innate immunity, primarily responding to interferon treatment and mediating anti-viral actions by multiple mechanisms.

Figure 1.14. TRIM proteins are modulated by type I and II IFN’s and regulate innate immune response

1.11.3 TRIM family proteins as E3 ubiquitin ligases

A number of TRIM proteins function as RING type E3 ubiquitin ligases (289) and are important regulators for the degradation of short-lived regulatory proteins (297). TRIM family members are also involved in ubiquitin-like modifications including SUMOylation and ISGylation (289). E3 ubiquitin ligases function as scaffold proteins between the E2 ubiquitin conjugating enzyme and the substrate (297). These E3 ubiquitin ligases may have roles in cell cycle regulation and cell signalling, DNA repair, protein quality control and regulation of transcription. Some TRIM proteins with ubiquitination function have been shown to be important in the pathogenesis of cancer (297). TRIM proteins 13, 19, 24, 28 and 31 have been identified as E3 ubiquitin ligases (296) and TRIM29 (lacking the RING domain) has also been identified as an E3 ligase. Recently, TRIM16 (also lacking the RING domain) has been shown to function with an E3 ubiquitin ligase capacity resulting in homodimerization and auto-poly-ubiquitination (321). In addition, TRIM16 has the ability to heretodimerize with TRIM family members, TRIM24, PML, and TRIM18 (MID1) (321).

1.11.4 TRIM family proteins in cancer

Translocation of TRIM proteins has been implicated in cancer pathogenesis. These include TRIM19 (PML) translocation t(15;17) in acute promyelocytic leukaemia (APL) giving rise to the PML-RARα fusion (322). TRIM19 is located in PML bodies and is important in cellular stress, DNA repair and viral infections (297). Myeloproliferative syndrome (EMS) development is associated with rearrangement of TRIM24 and fibroblast growth factor receptor 1 (FGFR1). In the TRIM24-FGFR1 translocation, constitutive activation of the tyrosine kinase domain of FGFR1 is achieved that results in cellular transformation (297). TRIM24 (TIF1α) is important in the regulation of RARα, thyroid receptor and oestrogen receptors (297). TRIM24 has been validated as a tumour suppressor protein in hepatocellular carcinoma (HCC) in mice by attenuating RARα-mediated transcription. This was performed using a TRIM24-null mouse background with RARα deletion

82 demonstrating suppression of HCC development and restoration of a normal retinoid responsive gene signature (323).

TRIM’s 13, 19, 24, 28 and 29 are involved in the regulation of p53 by means of stability or transcriptional regulation (324). TRIM29 expression results in cell proliferation and resistance to ionizing radiation (IR). Conversely, gene silencing of TRIM29 decreases cell proliferation and sensitizes cells to IR. TRIM29 can bind to p53 and antagonize p53 nuclear activity, abrogating the expression of p53 target genes, p21 and NOXA (324). TRIM’s 19, 24 and 25 regulate nuclear receptors RARα and other hormone receptors implicated in the progression of leukaemia, prostate and breast cancers (297). These findings implicate that TRIM proteins in both oncogene and tumour suppressive roles in the development of numerous cancers. Furthermore, some TRIM proteins function as oncogenes in the context of one cancer type, and tumour suppressors in another cancer type indicating another level of complexity in TRIM protein function. An example of this is TRIM27 which demonstrates complexity in function in the ability to either suppress or enhance tumour development (297). TRIM27 is involved in the regulation of RARα via interaction with PML evidenced in immunoprecipitation and co-localization studies with PLM-RARα (316).

TRIM32 is highly expressed in head and neck squamous cell carcinoma (325). TRIM32 harbours ubiquitin ligase activity and transient transfection of TRIM32 has been shown to bind to tumour suppressor and inhibitor of migration, Abl-interactor 2 (Abi2) resulting in ubiquitination of Abi2, promoting protein degradation and leading to increased HEK293T cell proliferation and cell motility (325). The ubiquitinating properties of TRIM32 were further confirmed by transfection of a dominant negative mutant lacking the ring domain of TRIM32, which no degradation of Abi was observed (325). TRIM32 is also involved in modulating type I interferon response by targeting MITA/STING for ubiquitination (326).

TRIM24 deletion in breast cancer results in apoptosis which is dependent on p53, and TRIM24 is a potential therapeutic target in human breast cancer. In contrast, TRIM13 over-

83 expression causes a stabilization of p53, suggesting a possible tumour suppressor role (297).

TRIM31 was initially identified as a retinoid responsive gene (327). TRIM31 is overexpressed in gastric cancer (296, 327) and is a candidate biomarker of this disease, but retains the ability to negatively regulate cellular proliferation in other cell types. Overexpression in HEK293 cells, TRIM31 is polyubiquinated and degraded via the proteasome pathway. This result was also confirmed in pancreatic cancer cells (296). Taken together, these findings highlight the cell-specific function of the TRIM family member.

It is important to note that there may be functional redundancy in the TRIM family of proteins that can be characterized by knock-out or knock-in mouse models and biochemical analysis to determine if some TRIM proteins compensate for others. Table 1.3 summarizes the alterations in TRIM genes in the pathogenesis of various cancers. The data is adapted from Hatekeyama et al, 2011 (297).

Table 1.3. TRIM gene alterations in cancer pathogenesis

Gene Cancer type Observed alterations Glioblastoma Gene deletion or loss of heterozygosity TRIM8 Laryngeal cancer Correlation with nodal metastatic progression B cell chronic lymphocytic leukaemia Gene deletion TRIM13 Chronic lymphocytic leukaemia Gene deletion TRIM19 Acute promyelocytic leukaemia Chromosome translocation: RARA Papillary thyroid cancer Chromosome translocation: RET Myeloproliferative syndrome Chromosome translocation: FGFR1 TRIM24 Liver cancer Chromosome translocation: BRAF MDS-related AML Overexpression Breast cancer Overexpression Ovarian cancer Overexpression TRIM25 Breast cancer Correlation with poor diagnosis Endometrial cancer Reduced expression Lymphoma Chromosome translocation: RET TRIM27 Breast cancer Overexpression TRIM28 Gastric cancer Overexpression: correlation with poor diagnosis Lung cancer Overexpression Bladder cancer Overexpression Colon cancer Overexpression TRIM29 Ovarian cancer Overexpression Endometrial cancer Overexpression Multiple myeloma Overexpression Gastric cancer Overexpression TRIM31 Gastric cancer Overexpression TRIM32 Head and neck cancer Overexpression TRIM33 Chronic myelomonocytic leukaemia Reduced expression TRIM40 Colon cancer Reduced expression TRIM68 Prostate cancer Overexpression

1.11.5 TRIM protein biomarkers and prognostic markers

Increased expression of TRIM59 has been proposed as a multi-tumour marker in early tumorigenesis in a cohort of 92 tumours of 8 different cancer types (breast, lung, parotid, gastrointestinal, female genital tract, bladder, kidney and prostate) and 75 patients with renal cell carcinoma, determined by IHC analysis (328). Furthermore, a correlation was determined between TRIM59 up-regulation and tumour progression in prostate and kidney cancers using normal adjacent tissues as a control which showed comparatively lower TRIM59 expression (328).

In a cohort of 119 endometrial cancer tissues, Ret finger protein (RFP, TRIM27) high expression is associated with poor OS and PFS (329). Positive immunoreactivity for RPF was determined to be an independent prognostic factor for poor survival in this cohort of 119 patients (329). In the analysis of 108 lung cancer patients by IHC, nuclear RFP expression was detected in 66.7% of cases. RFP correlated with poor prognosis in patients with epidermal growth factor receptor (EGFR) mutation and is proposed as a prognostic factor for this subset of patients (330). RFP has been shown to mediate the ubiquitiniation of PTEN and modulate AKT activation (331). Ubiquitination of PTEN has also been shown to influence protein stability and sub-cellular localization. Over-expression of RFP abrogates the suppressive effect of PTEN on AKT activation. Conversely, RFP gene silencing by siRNA enhances PTEN suppression of AKT in transfected HEK293 cells (331). Ectopic over-expression of RFP in HEK293 cells can induce apoptosis in a mechanism of activation of Jun N-terminal kinase and p38 kinase (332). RFP has also been shown to bind to retinoblastoma (Rb) and inhibit the ability of Rb to promote gene transcription required for differentiation and anti-proliferative function (333, 334).

A two stage skin carcinogenesis study using a sub-carcinogenic dose of DMBA followed by repeated application of TPA was performed in TRIM27-/- and wild-type control mice. TRIM27-/- mice were highly resistant to tumour development (8 of 14 mice were tumour free at 20 weeks) compared to wild-type litter mates (1 of 13 were tumour free) (333). These data indicate that TRIM27 is a modifier of disease incidence and progression and is a potential target for cancer therapy (333). 86

DEAR1 (TRIM62) mutations and homozygous deletions have been detected in breast cancer cell lines and patient samples (335). IHC staining of 123 young female patient samples showed that low DEAR1 expression is an independent prognostic factor for unfavourable prognosis in a 20 year follow-up study and is strongly correlated with the (ER-, PR-, HER2-) triple negative phenotype (335).

1.12 TRIM16 chromosomal location and tissue expression

In humans, TRIM16 is located at chromosomal position 17p11.2. In mice, TRIM16 is located at chromosomal position 11 B2; 11. It is expressed in all human tissues with particularly high expression in the digestive tract, liver, pancreas and blood of non- malignant tissues (336).

1.12.1 TRIM16 protein and post-translational modification

The human TRIM16 gene has 11 protein coding, and 14 processed transcripts (Ensemble).

In humans, the TRIM16 gene encodes five proteins. Three of the five encoded proteins are functional. TRIM16 exists in the α and β isoforms with the α-isoform being identified as the canonical form. The predominant isoform of TRIM16 has a molecular weight of 63,955Da and is comprised of 564 amino acids. TRIM16 has 8 putative post-translational modification sites in both humans and mice (Figure 14, Table 4&5). The crystal structure of TRIM16 is yet to be solved.

Figure 1.15. Schematic of the TRIM16 protein containing two B-Box groups, a coiled- coil domain and a PRY/SPRY grouping

Table 1.4. TRIM16 post-translational modification of human and corresponding mouse protein

Human and mouse TRIM16 protein post-translational modifications

Human modification Mouse modification Sequence Sequence T55-p SEKLGREtEEQDsDS T48 LSKSGEETQEQGHDP S60-p REtEEQDsDSAEQGD H53 EETQEQGHDPAELGA K264-a YRSAEMEkSkQELER K256 HRSLEMEKSKQELER K266-a SAEMEkSkQELERMA K258 SLEMEKSKQELERLA Y429-p QSLYLHRyyFEVEIF Y421 QSLYLHRYYFEVELS Y430-p SLYLHRyyFEVEIFG Y422 SLYLHRYYFEVELSG Y441-p EIFGAGTyVGLTCKG Y433 ELSGGGTYVGLTCKG Y531-p CKFSEPVyAAFWLSK Y523-p CKFSEPVyAAFWLSK

Table 1.5. Amino acid post-translational modification and corresponding region of TRIM16 protein motif

Amino acid Amino acid site Domain modification Threonine 55 - Serine 60 - Lysine 266 Coiled-coil Lysine 264 Coiled-coil Tyrosine 429 B30.2/SPRY Tyrosine 430 B30.2/SPRY Tyrosine 441 B30.2/SPRY Tyrosine 531 B30.2/SPRY

1.12.2 The role of TRIM16 in innate immunity

TRIM16 function was initially identified as a responding protein to both 4-hydroxy- tamoxifen (4HT) and estrogen (E2) in estrogen receptor (ER) transfected human mammary epithelial cells (HMEC) (337). In HMEC cells transfected with mutated ER with a deletion in the second zinc finger, the regulation by 4HT and E2 were markedly different. E2 treatment resulted in a reduction of TRIM16 gene expression, while 4HT increased TRIM16, suggesting that TRIM16 can be modulated via a defective receptor by 4HT, but not E2 suggesting a different mechanism of action for the two estrogenic agents (337).

1.12.3 The role of TRIM’s in type 1 Interferon response

The TRIM protein family has been identified as being regulated by interferons in innate immune signalling and direct involvement in anti-retroviral activity (338, 339). Interferon’s are classified into either type I (IFN-α and IFN-β) or type II (IFN-γ) interferon. Type I interferons are considered to be the first line of defence against viral infection (294). These are expressed by a number of cell types, whereas type II interferons are expressed exclusively by immune cells (313). Toll-like receptors (TLR’s) have the ability to induce type I interferons (294). TRIM62 is involved in the regulation of the TLR4 pathway and siRNA silencing of TRIM62 in primary macrophages resulted in defective TRIF-mediated late NF-кB, AP-1 and interferon production after lipopolysaccharide challenge (339). In turn, the type I interferons have been identified to regulate many members of the TRIM family including those required in innate immune response (286, 313). One example of this is the induction of TRIM22 by type I interferons that is required for the inhibition of HIV particle production in human cells (313). This mechanism was elucidated by siRNA mediated knockdown of TRIM22 and measurement of the ability of interferon to restrict HIV particle release. It was found that the knockdown of TRIM22 resulted in an increased HIV particle release indicating TRIM22 to be a key mediator inhibiting HIV particle production (313). Catalytic regions in the ring domain were further shown to be required

89 for this activity (313). In addition, TRIM5α has been shown to have an interferon- stimulated response element (ISRE) in its promoter region and TRIM5α expression is inducible by IFN-β in HeLa cells giving weigh to a role for TRIM5α in IFN-induced antiviral activity (340). TRIM21 is an interferon-inducible gene that is responsible for regulating cytokine production by targeting IRF8 for ubiquitination suggesting its role in innate immune regulation (341). It has further been shown to sustain IRF3 activation during antiviral response (342). Generally, though TRIM proteins have been identified as mediators of innate immune function, it has yet to be determined whether TRIM proteins are a help or hindrance in the case of autoimmune and auto inflammatory illness (307). What is clear is that the TRIM protein family are broadly involved at various processes of IFN induced innate immune signalling (286, 339, 343). Some TRIM family members have been shown to be regulators of the interferon pathway, or are directly involved in viral restriction (343). Key pathways that are regulated by TRIM family proteins are the NFкB pathway, Toll-like receptor pathway and JAK-STAT signalling pathways (286).

To broadly determine the responsiveness of TRIM proteins to interferon treatment, Carthagena and colleagues treated human primary lymphocytes and macrophages with type I and II interferon (344). They determined 27 of the 72 TRIM genes tested were sensitive to interferon. Their results showed both an up and down regulation of TRIM proteins. TRIM16 exhibited a decrease in gene expression in primary macrophages with type II IFN treatment and no significant change with type I IFN treatment (344).

While TRIM proteins have overwhelmingly been identified as effector proteins of type I interferon, TRIM38 has been identified as a negative regulator of IFNβ signalling by targeting TIR domain-containing adaptor inducing IFNβ (TRIF) for degradation (345). Further, TRIM38 negatively regulates TLR3/4 and RIG-1 mediated IFN-β production by targeting NAP1 (346). The mechanism of action is via toll-like receptor 3 (TLR3) of which TRIF is a key adaptor protein. Over-expression of TRIM38 resulted in decreased expression of TRIF and co-immunoprecipitation studies revealed binding of the TRIM38 PRYSPRY domain with the N-terminus of TRIF, an effect able to be inhibited by proteasome inhibitor, MG-132 (345). It is not yet determined whether other TRIM proteins have the ability to positively or negatively regulate interferon signalling. 90

Interferon inducible TRIM56 has been demonstrated as a responder to intracellular double- stranded DNA (dsDNA) and viruses (347). TRIM56 over-expression enhanced IFN-β promoter activation after 4 hours treatment with poly(dA:dT) and gene silencing of TRIM56 by siRNA abrogated this effect (348). Further, TRIM56 was found to interact with stimulator of interferon gene (STING) as a substrate for lysine-63-linked ubiquitination, inducing STING dimerization and recruitment of anti-viral kinase TBK1 required for induction of IFN-β. These data together suggest TRIM56 is an interferon inducible E3 ubiquitin ligase that mediates dsDNA immune response (348).

1.12.4 TRIM16 mutation in cancer

TRIM16 mutations have been identified in 2.49 and 2.13% of endometrium and stomach cancers respectively (349). In addition, TRIM16 mutation has been identified in 1.16% of all skin cancers (349). Over all cancer types, 71.79% of identified TRIM16 mutations are of the substitution-missense from the Catalogue of Somatic Mutations in Cancer (COSMIC database) database. Following this, substitution-synonymous (15.38%), substitution- nonsense (5.13%) and deletion-frame shift (2.56%) made up the remaining mutations with 5.13% falling into an unknown mutation.

1.12.5 TRIM16 loss of heterozygosity in cancer

Analysis of TRIM16 heterozygosity in the COSMIC database (349) through the Copy Number ANalysis (CONAN database) resources shows TRIM16 loss of heterozygosity (LOH) in a number of cancers. Most notable are pancreatic cancer with 15/16 samples analysed showing a loss of heterozygosity and breast and lung cancers with 35/45 and 120/149 LOH respectively. Table 6 summarizes the incidence of TRIM16 loss over a number of cancer types. Many cancers have a high incidence of TRIM16 loss such as adrenal gland 2/2, gastrointestinal tract 1/1, and biliary tract, 5/6; however, these sample

91 sizes are too small to gain meaningful data on the LOH of TRIM16 in these cancer types, though a larger sample size may indicate significance.

Table 1.6. Incidence of TRIM16 loss of heterozygosity across human cancer samples

Cancer type (unique samples analysed) TRIM16 loss of heterozygosity Incidence Adrenal gland (2) 2/2 1.00 Autonomic ganglia (37) 19/37 0.51 Biliary tract (6) 5/6 0.83 Bone (33) 18/33 0.55 Breast (45) 35/45 0.78 Central nervous system (59) 37/59 0.63 Cervix (12) 2/12 0.17 Endometrium (10) 4/10 0.40 Gastrointestinal tract (site indeterminate) (1) 1/1 1.00 Haematopoietic and lymphoid tissue (127) 42/127 0.33 Kidney (21) 8/21 0.38 Large intestine (39) 22/39 0.56 Liver (9) 3/9 0.33 Lung (149) 120/149 0.81 Oesophagus (22) 13/22 0.59 Ovary (22) 13/22 0.59 Pancreas (16) 15/16 0.94 Pleura (6) 2/6 0.33 Prostate (5) 1/5 0.20 Skin (51) 17/51 0.33 Soft tissue (19) 7/19 0.37 Stomach (21) 15/21 0.71 Testis (3) 1/3 0.33 Thyroid (12) 9/12 0.75 Upper aerodigestive tract (21) 12/21 0.57 Urinary tract (18) 7/18 0.39 Vulva (3) 1/3 0.33

1.12.6 TRIM16 as a candidate tumour suppressor in neuroblastoma

TRIM16 has been demonstrated to acts as a tumour suppressor protein in neuroblastoma via effects on cytoplasmic vimentin and nuclear E2F1 (350). TRIM16 binds vimentin in the cytoplasm and causes a reduction in vimentin protein expression. Conversely, gene silencing of TRIM16 mediated by siRNA results in an increase in vimentin protein expression. Further, the down-regulation of vimentin is required for the ability of TRIM16 to inhibit neuroblastoma cell migration (350). Additional work has demonstrated that TRIM16 inhibits neuroblastoma cell proliferation through cell cycle regulation and localization to the nucleus (351). High nuclear staining of TRIM16 has been observed in differentiating ganglia cells which is absent in the tumour initiating cells. TRIM16 protein translocates to the nucleus in the G1 cell cycle phase after being up-regulated. Cell cycle progression is attenuated through changes in cyclin D1 and p27 (351). These data implicate

TRIM16 as a regulator of G1/S progression and cellular differentiation (351). TRIM16 translocates to the nucleus upon treatment with all-trans retinoic acid and binds to and down-regulates E2F1 protein, reducing E2F1 protein half-life (350). TRIM16 has further been shown to induce apoptosis in neuroblastoma cells (BE-(2)-C) and breast cancer cells (MCF-7) in a caspase-2 dependant manner (352). In addition, TRIM16 has been shown to restore retinoid sensitivity to retinoid resistant breast and lung cancer cells via epigenetic mechanism of histone acetylation and restoration of RARβ2 transcription (353). Though these particular studies lack in vivo confirmation of TRIM16 as a tumour suppressor protein, evidence to date suggests that TRIM16 acts in a tumour suppressive in neuroblastoma.

1.12.7 TRIM16 as a candidate tumours suppressor in squamous cell carcinoma

In the histological progression of squamous cell carcinoma (SCC) from normal skin, TRIM16 was found to be significantly reduced (354). Further, there is a particular loss of nuclear TRIM16 in SCC cell lines compared to primary human keratinocytes. In like

94 manner to neuroblastoma, TRIM16 bound to and down regulated E2F1 and abrogated cell proliferation and cell migration mitigated by down-regulation of binding partner, vimentin. Treatment of SCC cells with retinoid increased nuclear TRIM16 in the retinoid-sensitive primary keratinocytes, but did not do so in the SCC cells that are retinoid resistant (354).

1.12.8 TRIM16 and retinoid signalling

Treatment of neuroblastoma cells results in the translocation of TRIM16 to the nucleus and binding to the RARβ2 retinoic acid response element (RARE), inducing RARβ2 transactivation by its coiled-coil domain (355). TRIM16 also undergoes serine/threonine phosphorylation and an increase in protein stability on treatment with all-trans retinoic acid (355). Figure 15 summarizes the know functions of the TRIM16 protein to date.

Figure 1.16. Summary of the known function of TRIM16 protein

1.12.8.1 RARβ as a tumour suppressor gene in melanoma

Retinoids inhibit melanoma cell proliferation and induce differentiation and retinoid receptors are frequently lost or de-regulated in melanoma (356). The RARβ2 gene is methylated in numerous cancers, and in a number of cancers, hyper-methylated, resulting in transcriptional repression of the gene. This results in a decrease in the RARβ2 protein and a subsequent resistance to retinoid induced anti-cancer treatment. In a study of 226 melanocytic lesions on tissue microarray, RAR and RXR (α, β, γ) were analysed by immunohistochemistry. Researchers noted a significant decrease in RARβ, nuclear RARγ, and RXRα in primary and metastatic melanomas compared to nevi (356).

1.12.8.2 TRIM16 regulation of retinoid signalling in neuroblastoma

Cheung et al have demonstrated that the tripartite motif 16 (TRIM16) protein of the TRIM family is a novel modulator of the retinoid pathway by binding to the retinoic acid response element β (βRARE) promoter region of RARβ2 and transcriptionally activating the gene in neuroblastoma (355). The retinoid acid (RA) response element β is the essential DNA sequence required for retinoid-induced RARβ transcription. Similarly, melanocytes, the cell type that gives rise to melanoma, and also of neural crest lineage (6, 10), thus have the potential to be sensitized to retinoid therapies by TRIM16 de-repression of RARβ2. Taken together, this data suggests a role for TRIM16 in numerous cellular processes and it is of interest to determine if loss of TRIM16 plays a role in melanoma development.

1.13 Innate immunity in melanoma

1.13.1 INF-β and melanoma

The human interferon gene region (containing IFNB1) is encoded on chromosome 9p21, a region that is deleted in numerous cancers including leukaemia, glioma, non-small-cell lung 97 cancer (357), and melanoma (357-361). In addition, homozygous and hemizygous deletion of the 9p region harbouring IFNB1 has been observed in lung cancer (362), acute lymphoblastic leukaemia and gliomas (363). Rearrangement of contiguous sequences has also been observed in lung cancers (362). It is important to note that this locus contains CDKN genes, but it is thought that the interferon gene family represent additional susceptibility genes that are deleted (359). Interferons have been shown to play a major non-redundant role in cancer immunoediting and are co-ordinators of immune system- tumour interactions (364).

In melanoma, IFN-β gene transfer has been shown to induce cell death in vitro (365-367). In vivo data is further encouraging; cationic liposomes containing human IFN-β were injected into xenograft mouse models. IFN-β treatment ablated engrafted tumours after 6 injections at 40 days after the initial injection. Histology analysis evidenced apoptosis in the IFN-β injected tumours demonstrating a highly effective candidate therapy against human melanoma (367). Of all interferons, human interferon-β demonstrates the highest anti-proliferative activity against human melanoma cell lines (367). Additionally, a study using a xenograft model of murine B16F1 melanoma cells also demonstrated a reduction in tumour volume with intratumoral liposome mediated administration of IFN-β and ablation of tumours in 18% of mice, 15 days after the first treatment (366). Histological analysis also indicated apoptosis in the form of cell shrinkage, nuclear condensation and bleb formation (366). Depletion of the natural killer (NK) cell infiltrates identified in immunocytochemical analysis resulted in reduced treatment efficacy suggesting cell death is mediated by these cells (366). In a similar study using B16 melanoma cells, the addition of p19ARF to IFN-β increased cell death and prolonged survival in a mouse xenograft model (365). Both IFN-α2b and IFN-β1a were shown to be important in inhibiting tumour growth and lymph node metastasis by the inhibition of lymphangiogenesis in human melanoma xenografts (368). Mechanistically, these two agents decreased vascular endothelial growth factors VEGF-C and VEGF-receptor 3 protein expression in vivo (368). Use of interferon treatment and siRNA’s targeting STAT3 abrogate the growth of B16 melanomas in vitro, suggesting a means to improve the efficacy of IFN-β treatment in melanoma (369). These 98 results demonstrate the potential for type I interferons in anti-metastatic and anti- proliferative application to melanoma treatment.

1.13.2 Clinical use of Interferon treatment

1.13.2.1 IFN-α2b

The use of adjuvant pegylated IFN-α2b (PEG-IFN-α2b) was approved by the FDA in 2011 for the treatment of resected stage III melanoma (165, 370). It is one of the few effective adjuvant therapies available for advance melanoma (371, 372). In a large study involving 1,256 patients with resected stage II melanoma, patients were randomly assigned PEG-IFN- α2b (n=627) or no treatment observation (n=629). Recurrence free survival (RFS, primary endpoint), distant metastasis free survival (DMFS) and overall survival (OS) were assessed. The median follow up for this study was 7.6 years. 384 patient disease recurrence or death was recorded for the PEG-IFN-α2b group vs. 406 in the observation group. The RFS was 39.1% in the PEG-IFN-α2b group vs. 34.6% in the observation group giving a slight advantage to PEG-IFN-α2b adjuvant therapy (370). Researchers noted that patients with ulcerated disease and lower burden of disease benefitted the most from PEG-IFN-α2b treatment. No difference in OS or DMFS was observed between the two groups (370).

1.13.2.2 IFN-β

In contrast to IFN-2α, IFN-β bound with higher affinity to its receptor, inducing higher levels of associated gene product and a greater degree of apoptosis in melanoma cells in pre-clinical studies (373, 374). IFN-β has been shown to have anti-tumour activity in mouse models of melanoma (375, 376). In a study using IFN-β knock-out mice, it was found that melanoma tumour angiogenesis is inhibited by endogenous IFN-β production by surrounding neutrophils (376). In vitro, DNA demethylation has been shown to overcome

99 resistance to IFN-induced apoptosis in renal carcinoma and melanoma cells (377) and IFN- β suppresses the proliferation of melanoma cells in an autocrine manner (378).

In a phase II trial of IFN-β1a for cutaneous (n=17) and uveal (n=4) melanoma patients, one patient demonstrated durable tumour regression and the response rate was less than 10%. There was a strong induction of IFN-stimulated gene expression in peripheral blood cells such as pro-apoptotic TRAIL and chemokines, CXCL10 and CCL8. Angiogenesis factors, VEGF-A and CXCL5 were both decreased post-treatment. This study demonstrates a pro- apoptotic, anti-angiogenic and immunomodulatory nature of IFN-β1a treatment, but a low overall response rate with the treatment in its current form (379). Furthermore, treatment of low antigen expressing melanoma cells with IFN-β resulted in the increase of melanocytic antigens and expression of class I human leukocyte antigens (HLA’s) (380). An additional study found that TLR4 stimulated B16 melanoma cells yielded smaller tumours when xenographed into nude mice and IFN-β was determined to be the critical factor improving anti-tumour immune response (381). These data suggest IFN-β treatment may be a useful co-treatment with immunotherapies (380).

Cationic liposome-mediated IFN-β gene therapy was used successfully to induce tumour regression in a pilot clinical trial in glioma patients. Two of five patients exhibited more than a 50% reduction in tumours while the remaining three patients maintained stable disease (363). This was observed with intra-tumoral changes to genes involved in apoptosis and immunoresponse. Additionally, tumour cells showed necrotic change and CD8-positive lymphocytes and macrophages infiltrating the tumour tissue and surrounds (363). Mechanistically, IFN-β has been found to down-regulate the DNA repair gene MGMT and sensitize resistant glioma cells to temozolomide (382).

An intriguing clinical case has been reported on the successful treatment of metastasis from nodular melanoma after failure of dacarbazine treatment and radiotherapy. In this case, two treatment cycles of intratumoral injection of IFN-β (Fiblaferon) resulted in tumour regression 14 days after the final injection (3 times weekly at 5 million IU per treatment, courses of 2 weeks and 4 weeks separated by 1 month). A five year follow-up of the patient revealed no sign of melanoma recurrence or systemic metastasis (383). Additional clinical

100 data provide support for the use of IFNβ in maintenance therapy for melanoma treatment, demonstrating that patients treated with a 10 day course of IFNβ at 3x106IU/day had a better overall survival and relapse-free survival than the control group without significant toxic side effects (384).

1.13.2.3 IFN-β transcription via the enhanceosome

Transcription of the IFNβ1 promoter is a tightly regulated process that involves recruitment of the enhanceosome complex to the IFNβ1 promoter (385, 386). Recruitment of proteins ATF2/c-JUN, IRF-3A&C, IRF-7B&D, p50 and RelA are required to form the IFNβ1 enhanceosome (385-388) (Figure 16). The transcriptional synergy demonstrated by the formation of the enhanceosome complex is greater than the transcriptional power of the individual members and a specific order of arrangement is required for transcriptional synergy (388). Enhanceosome formation is particularly dependant of correct ATF2/c-JUN heterodimer orientation (389). Overlap between complex members in binding to the IFNβ1 promoter DNA is required to ensure contact of almost every nucleotide enabling the enhanceosome to function as a single unit of regulation (390) though there is little protein- protein interaction (387). Additionally, co-activators and chromatin remodelling proteins are recruited to aid enhanceosome binding and transcriptional activation (386). Mitogen- activated protein kinase kinase kinase 1 (MEKK1) is responsible for inducing IRF3 and ATF2/c-JUN through the JNK pathway and also inducing NF-кB through the IKK pathway (391).

101

Figure 1.17. Schematic representation of the components of the IFNβ1 enhanceosome

102

1.14 Perspectives and experimental directions

It is clear that to effectively treat metastatic melanoma, a thorough understanding of disease pathology combined with a targeted molecular therapy is required. Identification of novel tumour suppressor proteins and understanding the de-regulated pathways in melanoma are important in order to identify new molecular targets and design new targeted therapies. As cancer treatment moves away from systemic chemotherapies and towards rationally designed targeted inhibitors, the treatments become less cytotoxic. However, there is an increasing need to understand the complexities of targeted treatments as drug resistance invariable develops. A combinatorial approach is essential in targeting such a heterogeneous form of cancer as melanoma. Development of enhancers of clinically used agents is a useful method to partner existing agents with drugs that effectively target a different pathway or position in the same pathway. A personalized approach for the treatment requires comprehensive molecular subtyping of individuals to tailor a treatment regime to individual melanoma pathogenesis. To address these concerns, this thesis comprises two main projects and three result chapters:

1. Identification, characterization of a candidate tumour suppressor protein, TRIM16, in melanoma

Hypothesis I: TRIM16 is decreased as melanoma becomes invasive and loss of TRIM16 is prognostic of poor patient survival. This aim was to determine the clinical relevance of TRIM16 loss and pinpoint which stage of disease this loss occurred. This gives insight as to the role of TRIM16, e.g. if TRIM16 is lost as melanoma becomes invasive we hypothesize that TRIM16 may play a role in suppressing melanoma metastasis. It was also to determine the relevance of TRIM16 as a candidate prognostic marker in melanoma.

Hypothesis II: TRIM16 reduces melanoma cell proliferation in an IFNβ1 dependant manner and IFNβ1 is also decreased as melanoma becomes invasive. This aim was to establish the relationship between IFNβ1 and TRIM16 in vitro and in vivo and to build understanding of the clinical relevance of TRIM16 and IFN-β protein loss in melanoma patients.

103

Hypothesis III: TRIM16 is partially required for vemurafenib drug action and increases vemurafenib efficacy. This aim was to determine the clinical relevance of TRIM16 in the use of the BRAF inhibitor, vemurafenib, and understand the importance of TRIM16 in mediating drug action and increasing drug efficacy.

2. Characterization of TRIM16 as a candidate tumour suppressor protein in melanoma in vivo

Hypothesis IV: TRIM16 is a novel tumour suppressor in melanoma and TRIM16 deletion in the knockout mice will accelerate tumour initiation and progression with skin carcinogen challenge. This aim was to test the in vivo significance of TRIM16 as a tumour suppressor in melanoma by applying the two-stage skin carcinogenesis model to TRIM16 knockout mice to determine tumour incidence and latency.

3. Identification, characterization, studying the mechanism of action of a novel small molecule compound 012 as an enhancer of vemurafenib

Hypothesis V: The compound 012 is an enhancer of vemurafenib in BRAF wild-type cells and is dependent on TRIM16 for the efficacy of combined drug actions. This aim was to identify a small molecule compound that could work in synergy with vemurafenib to increase drug efficacy. The aim was to determine the mechanism of action for compound 012 to develop novel combination therapy for the treatment of melanoma

Hypothesis VI: Compound 012 is an enhancer of vemurafenib in vivo. This aim was to determine the in vivo efficacy, and building on the in vitro characterization of compound 012 as an effective enhancer working in synergy with vemurafenib, and the potential to reduce tumour burden in a mouse xenograft model.

104

Chapter 2

Material and Methods

105

1 Materials

Melanoma cell lines were culture in Dulbecco’s Modified Eagle Media (DMEM) supplemented with 5% fetal calf serum (Life technologies, NSW, Australia). Cell transfection of siRNA and plasmid were performed using Opti-MEM (Life technologies, NSW, Australia). Specific siRNAs (siTRIM16-1 and siTRIM16-2) were custom synthesized by Dharmacon (ThermoScientific, Waltham, MA, USA). The pcDNA3.1 plasmid was obtained from Life Technologies (NSW, Australia) and the pCMV6 plasmid was obtained from Origene (Rockville, MD, USA). Transfection reagent was Lipofectamine 2000 (Life technologies, NSW, Australia). The TRIM16 antibody for western blotting was from Bethyl Laboratories (TX, USA). The TRIM16 antibody used for immunohistochemistry was custom made (Biosource, CA, USA). The IFNβ1 antibody for western blotting and immunohistochemistry was from Abcam (NSW, Australia). The secondary biotinylated antibody for immunohistochemistry was obtained from Dako (Vic, Australia). BrdU Cell Proliferation Assay Kit, BCA Protein Quantitation Assay Kit and Supersignal West Pico Chemiluminescent Substrate were from Thermo Scientific (IL, USA). Hybond ECL Nitrocellulose membranes were purchased from GE Healthcare (Rydalmere, Australia). Criterion gels were from Bio-Rad (Sydney, Australia). RNeasy Mini RNA Isolation Kit and Miniprep Kit for Plasmid Purification were from Qiagen (Vic, Australia). Alamar blue reagent was purchased from AccuMed International (IL, USA). AmpliTaq Gold Buffer and Polymerase were purchased from Applied Biosystems (Vic, Australia). All primers were custom designed and obtained from Sigma-Aldrich (MO, USA).

106

2 Methods

2.1 General techniques

2.1.1 Tissue culture conditions

Melanoma cell lines, A375 and G361 were purchased from ATCC. Lines, CHL-1, IPC-298, SK-Mel2 and MeWo were kindly gifted from Professor Grant MacArthur at the Peter MacCallum Institute, Melbourne. Lines, Mel-JD, Mel-RM, Mel-CV, M4405 and MM200 were kindly gifted from Professor Xu Dong Zhang at the University of Newcastle. All lines were cultured in Dulbeco’s modified eagle medium (Life Technologies Australia, NSW, º Australia) supplemented with 5% foetal calf serum and incubated at 37 C/5% CO2. TRIM16 overexpression was achieved by transient transfection of the pcDNA3.1/myc-his plasmid containing the full-length TRIM16 cDNA under a CMV promoter using lipofectamine 2000 (Life Technologies Australia, VIC, Australia). TRIM16-1 siRNA 5’AGTAATTCACCATGCAGGTTT-3’, TRIM16-2 siRNA 5’TCTCCCTCCTGCATTTGTGTT-3’ were custom designed and transfected using siControl non-targeting pool (Thermo Scientific, Waltham, MA, USA) as a control at 20 nM using lipofectamine 2000 as the transfection agent.

2.1.2 Plasmid preparation

Construction of TRIM16 is described in Cheung et al., 2006 (355), where the full-length cDNA was subcloned in to the pcDNA3.1(-)/MYC/His vector at the EcoRI and KpnI multi- cloning site. TRIM16 full-length-GFP was purchased from OriGene (Rockville, MD, USA) and was cloned into a pCMV6-AC-GFP vector. Plasmids were transformed into competent bacteria and grown on ampicillin (pcDNA3.1) or kanamycin (pCMV6) resistant plates. Plasmid containing bacteria was grown in LB broth overnight and plasmid was prepared using a Purelink HiPure Plasmid filter Maxiprep kit (Life technologies, NSW, Australia). DNA concentration was measured by Nanodrop ND1000 spectrophometer.

107

2.1.3 TRIM16 plasmids and siRNA transfection

TRIM16 overexpression was achieved by transient transfection of the pcDNA3.1/myc-his plasmid containing the full-length TRIM16 cDNA under a CMV promoter using lipofectamine 2000 (Life Technologies Australia, VIC, Australia). For a 10cm dish, 5µg of plasmid DNA was transfected with Lipofectamine 2000 in an optimized 1:3 ratio for all transfections. For 96 and 6 well plates 0.1µg/well and 1µg/well were transfected, respectively.

TRIM16-1 siRNA 5’AGTAATTCACCATGCAGGTTT-3’, TRIM16-2 siRNA 5’TCTCCCTCCTGCATTTGTGTT-3’ were custom designed and transfected using siControl non-targeting pool (Thermo Scientific, Waltham, MA, USA) as a control at 20 nM using lipofectamine 2000 as the transfection agent. Cultures were incubated for 6 hours before the addition of 20% FCS to double the seeding volume giving a 10% FCS final concentration. RNA and protein isolation was at 24, 48 and 72 hours.

2.1.4 Co-transfection of TRIM16 plasmid and siRNA’s, IFNβ1 and c-Jun

Mel-JD and G361 cells were seeded in 96 well plates at 2000 cells/well and 2500 cells/well, respectively and 6 well plated at 1x105 cells/well. Cells were transiently co- transfected with TRIM16 plasmid (as described above) and siRNA against IFNβ1 or c-Jun. Both siIFNβ1 and sic-Jun were used at a concentration of 20 nM in OptiMEM. Cells were incubated for 6 hours before an equal volume of media containing 20% FCS was added. Transfections lasted for 24 and 48 hours and cell proliferation and viability was assessed by BrdU incorporation and trypan blue. Cell lysates were prepared at 24 and 48 hours for molecular analysis.

108

2.1.5 Tissue microarray construction

Reference sections of the donor tissue block were cut, H&E stains performed and the slides marked with a 1mm circle to identify areas of tumour. Triplicate tissue cores 1 mm in diameter were taken from the donor paraffin block using the marked section as a reference and then arranged in a blank paraffin block by a MTA-1 Manual Tissue Arrayer (Beecher Instruments, Sun Prairie, WI, USA). Once constructed, the microarrays were baked for 30 minutes at 37° C on a glass slide to even out the surface and fuse the paraffin in the cores and donor block.

2.1.6 Cytoplasmic and nuclear protein fractionation

Cytoplasmic and nuclear protein fractionation was achieved using the NE-PER nuclear protein extraction kit (Thermoscientific, Rockford, IL, USA). Briefly, a cell pellet containing 2x106 cells was lysed with 200µL of CER I. The lysate was vortexed and chilled on ice for 10 minutes and 11µL of CERII added. The lysate was vortexed and chilled on ice for 1 minute. The suspension was centrifuged and the supernatant (cytoplasmic fraction) transferred to a new tube. The remaining pellet was washed with CER I to remove residual cytoplasmic extract. Ice-cold NER was added and the pellet was vortexed for 15 seconds every 10 minutes and placed on ice in between for a total of 40 minutes. The sample was centrifuged and the supernatant (nuclear extract) transferred to a new tube.

2.1.7 BCA Protein assay

Cell lysates were prepared using RIPA buffer and protein concentration were determined in cleared lysates using the Pierce BCA method (Thermoscientific, Rockford, IL, USA).

Briefly, the cell pellet was thawed on ice and 1-200 µL RIPA buffer with protease inhibitor added (10 µL stock in 1 mL RIPA). Cells were incubated on ice for 30 min with pipetting

109 up and down. Lysate was spun at 12,000 x g, 4°C for 20-30 min. The cleared lysate was transferred to a fresh tube and protein content assayed using the BCA method (1/5 dilution). BSA standards were prepared at 2, 1, 0.5, .25, .125 mg/mL with RIPA buffer as a blank. BCA reagent was prepared 50:1 (A:B) 9.8 mL to 200 µL. Protein was incubated at 37°C for 30 min and analysed spectrophotometrically at 570 nm with 450 as a reference wavelength. Protein was standardized to contain 15 µg in a 24 µL final volume with sample buffer (4 µL) included. The standardized protein was boiled for 5 min at 95°C.

2.1.8 Western blotting

Protein lysate was standardized using the BCA protein quantitation assay kit as per manufacturer’s instructions (Thermo Scientific, IL, USA). Western blotting used the following antibodies: polyclonal TRIM16 (Bethyl laboratories, TX, USA), anti-myc tag antibody (Cell Signalling Technology, Danvers, MA, USA), rabbit polyclonal actin antibody (Sigma, St Louis, MO, USA) and anti- GAPDH antibody (Abcam, NSW, Australia). The primary TRIM16 antibody was used at 1:700 dilution and the secondary rabbit antibody at 1:4000 dilution. The primary GAPDH antibody was used at 1:6000 dilution and the secondary mouse antibody at 1:10000 dilution.

Samples were run on a Bio-Rad criterion (Tris-HCl) 10.5-14% gradient gel (Bio-Rad, NSW, Australia) for 2.5 hours at 100V in 1 x running buffer. Transfer was performed at 4°C for 2 hours onto a nitrocellulose membrane in 1 x transfer buffer. Ponceau solution was used to assess equal protein loading. Membrane blocking was with 10% skim milk in TBST for 1 hour. Membranes were washed three times in TBST. Primary antibodies were applied overnight at 4°C. Membranes were washed and secondary antibody applied for 4 hours at ambient temperature. Membranes washed three times in TBST and enhanced chemiluminescence (SuperSignal) performed.

110

2.1.9 BrdU incorporation cell proliferation assay

Melanoma cells were seeded into 96 well plates at a density of 2 x103 cells/well. Cells were transiently transfected with pcDNA3.1 TRIM16 or EV in 96 well plates. At specified time points, 5-bromo-2-deoxyuridine (BrdU) (Life Technologies Australia, VIC, Australia) 1:1000 was added to proliferating cells. BrdU incorporation was allowed for two hours after which time the culture media was aspirated and cells were fixed. Incorporated BrdU was labelled with a peroxidase-conjugated anti-BrdU before adding the peroxidase substrate. Colour was allowed to develop before reading the absorbance of the wells with a multiplate reader with a test wavelength of 370 nm and a reference wavelength of 490 nm. Rate of proliferation for the transiently co-transfected cells were determined by analysing the raw data using the GraphPad Prism version 6.01 (GraphPad software Inc, La Jolla, CA)

2.1.10 Alamar Blue cell viability assay

Alamar blue assays were performed on melanoma cells that had been transiently transfected for 24, 48 and 72 hours, or melanoma cells that had been treated with drugs for 72 hours. At the specified time, Alamar blue was added (1:10) dilution to proliferating cells in the culture media and a baseline colourimetric measurement taken using a Victor3 multiplate reader. Cells were allowed to proliferate for 4-6 hours at 37 ºC with 5% CO2 incubation, after which time a measurement was taken to record the colour change due to Alamar Blue dye cleavage. The reading was normalized to the baseline reading for each well and colourmetric changes were analysed compared to their respective controls using GraphPad Prism version 6.01.

2.1.11 ELISA apoptosis assay

Apoptosis was measured by the quantification of histone-complexed DNA fragments (mono- and oligonucleosomes) from the cell cytoplasm using a cell death detection 111

ELISAPLUS kit from Roche Applied Science (Mannheim, Germany). Briefly, cells are lysed and intact nuclei pelleted by centrifugation. An aliquot of supernatant is transferred to streptavidin-coated wells in duplicate. Streptavidin bound nucleosomes in the supernatant bind with two monoclonal antibodies, anti-histone (biotin-labeled) and anti-DNA (peroxidase-conjugated). Bound complexes are washed and incubated with a peroxidase substrate. The amount of resulting coloured product is measure spectrophotometrically.

2.1.12 Scratch wound migration assay

For the scratch assay, G361 cells were seeded at 1x105cells/6well plate and allowed to culture overnight. Cells were transiently transfected with pcDNA3.1 expressing empty vector or TRIM16. After 24 hours transfection, a scratch wound was applied using a pipette tip and a baseline image obtained. Scratch wound closure was monitored over a period of 24 hours. For the transwell assay, NHEM cells were seeded at 2x104cells/transwell chamber (Life Technologies Australia, VIC, Australia) in a cell suspension of 20 nM siTRIM16-1, siTRIM16-2 or siControl. After 5 hours transfection, the transfection complexes were removed and melanocyte media placed into the upper chamber. Conditioned media was placed in the lower chamber as a chemo-attractant. Melanocytes were allowed to migrate for 48 hours before the chamber was fixed and stained for cell counting.

2.1.13 Transwell migration assay

The trans-wells were seeded with 20,000 cells/well and transfected in-well. Gene silencing using siTRIM16 transfection was performed as previously described. 20 nM siRNA was used and transfection period was 5 hours after which. After 5 hours the transfection complexes were removed and the cell monolayer was washed with media and fresh media was added to the cells. Conditioned media was added to the lower chamber to function as a chemo-attractant, fresh NHEM media was added to the top chamber. Cells were allowed to

112 migrate for 48 hours before being fixed and counted for the number of cells that had migrated through the transwell.

2.1.14 PCR cancer pathway array

RNA was extracted from TRIM16 transfected G361 cell lines using the PureLink RNA mini extraction kit (Ambion, VIC, Australia), in accordance with the manufacturer’s instruction. Cancer PathwayFinder PCR array (SABiosciences, PAHS-033A) was performed according to the manufacturer’s instruction using an ABI Prism 7900 sequence detection system (Applied Biosystems, VIC, Australia).

2.1.15 Quantitative real-time polymerase chain reaction (RT-qPCR)

Wells were prepared in duplicate and no template control (NTC) of water prepared for each primer. Primers used targeted the TRIM16 gene with β2M as the housekeeping gene. Samples were run on an ABI 7900 system (Absolute quantification and melt) following standard PCR conditions. A dissociation curve was produced to determine agreement between duplicates of dsDNA dissociation (by estimating the reduction in fluorescence of SYBR green as the double strand dissociates). qPCR was the average of two wells and analysis by the DDCt method (392) giving a log Ct value relative to 1 (no change).

2.1.16 Microarray and analysis

The microarray study for the 5-aza work was performed on cDNA using the human Affymetrix 2.0 array prepared from A375 and Mel-CV cells both treated with control or 10 µM of 5-aza for 48 hours. The microarray study for the determination of compound 012/vemurafenib candidate genes was performed in duplicate using Mel-JD cells treated with control (DMSO), compound 012 (4 µM), vemurafenib (5 µM) and combination compound 012/vemurafenib treated for 6 hours. The platform was the human Agilent 113

8x15K array. Analysis was performed to determine differential gene expression and gene set enrichment analysis (GSEA) using software provided by the Broad Institute (MA, USA).

2.1.17 PCR of genomic DNA and sequencing

Genomic DNA was extracted using TRIreagent (Sigma-Aldrich, NSW, Australia) in accordance with the manufacturer’s instruction. PCR of six TRIM16 exons was performed using the following primer pairs: Exon 1 Forward (F) 5’-TGGCCGAGCTTCCTCTGGGA- 3’, Reverse (R) 5’- ATGAATGGTCCCCAAGCACTCAC3’. Exon 2 F 5’- GTGCATTGGGCTCCTTCCTCCTTA-3’, R 5’- CTGGCTGACCCAGGCTGGTCTT-3’. Exon 3 F 5’- TTGGCTGCCTTCCACCCCCA-3’, R 5’- TGGGGCAGCTGTGGGATGCC- 3’. Exon 4 F 5’- GCAAGTTCTGCTGTTTCCTTTTCTGC-3’, R 5’- CACTATAGTCCATGGCCCAGAATGC-3’. Exon 5 F 5’- CTGGACAGGTGTGCATTCACAACTC-3’, R 5’- ACTGATATCAACAACTGGAGAAGCGG-3’. Exon 6 F 5’- TCTCCTGCCTTCTGTGTCTCCTCAG-3’, R 5’- AGCCAGCTACCATCAGCAGTTATTTC-3’. PCR product clean-up was performed using ExoSAP-IT (Affymetrix, CA, USA). Sequencing was performed at the Australian Genome Research Facility (Westmead, NSW, Australia) using BigDye Terminator sequencing (Life Technologies Australia, VIC, Australia).

2.1.18 Chromatin immunoprecipitation assay

Chromatin immunoprecipitation was performed using the Millipore ChIP kit (Merck/Millipore, Germany) with modification to manufacturer’s instruction. G361 cells were sonicated at 45 min using high sonication and 45 min low sonication in the DNA fragmentation step. Specific antibodies for TRIM16 (Bethyl, TX, USA), c-JUN (Cell Signaling, Massachusetts, USA catalogue #9165) and IgG (Dako, Victoria, Australia) were

114 used for respective protein pull-down. DNA from the pull-down was purified using the MiniElute PCR purification kit in accordance with the manufacturer’s instructions (Qiagen, VIC, Australia). PCR protocol was as follows: 95C: 2min/95C:45 sec, 58C:45 sec, 72C:45 sec/72C:7 min (x35 cycles). Primer pairs were as follows: IFNβ1 promoter (forward 5’- AGGTCGTTTGCTTTCCTTTGC-3’, reverse 5’-GACAACACGAACAGTGTCGC-3’) with negative control primers (forward 5’ACTGCCTGCATTAAGGGCAA-3’, reverse 5’- ACAGAAGGCCTCATCACTGC-3’).

2.1.19 Cycloheximide and 5-aza-2’-deoxycytidine (5-Aza) treatment

For the cycloheximide (CHX) chase assays, A375 and Mel-CV cells were seeded at a density of 1x104 cells/6well plate and cultured overnight. Baseline samples were taken in the absence of CHX and CHX applied at 100μg/mL with lysates prepared at 2, 4, 6, 8, 14 and 24 hours for western blot analysis using TRIM16 and GAPDH specific antibodies. For the 5-aza treatment, cells were seeded at 1x106 cells in T-75 flasks and at the following day, 30 μM of 5-Aza was applied and cultured for 24 hours (harvested for RNA) and 72 hours for protein lysates.

2.1.20 MG-132 proteosomal inhibitor treatment

For the MG-132 treatment, A375, Mel-CV and G361 cells were seeded at 1x106 cells in T- 25 flasks. The following day, 30 µM of MG132 (Biomol, USA) was applied and cultured for 8 and 16 hours for protein lysates.

2.1.21 Clonogenic assay

Melanoma cells are seeded at 100 cells/well in 6-well plates. Cells are cultured with control (DMSO only), compound 012 at 4 µM, vemurafenib at 0.5 µM (BRAF mutant cells) or 5 µM (BRAF wild-type cells) or combination compound 012/vemurafenib in 2 mL of media. 115

Colonies are allowed to form over 14 days and are then fixed and stained with 2 mL of 6% gluteraldehyde and 0.5% crystal violet solution prepared in Milli-Q water. A colony was determined to be 50 cells or more. Colonies were counted for each treatment group.

2.1.22 Immunofluorescence

Melanoma or melanocyte cells were plated in an 8-well chamber slides and cultured to until 80% confluent. Barriers were removed and cells washed with PBS. Slides were fixed with 4% paraformaldehyde in PBS for 15 minutes at room temperature. Slides were washed with PBS and placed in cold methanol for 10 minutes on ice. Slides were washed with PBS and blocked with 10% FCS for 1 hour at room temperature. Primary antibody 1/200 polyclonal TRIM16 was applied in 10% FCS/PBS for 1 hour at room temperature. The secondary antibody (1/1000 rabbit Gt488 (Alexa Fluor)) was applied in 10%FCS/PBS for 1 hour. Mounting media with Dapi (Life technologies, NSW, Australia) was applied and coverslip secured before viewing on the Olympus Fluoview FV1000 fluorescent microscope.

2.1.23 Immunohistochemistry

Following Institutional Review Board Human Research Ethics Committee approval (Protocol X11-0023 & HREC/11/RPAH/32 and Protocol X10-0300 & HREC/10/RPAH/530), archival formalin-fixed, paraffin-embedded tissue blocks of excised human skin specimens were retrieved from the Department of Tissue Pathology and Diagnostic Oncology at the Royal Prince Alfred Hospital, Sydney, Australia. Five- micrometre-thick tissue sections were cut from the paraffin blocks and placed on saline- coated slides. Whole tissue sections from compound and dysplastic nevus, in situ melanoma, dermal invasive melanoma, lymph node and distant metastasis were probed with specific TRIM16 custom made antibody at 1:500 dilution (Biosource, CA, USA) and IgG at 1:500 dilution (Dako, VIC, Australia) was used as a negative control. A rabbit- biotinylated secondary antibody was used at 1:500 dilution (Dako, VIC, Australia). Each

116 stage was represented by 13 samples giving a total of 91 samples. Tissue microarrays of 170 patient lymph node metastasis samples (in triplicate) were stained as described for the whole tissue sections. Samples were blindly graded independently by two people on an arbitrary scale of 0-4, with 4 being the highest staining intensity and 0 as negative staining. Tissues were graded at two different sites to allow for heterogeneity of staining intensity within the tissue. Data were analysed using the Student’s t-test. Results were considered statistically significant with a p value of <0.05.

2.1.24 Drug synergy assays

Assays to determine additive or synergistic drugs used the drug IC50 value and a 1:1.25 ratio (BRAF mutant cells, compound 012:vemurafenib) or 1:12 ratio (BRAF wild-type cells, compound 012:vemurafenib). Treatment duration was for 72 hours. Synergy was used by applying the Calcusyn algorithm (Biosoft) where a value <1 was designated as synergy and a value >1 was determined as additive effect.

2.1.25 Microsomal stability assay

A reaction mixture of 1640 µL of purified water, 460 µL of 0.5 M potassium phosphate (pH 7.4), 115 µL NADPH regenerating system solution A (BD Biosciences, NSW, Australia), and 23 µL NADPH regenerating system solution B (BD Biosciences, NSW, Australia) was prepared. 447.5 µL of reaction mixture was pipetted into two eppendorf tubes (one compound performed in duplicate). 0.92 µL of 0.5 mM test compound (prepared in DMSO) was added to each tube. 0.25 mg of human liver microsomes were added to each tube and mixed. 70 µL was removed and stored on ice as a baseline sample. Tubes were incubated at 37oC and 70 µL of sample was aspirated at 5, 15, 30, 60 and 90 minutes of incubation and placed in tubes that contained 70 µL of acetonitrile. Samples were stored on ice. All samples were centrifuged at 10,000xg for 3 minutes at 4oC. The supernatant was

117 stored at -80oC before analysis by mass spectrometry (Bioanalytical Mass Spectrometry Facility, UNSW, Sydney, Australia).

2.1.26 Statistical analysis

Average replicated of three individual experiments was conducted for the tissue culture and molecular studies. Data were analysed using the unpaired Student’s t-test or log-rank test where appropriate for the analysis. Results were considered statistically significant with a p value of <0.05. Statistical analysis was performed using GraphPad Prism version 6.01 (GraphPad software Inc, La Jolla, CA). All statistical tests were two sided.

2.2 Generation of TRIM16 skin specific knock-out mice

2.2.1 Preparation of TRIM16 floxed mice

For modelling of the function of TRIM16 in melanomagenesis, we chose a LoxP-Cre knockout system that involves knocking in a LoxP-Cre construct with flanking homology to the TRIM16 gene resulting in a ‘floxed’ mouse and gene deletion by crossing with a Cre expression system under the control of a promoter for the tissue of interest. We have produced two knockout mice from this system. The first is the mouse produced from crossing our floxed mice with a Cre delete mouse under the control of Ketatin14. Keratin14 is expressed in mitotically active basal layer cells and expression is down-regulated as cells differentiate (393). Knocking out TRIM16 in keratin14 expressing cells ensures a specific knockout of TRIM16 in the epithelial cells (394), (395), (393), while the remaining mouse tissues express TRIM16. Knocked-out tissues include the skin, tongue and cornea epithelial cells (393). The second model is a full tissue TRIM16 knockout by crossing the floxed mice with B6-Tg (Cmv-Cre)1Cgn deleter strain for the deletion of Lox-P sites in all tissues. This ensures that all tissues are knocked out for TRIM16 expression. 118

2.2.2 Development of skin lesions using a two-stage skin carcinogenesis model

TRIM16 wild-type, heterozygous and homozygous skin-specific littermate mice between 7- 9 weeks of age were shaved on the right dorsal flank prior to initiation. The skin lesions of the mice were initiated with a single topical dose of 7,12-Dimethylbenz(a)anthracene (DMBA) at 97.5 nmol in 0.2 mL of acetone. Two weeks following initiation, mice were treated with either 0.2 mL 6.8 nmol of 12-O-tetradecanoylphorbol-13-acetate or acetone control twice weekly for a period of 21-42 weeks. The skin lesion development of the mice was scored for weekly throughout the study.

2.2.3 Design of the TRIM16 knock-out construct

TRIM16 knockout mice were generated at Ozgene Australia, utilizing a construct designed to ablate the full-length TRIM16 coding sequence flanked by lox P sites (Figure 2.1). We crossed TRIM16 wild-type/flox knockout (KO) mice with skin-specific Cre mice, which express Cre under the control of the human keratin 14 promoter, to allow excision of full- length TRIM16 and Neo selection cassette in tissues expressing keratin-14. The heterozygous and homozygous of skin specific TRIM16 KO mice are viable and fertile.

119

Exon and intron schematic

Figure 2.1. The FLOX construct and homology arms. The Flox construct features two homology arms of 7.4 (pink) and 7.2 (yellow) kb that flank the Lox-P cassette. The Lox-P cassette contains the Lox-P cre recognition sites (red) containing exons 1-6 of the murine TRIM16 gene. The Neo cassette (orange) for Flox identification by PCR is positioned between the Lox-P sites.

120

2.2.3.1 Genotyping PCR design to determine TRIM16 knockout

Genotyping of the floxed mice required the recognition of the Neo sequence that is present in the knocked-in flox construct (Figure 2.1). Amplification of the Neo cassette confirms at least one copy of the flox construct, while amplification of Exon 6 of the wild-type TRIM16 (spanning an intronic sequence), confirms the presence of wild-type TRIM16. Combination of these two primer pairs in a multiplex assay allows the determination of the three genotypes where wild-type mice will only amplify the Exon 6 PCR, homozygous floxed mice will only amplify the neo cassette, and heterozygous floxed mice will amplify for both Exon 6 and the Neo cassette (Figure 2.2A). For the skin specific TRIM16 knockout mice a KRT14Cre delete stain is used to excise the flox construct and remove the TRIM16 gene. As this Cre delete strain is under the control of a KRT14 promoter, only the keratinocytes expressing Keratin 14 will be knocked out for the TRIM16 gene. For the full- tissue knockout mice, a B6-Tg (Cmv-Cre)1Cgn delete mouse is selected that ensures the TRIM16 gene is knocked out in all tissues (Figure 2.2B). To determine the deletion of the TRIM16 gene, a knockout (KO) PCR is used, which is designed to amplify spanning the region where the TRIM16 was before excision. Confirmation of the present of the TRIM16 gene is determined by amplification of the first exon (Exon 1). Thus, combination of the KO PCR and Exon 1 PCR distinguishes between the presence of all three genotypes (Figure 2.2B). Genotyping of mice was performed using DNA extracted by Chelex (Sigma- Aldrich, NSW, Australia) from tail tips using the following primer pairs (Table 2.1)

121

Table 2.1 Primer design for genotyping of TRIM16 knockout mice

PCR Direction Sequence 5'-3' Neo Forward AGAGGCTATTCGGCTATGACTGG Reverse GGACAGGTCGGTCTTGACAAAAAG Exon 6 Forward TGCCTTGTGGGGGTCACTTGGA Reverse GGTGTTCCCAGGGCGTGGTG KO PCR Forward GAGCCTCGTCCTGTCTGAGTAAC Reverse AAACCAAGAAGTGCCAGAAATA Exon 1 Forward GAGCCTCGTCCTGTCTGAGTAAC Reverse TCTTCTTTTTCTGCTGGGATAG β2 microglubulin Forward TCTCACTGACCGGCCTGTAT Reverse GGAACTGTGTTACGTAGCAG

122

Figure 2.2A.

Figure 2.2B.

Figure 2.2. Schematic of PCR to determine presence of the flox construct and hetero or homozygosity. A). Heterozygous mice harbouring the flox construct are crossed, giving rise to flox/flox offspring, identified by the Neo cassette and absence of the Exon 6 band present in wild-type TRIM16 mice. Flox/flox mice are crossed with a deleter strain, KRT14-Cre (skin-specific TRIM16 knockout mice) or B6-Tg (Cmv-Cre)1Cgn (Full-tissue TRIM16 knockout mice). B). The offspring from this cross are heterozygous from TRIM16 knockout in their respective tissues. Identification of TRIM16 knockout is determined by the KO PCR and Exon1 PCR. Offspring from heterozygous mice gives rise to three genotypes, TRIM16 wild-type (+/+), TRIM16 heterozygous (+/-), and TRIM16 homozygous (-/-) mice.

123

2.2.4 Development of skin lesions using a two-stage skin carcinogenesis model

TRIM16 wild-type, heterozygous and homozygous skin-specific littermate mice between 7- 9 weeks of age were shaved on the right dorsal flank prior to initiation. Mice were initiated with a single topical dose of 7,12-Dimethylbenz(a)anthracene (DMBA) at 97.5 nmol in 0.2 mL of acetone. Two weeks following initiation, mice were treated with either 0.2 mL 6.8 nmol of 12-O-tetradecanoylphorbol-13-acetate or acetone control twice weekly for a period of 21 weeks. Mice were scored for skin lesion development weekly throughout the study.

2.2.5 Culture of primary keratinocytes

Primary keratinocytes were cultured from the tails of 7 week old mice. Mice were euthanized with CO2 and tail removed and placed in 80% ethanol. Tail skin was removed with sterile dissecting tools and cut into 1cm sections and washed in epidermal keratinocyte culture media (CnT) (CellnTec, Bern, Switzerland). Sections were cultured overnight in CnT media with 5mg/mL dispase and penicillin/streptomycin antibiotic. After rubbing tissue to release a single cell suspension, the tissue was centrifuged and the resulting pellet cultured in CnT media with antibiotics in collagen IV (BD Biosciences, NSW, Australia) coated dishes. Western blotting for TRIM16 protein was performed from cultured keratinocytes using a TRIM16 specific antibody. Gene expression of TRIM16 was performed by RT-PCR using TRIM16 primers forward 5’- GGCTCTCTGGTTTGACTTGG-3’, reverse 5’-GGTTTCTTCGGTGGAAAACAA- 3’ and β2M primers forward 5’-TCTCACTGACCGGCCTGTAT-3’, reverse 5’- GGAACTGTGTTACGTAGCAG-3’ in PCR multiplex.

124

Chapter 3

Identification and characterization of a candidate tumour suppressor protein, TRIM16, in melanoma

125

This chapter of my thesis presents data on the loss of TRIM16 expression in melanoma in vitro and in vivo and provides insight on the mechanism and consequence of repressed TRIM16 expression. The first hypothesis we put forward stated that TRIM16 protein expression is decreased during melanoma progression and is associated with poor patient survival. Here we sought to understand the clinical relevance of TRIM16 loss and determine the strength of TRIM16 as a candidate prognostic marker. The second hypothesis we put forward states that TRIM16 reduces melanoma proliferation in a manner dependant on IFNβ1 and that IFNβ1 is decreased in melanomagenesis in like-manner to TRIM16. Here we propose a novel mechanism of TRIM16 loss in melanoma and provide new insight into the function of IFNβ1 in melanoma progression in vitro and in vivo. The third hypothesis we put forward is that TRIM16 is partially required for vemurafenib drug action. This provides a clinical basis for the importance of TRIM16 in reducing melanoma proliferation and as a potential molecular target of BRAF inhibitor, vemurafenib.

3.1 Introduction

The incidence of cutaneous malignant melanoma is rapidly increasing with an American study estimating an 8-fold increase in women, and a 4-fold increase in men between 1970 and 2009 (18). Melanoma is an extremely aggressive malignancy making up only 4% of total skin cancers, but it is responsible for 80% of skin cancer deaths (48). While early detection and removal by surgical excision results in an almost complete cure, the patients with metastatic melanoma have a poor prognosis with median survival estimated between 6-9 months (396, 397). Metastatic melanoma is highly chemo-resistant with traditional chemotherapies such as dacarbazine, achieving only a 5% response rate (226, 397). New targeted melanoma treatments, notably vemurafenib, offer a much higher response rate due to specific BRAFV600E targeting in patients (226). However, disease progression is invariably observed at 5-7 months, further highlighting the evasive nature of metastatic melanoma and a need for a more comprehensive understanding of disease pathology (227).

With targeted therapy becoming the mainstay of melanoma treatment (55, 84), an intimate understanding of the molecular pathology of melanoma is paramount to determine patients that are likely to respond to treatment, such as BRAFV600E patients responding to vemurafenib. 126

Determination of patients that are likely to respond to specific treatment requires an understanding of the molecular subtype that preferentially responds to a given treatment (72, 221, 222). This results in a greater number of clinical responses to therapy and also highlights molecular subtypes in need of development of targeted therapy. Specifically, understanding of oncogenes that drive melanomagenesis (84, 123) and tumour suppressors that are lost during disease pathology are valuable in designing targeted inhibitors to oncogenes, or screening for compounds that can restore tumour suppressor proteins. As melanomas are deregulated in multiple pathways and are able to by-pass drug inhibition and reactivate oncogenic signalling (57, 251, 252), a detailed understanding of major as well as minor molecular signalling pathways is required. Furthermore, de-regulated pathways in melanoma do not occur in isolation, but impact other signalling nodes within the cell (114). Loss of tumour suppressor proteins can result in failure of the cell to undergo apoptosis, un-regulated cell proliferation or up-regulation of an oncogenic protein that is no longer suppressed (47). Tumour suppressor proteins may be lost by DNA methylation, DNA mutation or targeting by an ubiquitin ligase or a combination thereof (47). By gaining understanding of the mechanism of tumour suppressor loss, a strategy to restore the tumour suppressor may be devised. This may be in the form of restoration of gene expression by epigenetic agents if the tumour suppressor gene is methylated or histones acetylated (160, 398), or restoration of tumour suppressor expression by small molecules (399), targeting a melanoma oncogene that down- regulates a tumour suppressor protein (400), or restoring anticancer pathways usually activated downstream of a tumour suppressor. Treatment with small molecules that increase tumour suppressor half-life may allow increased tumour suppressor activity and be beneficial for the subtype of patients that have lost protein activity. Furthermore, tumour suppressor proteins may also act as prognostic indicators of patient survival and determine melanoma subtype that is responsive to a particular treatment modality. Re-establishing tumour suppressors may confer sensitivity to some anti-cancer agents (265, 401) providing a rational basis for a combination therapy approach.

The TRIM family of proteins has been implicated in the pathogenesis of numerous cancers, both as oncogenes and tumour suppressor proteins (297). To date, no TRIM protein has been described in the pathogenesis of melanoma (297). TRIM16 is a member of the tripartite motif family of proteins comprising over 100 members that typically carry the characteristic RING b-box-Coiled-coil architecture (297, 322). TRIM16 has been identified as having a role in innate immune function

127

and being identified as part of the NALP1 inflammasome (304). TRIM16 is secreted by keratinocyte cells in a caspase-1 dependent manner, a process that is enhanced by interleukin-1β (IL-1β). In addition, TRIM16 has been shown to enhance IL-1β secretion (304). TRIM16 is highly expressed in epidermal growth factor responsive basal keratinocytes and TRIM16 protein down- regulation is observed in the hyper-thickened epithelium of skin wounds (305). TRIM16 has been shown to increase differentiation markers such as keratins 6&10 and involucrin in keratinocytes (305). Intriguingly, in the histological progression of squamous cell carcinoma (SCC) from normal skin, TRIM16 was found to be significantly reduced in vivo and TRIM16 reduces SCC cell migration in vitro (354). TRIM16 protein has emerged as acting as a tumour suppressor in neuroblastoma (350), a neuroectodermal tumour type like melanoma. In neuroblastoma, TRIM16 has been shown to inhibit cell proliferation by binding to nuclear E2F1 and down-regulating cytoplasmic vimentin. Furthermore, the down-regulation of vimentin is required for TRIM16 to inhibit neuroblastoma cell migration (350). A recent study has demonstrated that TRIM16 inhibits neuroblastoma cell proliferation through cell cycle regulation and TRIM16 protein localization to the nucleus (402). High nuclear staining of TRIM16 has been observed in differentiating ganglia cells which is absent in the tumour initiating cells (402). The role of TRIM16 in the pathogenesis of melanoma is presently unknown. Elucidating the expression, functional phenotype and mechanism of action of TRIM16 in the pathogenesis of melanoma is the focus of this chapter.

The human interferon gene region (containing Interferon-β1 (IFNβ1)) is deleted in numerous cancers including leukaemia, glioma, non-small-cell lung cancer (357), and melanoma (357, 361). In melanoma, IFN-β gene transfer has been shown to induce cell death in vitro (365-367) and suppress melanoma cell proliferation in an autocrine manner (378). Both IFN-α2b and IFN-β1a were shown to be important in inhibiting tumour growth and lymph node metastasis by the inhibition of lymphangiogenesis in human melanoma xenografts (368). Clinically, The use of adjuvant pegylated IFN-α2b (PEG-IFN-α2b) was approved by the FDA in 2011 for the treatment of resected stage III melanoma as an adjuvant therapy (165, 370). Mechanistically, these two agents decreased vascular endothelial growth factors: vascular endothelial growth factor (VEGF)- receptor 3 and VEGF-C protein expression in vivo (368). Of all interferons, human IFNβ demonstrates the highest anti-proliferative activity against human melanoma cell lines (367). In contrast to IFN-2α, IFN-β bound with higher affinity to its receptor, inducing higher levels of associated gene product and a greater degree of apoptosis in melanoma cells in pre-clinical studies

128

(379). In a phase II trial of IFN-β1a for cutaneous (n=17) and uveal (n=4) melanoma patients, one patient demonstrated durable tumour regression and the response rate was 10%. There was a strong induction of IFN-stimulated genes in peripheral blood cells such as pro-apoptotic TRAIL, chemokines CXCL10 and CCL8. Angiogenesis factors, VEGF-A and CXCL5 were both decreased post-treatment (379). This study demonstrates a pro-apoptotic, anti-angiogenic and immunomodulatory nature of IFN-β1a treatment, but a low overall response rate with the treatment in its current form (379). Additional clinical data also provide support for the use of IFN-β in maintenance therapy for melanoma treatment demonstrating that patients treated with a 10 day course of IFNβ at 3x106 IU/day had a better overall survival and relapse-free survival than the control group without significant toxic side effects (384).

3.2 Results

3.2.1 TRIM16 protein expression is reduced in metastatic melanoma and correlates with overall survival risk in Stage III disease

As TRIM16 has been identified to act as a tumour suppressor in a neural crest derived cancer neuroblastoma, and squamous cell carcinoma, the first question we sought to address was the expression levels of TRIM16 in melanoma compared to normal melanocytic tissue. Gene expression data retrieved from the National Centre for Biotechnology Information (NCBI) indicated that TRIM16 mRNA expression is decreased in metastatic melanoma compared to normal skin or normal melanocytic nevi (GEO accessions GDS1375; (403)). This provided additional rationale for a more comprehensive investigation of TRIM16 expression in melanoma.

129

3.2.1.1 TRIM16 protein expression is reduced during melanomagenesis

We next sought to determine TRIM16 protein expression in human patient samples of melanoma and compound nevus. This approach provides data on the expression of TRIM16 in clinical samples and, if reduced, provides a basis to investigate whether TRIM16 is important in the progression of melanomagenesis. Furthermore, the loss of TRIM16 protein in a particular melanoma stage can be investigated for prognostic significance, which adds further evidence to the potential of the TRIM16 protein to act as a growth and/or metastasis suppressor in melanoma. Other clinically relevant data can be assessed (tumour thickness, ulceration, mitotic index, age, and gender) to determine whether specific correlations are observed between protein expression and a given parameter.

We investigated the pattern of TRIM16 expression by immunohistochemistry (IHC) among 91 patient tumour samples, representing all clinical stages (Figure 3.1A&B). A significant progressive reduction of TRIM16 protein expression level was found for dermally invasive primary melanoma, lymph node metastases and distant metastases, when compared to compound nevi (P<0.001) (Figure 3.1 A&B).

130

Figure 3.1A

131

Figure 3.1B

Figure 3.1. TRIM16 protein expression is reduced in metastatic melanoma compared to nevi and primary melanoma A) The level of TRIM16 protein expression was analysed by immunohistochemistry by two observers blinded to clinical outcome using a polyclonal TRIM16 antibody and graded on an arbitrary scale of 0-4. TRIM16 expression score was collected from 91 patient samples with 13 samples per stage. The statistical comparisons were performed using the Student’s t-test. A statistically significant difference for TRIM16 expression level between melanoma and compound nevus is indicated by *p<0.05, **p<0.01 or ***p<0.001. B) Representative immunohistochemistry images of TRIM16 staining in two samples of dysplastic compound nevus and distant metastasis

132

3.2.1.2 High TRIM16 protein expression is associated with favourable patient outcome in a cohort of lymph node metastasis patients

We then sought to determine the prognostic significance of TRIM16 loss in a cohort of lymph node metastasis patients (n=170) to determine TRIM16 protein expression level and patient overall survival. This cohort was chosen as TRIM16 is significantly lost as melanoma becomes invasive (Figure 3.1A) suggesting TRIM16 may play a role in melanoma metastasis. As we are assessing TRIM16 as a candidate tumour suppressor protein, understanding the prognostic impact of TRIM16 protein expression is required. We performed immunohistochemistry in a cohort of stage III melanoma patient samples to determine TRIM16 protein expression levels. Low TRIM16 (<1 median value) expression in melanoma cells metastatic to lymph nodes was significantly associated with poor prognosis (Hazard ratio 0.6322 with 95% confidence of 0.4322 to 0.9383) vs. (Hazard ratio 1.582 with 95% confidence of 1.066 to 2.314) (P = 0.0255, two-sided log-rank test; n = 170). Our data show that low TRIM16 expression gives a poor prognosis with a median survival of 16 months compared to a median survival of 59 months for a high TRIM16 score (>1 median value), *p=0.02 (Figure 3.2A).

133

Figure 3.2A

Figure 3.2B

134

Figure 3.2. High TRIM16 is prognostic of favourable overall survival in lymph node metastasis patients A) Kaplan-Meier curve from immunohistochemistry staining for TRIM16 protein expression with the value cut off at 1 (median value) giving ‘high’ and ‘low’ staining. Patient median survival was 59 months for TRIM16 expression level of >1, and 16 months for TRIM16 expression level of <1. Data represents overall survival and was analysed with the Log- rank test. B) Representative immunohistochemical staining of lymph node metastases showing lymph node metastasis with high or low TRIM16 expression. The total number of patients was n=170.

135

3.2.1.3 TRIM16 protein expression does not correlate with primary tumour ulceration or thickness

We also sought to determine if TRIM16 protein expression was linked to either the presence of ulceration or primary tumour thickness in this cohort (Figure 3.3 A&B). This allows investigation into whether TRIM16 may be prognostic at an earlier melanoma stage. A cohort of 87 stage III patients that had ulceration present in the primary tumour were assessed for TRIM16 expression levels by immunohistochemistry. Between the groups of ulceration absent or present, there was no significant difference in TRIM16 expression (Figure 3.3A). Indeed, we observed a decrease in TRIM16 expression and higher primary tumour thickness lending to the hypothesis that TRIM16 loss may facilitate melanoma cell invasion. However, in an analysis of 94 stage III patient samples, high TRIM16 score (>1) vs. low TRIM16 score (<1) revealed no significant difference (p=0.26) (Figure 3.3B) so this hypothesis could not be confirmed. TRIM16 loss is summarized schematically in Figure 3.4.

136

Figure 3.3A

Figure 3.3B

137

Figure 3.3. TRIM16 expression is not higher in lymph node metastasis from ulcerated melanoma compared to non-ulcerated melanoma. TRIM16 expression was measured (arbitrary scale of 0-4) in a cohort of 87 and 94 lymph node metastasis patients with data available on the presence of ulceration (A) and tumour depth in mm with in the primary melanoma (B), respectively. Statistical analysis was performed using the Student’s t-test.

138

These data show that TRIM16 is lost during melanomagenesis and is significant as the melanoma becomes invasive in the dermis. This data does not determine if TRIM16 loss is a driving event or a passenger event in the progression of melanoma. One definitive method of addressing the significance of TRIM16 loss as a driver or metastasis is the generation of a mouse model of knockout of TRIM16 in the melanocytes crossed with a known model to induce melanoma and assess the degree of metastasis with TRIM16 loss. Tumour ulceration and increased primary tumour thickness are associated with poorer patient prognosis (40, 404), though there is a trend towards lower TRIM16 expression and ulceration and increased primary tumour thickness, this did not reach significance. From the analysis of this cohort, TRIM16 does not associate with ulceration or tumour thickness and is unlikely to serve as a biomarker marker in early melanoma. This data does indicate that TRIM16 is associated with favourable patient outcome in a cohort of lymph node metastasis patients and provides a basis for further investigation to determine if TRIM16 may serve as a prognostic marker in stage II & III disease.

139

Figure 3.4. Schematic summary of TRIM16 loss and melanoma stage. TRIM16 is decreased between nevus and metastatic melanoma and is significantly decreased between nevus and dermal invasive melanoma <1 mm. This highlights loss of TRIM16 may be an early event in melanomagenesis immediately before metastasis occurs. 140

3.2.2 Nuclear and total TRIM16 protein is lost in melanoma cells compared to normal melanocytes

To determine the loss of TRIM16 in cultured cells and to perform forced biology studies, we obtained a panel of melanoma cell lines and normal human epidermal melanocyte (NHEM) cells. Importantly, we have a selection of BRAFV600E, NRASQ61R/L and BRAFWT/NRASWT melanoma lines as these are the dominant oncogenes that are screened for clinically and provide insight into responsiveness to clinical treatments. For example, only patients presenting with BRAFV600 positive tumours are treated with the BRAF inhibitors, vemurafenib or dabrafenib (231, 405). Our melanoma cell panel also has additional mutational status, including p53 mutant melanoma cell lines (Figure 3.5). We analyzed the TRIM16 protein expression level by Western blotting in 10 melanoma cell lines, and showed a significantly lower level compared to the normal human epidermal melanocyte cell line (NHEM) (Figure 3.6). The two protein bands at 67kDa and 64kDa on the Western blot using the TRIM16-specific antibody represent two isoforms of the TRIM16 protein caused by differential post-translational modification. The immunoblot demonstrates a clear decrease in TRIM16 expression in all melanoma cell lines compared to normal melanocytes (Figure 3.6). TRIM16 expression did not correlate with any mutational status. Cell lines, Mel-CV and M4405 retained relatively high levels of TRIM16 expression. The two protein bands at 67kDa and 64 kDa on the Western blot using the TRIM16-specific antibody represent two isoforms of the TRIM16 protein and may be caused by differential post-translational modification. Different isoforms and post-translational modification states are present for the cell lines, while the normal melanocyte cells retain all states of TRIM16 (Figure 3.6).

Because previous research has shown that TRIM16 sub-cellular localization is important for cellular function, we sought to determine the localization of TRIM16 in melanoma cells compared to melanocytes. In neuroblastoma cells, TRIM16 nuclear loss has been shown (350) and, TRIM16 has been demonstrated to interact with and down-regulate cytoplasmic vimentin. The nuclear activity of TRIM16 in neuroblastoma binds to, and inhibits E2F1, thus reducing cell cycle progression (350). We found that the TRIM16 expression in melanocytes is both nuclear and cytoplasmic. In melanoma cells, TRIM16 expression level is entirely cytoplasmic (Figure 3.7A). This suggests that loss of TRIM16 in the nucleus may be important in the progression of

141

melanoma. No change was seen in levels of vimentin or E2F1 expression with TRIM16 forced over-expression (data not shown), suggesting a different mechanism of action of TRIM16 in melanoma compared to neuroblastoma.

142

Figure 3.5. Melanoma cell line panel and known mutations in key melanoma pathogenesis genes. A panel of genetically diverse melanomas are assembled harbouring the BRAFV600E mutation, NRASQ61R/L or melanomas with no BRAF or NRAS mutation. This panel also displays diversity of p53 mutational status. Normal human epidermal melanocytes (NHEM) cells are wild- type for all mutations. Blank boxes indicate the mutational status is unknown.

143

Figure 3.6. TRIM16 protein expression is decreased in melanoma cell lines compared to normal human epidermal melanocytes Immunoblotting for TRIM16 protein expression was performed in proliferating normal human epidermal melanocytes (NHEM) and 10 melanoma cell lines of diverse genetic background. Whole cell lysates were prepared and immunoblots were probed with a specific anti- TRIM16 antibody. An anti-Actin antibody served as a loading control. Corresponding densitimetry plot shows TRIM16 protein expression as a percentage of β-Actin loading control. ****p<0.0001 of all melanoma lines compared with NHEM cells. 144

Figure 3.7A

Figure 3.7B

Figure 3.7. Nuclear TRIM16 localization is lost in melanoma cells compared to normal human melanocytes. A) A panel of normal melanocytes and 7 melanoma cells were fractionated into cytoplasmic and nuclear fractions and an immunoblot performed using anti-TRIM16, anti- GAPDH (cytoplasmic marker) and anti-Histone-H3 (nuclear marker) antibodies. B) Endogenous TRIM16 in normal melanocytes (NHEM) and melanoma cells (Mel-CV) and A375) were tagged with GFP protein and fixed using 4% paraformaldehyde and sub-cellular localization was observed using fluorescence microscopy. N=3

145

3.2.2.1 TRIM16 is reduced by the proteosomal degradation pathway in melanoma cells

To determine the mechanism of TRIM16 loss in melanoma, we used the proteasome inhibitor MG- 132. Cell lines, A375, Mel-CV and G361 were treated with 30 µM of MG-132 for 8 and 16 hours and TRIM16 expression was determined by western blotting. The proteasome inhibitor MG-132 increased TRIM16 protein levels in melanoma cells at both 8 and 16 hours (Figure 3.8). This indicates that TRIM16 is degraded by the proteasome pathway in melanoma cells.

146

Figure 3.8. TRIM16 is reduced by the proteosomal degradation pathway in melanoma cells. TRIM16 degradation pathway was analysed by proteasome inhibitor, MG-132. A375, Mel-CV and G361 cells were treated with 30 µM of MG-132 for 8 and 16 hours, and western blots from whole cell lysates were probed for TRIM16, Cyclin E2, and GAPDH. Data was provided by Jessica Koach.

147

These data indicate that nuclear TRIM16 is decreased in melanoma compared normal melanocytes. This suggests that TRIM16 may be required in the nucleus to mediate function. TRIM16 is required in the nucleus of neuroblastoma cells to inhibit the cell cycle and reduce cell proliferation (351). Furthermore, we have determined that TRIM16 is degraded in melanoma cells via the proteosomal degradation pathway, which is in agreement with the pathway of degradation of TRIM16 in neuroblastoma (321).

3.2.3 TRIM16 protein half-life is reduced in melanoma by an epigenetic mechanism

In the previous section, we have established the loss of TRIM16 in a cohort of human melanoma patient samples and melanoma cell lines and determined that high TRIM16 expression is prognostic of favourable patient outcome. In addition, TRIM16 is shown to be lost from the nucleus of melanocyte cells as an early event before melanoma migration occurs.

Tumour suppressor genes can be mutated, methylated (106, 398), gene deleted (113), targeted for degradation (116), or excluded from the required subcellular compartment (406-408) as mechanisms that abrogate their tumour suppressor function. Understanding of the mechanism of loss of tumour suppressor function provides a rational basis for development of targeted treatment for protein restoration. For example, the tumour suppressor p53 that is mutated in many cancers may be restored by small molecules (409, 410). Restoration of down-stream targets of tumour suppressor (411) or inhibiting proteins that bind and degrade the target tumour suppressor (412) are also approaches for restoring tumour suppressor function. This highlights the many paths to tumour suppressor inactivation and a targeted approach to protein and signalling restoration.

To gain understanding into the mechanism of TRIM16 loss in melanoma, we sought to characterise the loss of TRIM16 compared to normal melanocytes by examining protein half-life, mutation status and DNA methylation of the gene promoter region. These data will provide insight into approaches to restore TRIM16 function and candidate agents that may be investigated to

148

increase TRIM16 half-life. Melanocytes (NHEM), normal human fibroblasts (WI-38) and melanoma cell lines, Mel-CV and A375 were treated with cycloheximide (CHX) for up to 24 hours. TRIM16 protein expression was determined. The TRIM16 protein half-life in melanoma cell lines was 12 hours and 6 hours, compared with more than 24 hours in NHEM and normal human fibroblasts (WI38) (Figure 3.9A&B).

To determine whether loss of TRIM16 expression was also due to DNA methylation, we treated NHEM, WI38 and four melanoma cell lines with the demethylating agent, 5-Azacytidine (5-Aza), at 30 µM for 72 hours (Figure 3.10A&B). TRIM16 protein levels were significantly increased in the four melanoma cell lines following 5-Aza treatment (p<0.001 for Mel-JD and Mel-RM; p<0.01 for A375 and Mel-CV), compared with normal cells (Figure 3.10A). TRIM16 transcription also increased following 5-Aza treatment (Figure 3.10C) in melanoma cell lines (p=0.05), indicating direct or indirect de-repression of the TRIM16 gene transcription upon treatment with the demethylating agent.

149

Figure 3.9A

Figure 3.9B

Figure 3.9. TRIM16 half-life is shorter in melanoma cells compared to normal human melanocyte and fibroblast cells TRIM16 protein half-life was analysed in melanoma cell lines (Mel-CV, A375), NHEM and WI-38 human fibroblasts following CHX treatment. Cells were treated with cycloheximide at a final concentration of 100 μg/ml over 24 hours. At the specified time points, the cells were harvested and total cellular protein was extracted for western blotting. The western blots were probed with an anti-actin antibody as a loading control.

150

Figure 3.10A

Figure 3.10B

Figure 3.10C

151

Figure 3.10. TRIM16 protein expression and half-life is decreased in melanoma cell lines compared to normal melanocytes and is increased following treatment with the demethylating agent, 5-aza-cytidine. A) Melanoma cell lines (Mel JD, Mel RM, A375, Mel-CV), NHEM and WI-38 human fibroblasts were treated with solvent control or 30 μM 5-Aza for 72 hours. In the top panel, the whole cell lysates were prepared and western blots were probed with anti-TRIM16 and anti-Actin antibodies. B) TRIM16 protein expressions in 5-Aza treated samples were quantified and represented as percentage of solvent control. C) TRIM16 mRNA expression was measured by RT-qPCR in melanoma cell lines, Mel-CV, A375, Mel-JD and Mel-RM. Data is expressed as fold change compared to untreated controls. A statistically significant difference for TRIM16 expression level between melanoma and compound nevus is indicated by **p<0.01 or ***p<0.001. N=3. Data was provided by Jessica Koach.

152

In this section we have established that TRIM16 half-life is reduced in melanoma cells compared to normal melanocytes and cultured human fibroblasts. We have shown that TRIM16 protein and mRNA is increased with treatment by the demethylating agent, 5-aza, indicating the potential for the TRIM16 promoter to be methylated in melanoma. In the next section, we explore additional mechanisms of TRIM16 inactivation and loss in melanoma.

3.2.3.1 E3 ubiquitin ligase, NEDD4, is decreased with TRIM16 overexpression, but does not correlate with TRIM16 expression in melanoma

We have previously shown that TRIM16 can be ubiquitinated (321). As we have demonstrated that TRIM16 half-life is reduced in melanoma compared to normal melanocyte and fibroblast cells (Figure 3.9), we aimed to identify candidate E3 ligases may target TRIM16 for degradation. We treated melanoma cells for 72 hours with 10 µM 5-Aza, and then performed a cDNA expression profiling. Candidate genes that may target TRIM16 for degradation were selected based in differential gene expression. The E3 ubiquitin ligase, NEDD4, was the most strongly repressed transcript, by 12-fold. NEDD4 is one candidate gene that has known E3 ubiquitin ligase function (413, 414). As we hypothesize that TRIM16 may be targeted by an E3 ubiquitin ligase, reducing its half-life, we examined the NEDD4 protein expression in melanocytes and melanoma cells. It should be noted that the expression level of NEDD4 was low in the melanoma cell line panel (Figure 3.11A). Cell lines CHL-1 and IPC-298 were selected for study as they expressed relatively high levels of NEDD4 compared to other melanoma lines (Figure 3.11A). Using transient transfection of siRNA targeting NEDD4 in CHL-1 and IPC-298 melanoma cell lines, we found that TRIM16 protein expression is increased with knockdown of NEDD4 expression (Figure 3.11B). This provides further evidence towards the hypothesis that TRIM16 may be targeted by the E3 ubiquitin ligase, NEDD4, for degradation. In addition, the knockdown of NEDD4 reduced cell proliferation (Figure 3.11C).

153

Figure 3.11A

154

Figure 3.11B

Figure 3.11C

155

Figure 3.11. Knockdown of NEDD4 increases TRIM16 protein and decreases cell proliferation. A) Immunoblot of endogenous protein expression using anti-NEDD4 with anti- Actin antibody as a loading control. B) NEDD4 mRNA was silenced using siRNA targeting NEDD4 transcript and an immunoblot against anti-TRIM16 and anti-NEDD4 was performed using anti-Actin as a loading control. C) Cell proliferation is measured by BrdU incorporation in IPC- 298 and CHL-1 melanoma cell lines at 48 hours. ***p<0.0001 compared with siRNA control.

156

NEDD4-2 is increased in 34 of 79 cutaneous melanomas and 9 of 32 nodal metastatic tumours, but is absent in melanocytes (415). Furthermore, gene silencing of NEDD4-2 in G361 cells reduced cell proliferation in vitro (415). The function of NEDD4 in melanoma is not presently known. Here, we demonstrate the potential relationship between TRIM16 and the E3 ubiquitin ligase, NEDD4. It is possible that NEDD4 directly targets TRIM16 for degradation. As the increase in TRIM16 protein expression with gene silencing of NEDD4 was not large, and NEDD4 expression in melanoma cells was not high, investigation of NEDD4 as an agent that targets TRIM16 for degradation was not pursued further in favour of determining additional mechanisms of TRIM16 loss in melanoma.

157

3.2.3.2 TRIM16 is mutated at low levels in melanoma cell lines

We next performed DNA sequencing of the endogenous TRIM16 coding and promoter regions for 9 melanoma cell lines. The TRIM16 gene in melanoma cell lines, IPC-298, MM200, A375, G361, SK-Mel-2, Mel-CV, Mel-JD, Mel-RM, and CHL-1 was sequenced. We found an E121D missense mutation in exon 1 of IPC-298, in the B-Box1 domain of the TRIM16 protein in 1/9 melanoma cell lines (Figure 3.12). The E121D mutation has also been reported in a publicly available database, Catalogue of Somatic Mutations In Cancer (COSMIC) in 1/71 stomach cancers (349). In addition, a missense mutation at V123M was reported among 1/31 primary melanomas (349). The B-box1 domain of TRIM protein, MID1, structurally resembles the ring domain of E3 ubiquitin ligases (416). TRIM16 has a 46% sequence similarity with MID1, can adopt ring domain-like folds and has E3 ligase function in vitro and in vivo (321), indicating that it is possible that mutation in TRIM16 B-box1 may perturb E3 ligase activity.

158

Figure 3.12. TRIM16 harbours a mutation in B-Box-1 in 11% of melanoma cell lines. PCR amplification and BigDye Terminator DNA sequencing of all six TRIM16 exons was used to determine TRIM16 DNA mutation and corresponding amino acid change at E121D in the IPC-298 melanoma cell line. Melanoma cell lines MM200, A375, G361, SK-Mel-2, Mel-CV, Mel-JD, Mel-RM, and CHL-1 were also sequenced. Additional mutation at V123M was sourced from the Catalogue of Somatic Mutations in Cancer (COSMIC) database, present in the MZ7-mel melanoma cell line.

159

Taken together, these data indicate that the half-life of TRIM16 is significantly reduced compared to normal melanocytes and fibroblasts.TRIM16 is also mutated in 1/9 melanoma cell lines tested and 1/31 tumours accessed by the catalogue of somatic mutations in cancer resource. Though this data is limited and conclusions are unable to be drawn from small numbers of cell lines analysed, it does allow for the possibility of TRIM16 to be mutated in melanoma at a low level. Mutation rates of tumour suppressors genes, PTEN (12%) and p16INK4a (19%) (117) are comparable to the conceivable mutation rate for TRIM16 (11%). Analysis of a larger panel of cell lines and/or patient samples is needed before a mutation rate of 11% is given to TRIM16 in melanoma.

160

3.2.4 TRIM16 protein expression reduces melanoma cell proliferation and migration and TRIM16 loss in melanocytes results in increased migration capacity

To gain insight into the functional significance of TRIM16 loss in melanoma, a series of overexpression or gene silencing experiments were conducted. A limitation to the approach of protein over-expression is the artificial consequences that may occur by ectopic expression of a protein that may exceed normal biological expression limits. Gene silencing is a more informative method of understanding the relevance of protein function. However, as melanoma cells have low expression levels of TRIM16 protein, an over-expression approach is required to determine the functional consequence of re-introducing the protein back into the cell. We have performed informative gene silencing studies in the high TRIM16 expressing normal melanocytes to gain understanding of the role TRIM16 may play in metastasis. This is due to the observed loss of TRIM16 in melanoma patient samples that reaches statistical significance at the dermal invasive melanoma stage (Figure 3.1); suggesting TRIM16 loss may play a role in melanoma metastasis and/or invasion. In Chapter 4, we have developed two murine TRIM16 knockout models that are used to investigate the loss of TRIM16 in an in vivo setting. We have developed a keratinocyte specific knockout model and full-tissue knockout model to assess the role of TRIM16 in melanoma development.

To determine the effects of TRIM16 overexpression on melanoma cell growth, we transiently transfected 5 melanoma cell lines with the pcDNA3.1 empty vector or pcDNA3.1/TRIM16 /myc.His expression vector. Overexpression of TRIM16 reduced cell proliferation for all 5 melanoma cell lines. We found that over-expression of TRIM16 in melanoma cells reduces cell proliferation measured by a reduction in BrdU incorporation in A375, Mel-JD, CHL-1 (***p<0.001) and Mel-RM and MeWo (**p<0.01) melanoma cell lines (Figure 3.13A) compared to empty vector controls. In addition, an induction of apoptosis is observed in normal melanocytes (**p<0.01), A375 and G361 melanoma cell lines (*p<0.05) (Figure 3.13B). Transient overexpression of TRIM16 for 48 hours in G361 melanoma cells caused a reduction in cell migration into a scratch wound in vitro over an 8 hour period (**p<0.01) (Figure 3.14).

161

Figure 3.13A

162

Figure 3.13B

Figure 3.13. TRIM16 over-expression reduces melanoma cell proliferation and induces apoptosis A) A panel of melanoma cell lines were transiently transfected with pcDNA3.1/TRIM16/myc.His expression vector (TRIM16) or pcDNA3.1 empty vector (EV) for 48 hours. Cell proliferation was measured by BrdU incorporation, with the lower panel immunoblot confirming TRIM16 transfection using an anti-Myc tag antibody. Anti-Actin antibody is used as a loading control. B) melanoma cell lines, G361 and A375 and normal melanocytes, NHEM were transiently transfected with TRIM16 or EV expression vectors for 48 hours and induction of apoptosis was determined using an enzyme linked immunosorbent (ELISA) colorimetric assay. A statistically significant difference between empty vector (EV) and TRIM16-transfected cells is indicated by **p<0.01; ***p<0.001. Apoptosis was quantified by colorimetric ELISA assay of plasmid transfected cell lysates at 24 hours N=3

163

164

Figure 3.14. TRIM16 overexpression reduces melanoma cell migration G361 cells were transiently transfected pcDNA3.1/TRIM16/myc.His expression vector (TRIM16) or pcDNA3.1 empty vector (EV) for 48 hours. Representative phase contrast micrographs of the closure of scratch-wounded confluent cultures of melanoma cells photographed either immediately or 8 hours after wounding. The histogram displays the proportion of wound closure over 8 hours from the time of wounding. **p<0.05 N=3

165

Here, we demonstrate a functional role for TRIM16 in reducing melanoma proliferation and migration. This demonstrates that re-introduction of TRIM16 into melanoma cells by transient overexpression actively reduces proliferation and migration and shows the potential for loss of TRIM16 to contribute to melanomagenesis.

3.2.4.1 TRIM16 gene silencing reduces melanocyte proliferation and increases migration

Our main interest was in assessing the role of TRIM16 in cell migration. In the previous section, we performed an over-expression of TRIM16 in a low TRIM16 expressing melanoma cell line, G361. We performed the scratch assay, applying a scratch wound and allowing melanoma cells to migrate into the space. We show that TRIM16 over-expression reduces melanoma cell migration (Figure 3.14). Here we examine the role of TRIM16 in melanocyte cell migration. Knockdown of TRIM16 using TRIM16-specific siRNAs in the normal cellular counterparts of melanoma cells, NHEMs, resulted in a significant increase in cell migration (***p<0.001) compared to siRNA control (Figure 3.15A). Curiously, TRIM16 gene silencing also resulted in reduced cell proliferation by BrdU incorporation in melanocytes (Figure 3.15A) and melanoma cells (Figure 3.16A).

166

Figure 3.15A

Figure 3.15B

167

Figure 3.15. TRIM16 gene silencing reduces melanoma cell proliferation and increases migration of NHEM cells. NHEM cells were transfected with siRNA siControl or two different TRIM16 specific siRNAs (siTRIM16-1 and siTRIM16-2) for 24 hours. A) Cell proliferation was measured by BrdU incorporation into the DNA. B) The transwell migration assay was performed using conditioned media as a chemo-attractant. The histogram displays the percentage of the migrated cells divided by the total number of cells initially loaded into the wells. Reduced TRIM16 protein was confirmed by western blotting using a TRIM16 specific antibody. Anti-GAPDH antibody was used as a loading control, N=3.

168

3.2.4.2 TRIM16 gene silencing reduces melanoma cell proliferation and induces EMT marker, Snail in A375 cells

As TRIM16 gene silencing in melanocytes resulted in decreased cell proliferation (Figure 3.15A), we sought to determine the result of TRIM16 gene silencing in melanoma cells. We demonstrate that using two specific siRNA’s against TRIM16 (siTRIM16-1 and siTRIM16-2) also resulted in a decrease in cell proliferation evidenced by reduced BrdU incorporation in A375 cells with siTRIM16-1 (***p<0.001) and Mel-JD cells with both siTRIM16-1 (***p<0.001) and siTRIM16- 2 (***p<0.001) compared to siRNA controls (Figure 3.16A). We observed a change in the morphology where dendrites were observed in A375 and Mel-JD cells with TRIM16 gene silencing (Figure 3.16B). As this morphology change can indicate the cells shifting from a epithelial to mesenchymal (EMT) state (152, 417), we sought to determine if changes in the EMT master regulator, Snail, and it’s downstream target, cyclin D1, were observed. A positive increase in Snail and in target cyclin-D1 was determined in the A375 cell line. This process appears to be more involved as the second cell line tested, Mel-JD, did not show an increase in EMT markers (Figure 3.17). This demonstrates that TRIM16 gene silencing may induce an EMT state in a subset of melanomas.

169

Figure 3.16A

Figure 3.16B

Figure 3.16. TRIM16 gene silencing reduces melanoma cell proliferation and results in a phenotype change. A) A375 and Mel-JD cells were transiently transfect with siRNA’s siControl, siTRIM16-1, or siTRIM16-2 for 48 hours and cell proliferation was measured using the BrdU incorporation assay. A statistically significant difference is indicated by ***p<0.0001 compared to siRNA control. B) An increase in dendrite formation is observed with TRIM16 gene silencing in A375 and Mel-JD cells.

170

Figure 3.17. TRIM16 modulates epithelial to mesenchymal marker, Snail, in A375 cells. A375 and Mel-JD cells were transiently transfected with 20 nM of siRNA Control, siTRIM16-1 or siTRIM16-2 for 48 hours. Immunoblotting of whole cell lysate was performed using anti-TRIM16, anti-Snail and anti-Cyclin D1 specific antibodies and anti-Actin was used as a loading control.

171

These data indicate that TRIM16 expression can reduce cell migration and most importantly, loss of TRIM16 by gene silencing in normal melanocytes can increase cell migration. There is evidence that TRIM16 gene silencing may induce an epithelial to mesenchymal transition, a state known to be characterized by reduced cell proliferation and an increase in cell migration. Analysis of a greater number of melanoma cell lines may clarify whether TRIM16 loss contributes to an EMT phenotype in a particular subset of melanoma cell lines. This data supports the role for TRIM16 in reducing melanoma migration and warrants further investigation into the role of TRIM16 as a suppressor of melanoma metastasis.

3.2.5 TRIM16 directly binds to the IFNβ1 promoter and induces IFNβ1 transcription

In order to determine TRIM16 mechanism of action we sought to utilise a cancer pathway specific PCR array as an unbiased approach to screen for candidate genes (418, 419) that are modulated by TRIM16 and may be required for TRIM16 to mediate its anti-proliferative and pro-apoptotic function.

In this section we sought to determine the target genes of TRIM16 that mediate its anti- proliferative and anti-migratory action. To investigate potential downstream target genes of TRIM16 in melanoma, we performed a Cancer Pathway PCR Array of 96 genes using mRNAs from G361 cells transiently transfected with empty vector or TRIM16 expression vector for 24 hours (Figure 3.18A). The most highly induced gene in the cDNA expression array studies of melanoma cells overexpressing TRIM16 was IFNβ1 (5.2-fold). Differentially expressed genes were predominantly involved in angiogenesis and invasion & metastasis (Figure 3.18A). We performed PCR validation studies on target genes that displayed >2 fold increase or decrease.

Key genes of interest were validated in G361 cells by RTq-PCR. Significant differences between empty vector and over-expression of TRIM16 in Interleukin-8 (**p<0.01), c-JUN (*p<0.05) and IFNβ1 (***p<0.001) gene expression is shown in Figure 3.18B. Putative target gene, TIMP3, was not validated by RT-qPCR and was determined to be a false positive on the PCR array. The key gene of interest, IFNβ1 was also validated in A375 cells (**p<0.01) (Figure 3.18C).

172

3.2.5.1 TRIM16 over-expression induces transcriptional changes in gene involved in angiogenesis and invasion & metastasis

Figure 3.18A

173

Figure 3.18B

Figure 3.18C

174

Figure 3.18. IFNβ1 mRNA expression is increased with TRIM16 overexpression. A) G361 cells were transiently transfected with pcDNA3.1/TRIM16/myc.His expression vector (TRIM16) or pcDNA3.1 empty vector (EV) for 24 hours. Gene expression was analysed by a Cancer PathwayFinder PCR array. B) Candidate genes, Interleukin-8, c-Jun and IFNβ1 were validated in G361 cells by RT-qPCR at 24 hours following transient TRIM16 overexpression N=2 C) Candidate gene IFNβ1 was validated in G361 and A375 cells by RT-qPCR. N=3 TIMP3 was not validated by RT-qPCR and was a false positive on the PCR array screen

175

Extensive literature has implicated TRIM family proteins as playing a role in innate immunity (286, 288, 303, 348). The relationship between immunity and cancer development is complex (420). Chronic activation of innate immune cells around pre-malignant tissues may promote tumour development (420). The interplay between the TRIM proteins, innate immunity and cancer development is not well understood. However, interferon proteins have been shown to play a significant and non-redundant role in cancer immunoediting (364). Furthermore, the 9p21 chromosomal region has been shown to be deleted in numerous cancer types (357, 362), including melanoma (357-361). In melanoma, IFN-β has been shown to induce cell death in vitro (365-367). Of all interferons, human IFN-β demonstrates the highest anti-proliferative activity against human melanoma cell lines (367). IFN-β has also been demonstrated to reduce tumour burden in an in vivo mouse xenograft models (365, 366). IFN-β1a has also been shown to be important in inhibiting lymph node metastasis in human melanoma xenografts (368). These data indicate a potential melanoma therapy that has yet to be realised. Gaining further insight into the mechanisms of IFN-β regulation is important in developing mechanisms for controlled IFN-β expression and restoration of IFN-β signalling in advanced melanoma. We selected IFNβ1 for further analysis based on the literature indicating IFNβ1 as having a role in reducing melanoma proliferation and metastasis. Due to the anti-proliferative action of IFNβ1, further investigation of IFNβ1 is warranted to determine if IFNβ1 is required for TRIM16 anti-proliferative action. Interestingly, Panne and colleagues have shown that c-JUN is required to form the enhanceosome complex and is essential for the transactivation of IFNβ1 (390). c-JUN was therefore included for further study.

3.2.6 TRIM16 binds to the IFNβ1 promoter in the enhanceosome region

As TRIM16 over-expression has been shown to increase IFNβ1 mRNA expression (Figure 3.18A, B&C), we put forward the hypothesis that TRIM16 binds to the IFNβ1 promoter and induces IFNβ1 gene transactivation. To strengthen this, we have also shown that c-JUN is a target gene of TRIM16 and c-JUN is known to be an essential component of the enhanceosome complex that is required for IFNβ1 transactivation (386, 390). To investigate whether TRIM16 directly bound the IFNβ1 promoter, we performed a chromatin immunoprecipitation (ChIP) assay, and showed that both anti-TRIM16 (p=0.0057) and anti-c-Jun (p=0.0065) antibodies efficiently

176

immunoprecipitated the region of the IFNβ1 gene core promoter carrying the enhanceosome protein binding region (Figure 3.19A, B&C). Enrichment of c-JUN at the promoter region also serves as a positive control of known protein binding (Figure 3.19C). The control for IgG protein pull down and PCR was negative indicating that TRIM16 and c-JUN protein pull-down is specific (Figure 3.19C).

177

Figure 3.19A

Figure 3.19B

178

Figure 3.19C

Figure 3.19D

179

Figure 3.19. TRIM16 directly binds the IFNβ1 gene promoter and induces IFNβ1 transcription. A). A diagrammatic representation of the IFNβ1 gene promoter indicating transcription start site (TSS) and enhanceosome binding site is shown. A chromatin immunoprecipitation (ChIP) assay is performed using anti-TRIM16 or anti-c-Jun antibodies in the immunoprecipitation with IgG antibody as a negative control. PCR is performed with IFNβ1 primers targeting the indicated regions of IFNβ1 promoter region in G361 cells. B) Agarose gel of TRIM16 amplification is shown compared to IgG control and input. C). Fold enrichment of the IFNβ1 promoter is shown for binding of TRIM16 and c-Jun antibodies and the IgG control to the IFNβ1 promoter. A control PCR 2000bp upstream of the IFNβ1 primer PCR site is used as a negative control. Promoter enrichment was calculated by dividing the PCR product from the TRIM16 and c-JUN and IgG antibody immunoprecipitation, by the PCR product from the input. D) Western blot confirming TRIM16 over-expression by transient transfection with corresponding reduced cell proliferation, N=2.

180

We have identified TRIM16 as a binding protein for the IFNβ1 promoter alongside known binding protein, c-JUN. TRIM16 binds in the same region as the enhanceosome complex and, like c-JUN, may be required for gene transactivation. This data highlights a unique mechanism of TRIM16 action and the potential for TRIM16 to mediate anti-proliferative and anti-migratory activity via IFNβ1. Binding of TRIM16 with the enhanceosome complex is proposed in the schematic on Figure 3.20.

181

Figure 3.20. TRIM16 binds to the IFNβ1 promoter in the same region as c-JUN, an integral part of the enhanceosome complex essential for IFNβ1 transcription. Schematic representation of the binding sites of c-JUN (maroon), ATF (red), IRF-3A&C (light green), IRF-7B&D (yellow), p50 (light blue) and RelA (dark blue) (binding site data from (386)). Schematic indicates ChIP forward and reverse primer sites encompassing the enhanceosome binding complex. TRIM16 is proposed to form part of the enhanceosome complex.

182

3.2.7 IFNβ1 expression is significantly decreased in primary human metastatic melanoma tissues and is required for TRIM16 mediated inhibition in cell proliferation.

Presently, there is no research on the expression levels of IFNβ1 in the progression of melanomagenesis. To further elucidate the role of IFNβ1 in melanoma progression, we performed immunohistochemistry on the same 91 human patient samples analyzed for TRIM16 expression in Figure 3.1. We found that IFNβ1 protein expression levels were lower as melanoma became more invasive (i.e. thicker) and metastatic (Figure 3.21A&B). The relationship between IFN1 expression and clinical stage closely mirrored the patterns seen for TRIM16 expression. We determined that IFNβ1 expression is decreased in stage III & IV melanoma compared to primary melanoma in a strikingly similar manner to that of TRIM16 (Figure 3.21A). Samples were analysed independently by two researchers in a blinded manner and two regions of the tumour were graded for protein expression. The loss of IFNβ1 expression is significant at the dermal invasive melanoma >1mm stage (**p<0.01) (Figure 3.21A) compared to normal compound nevus. This is the same stage that TRIM16 is significantly lost in melanomagenesis (Figure 3.1A). This highlights the potential for TRIM16 and IFNβ1 to regulate each other and gives a basis towards continued study to elucidate the relationship between the two proteins. Western blot analysis of IFNβ1 protein expression in melanoma cells compared to normal melanocytes revealed that IFNβ1 is not decreased in cell lines compared to normal melanocytes (Figure 3.22). However, as only one normal melanocyte line was tested here, it is possible that this particular line does not express high levels of IFNβ1 and testing of other melanocyte lines may show higher expression compared to the melanoma cell line panel.

183

Figure 3.21A

184

Figure 3.21B

Figure 3.21. IFNβ1 expression is significantly decreased in primary human metastatic melanoma tissue. A). The level of IFNβ1 expression in primary melanoma tissue was measured for 91 melanoma patients using immunohistochemical grading after staining with an anti-TRIM16 antibody. There were 13 samples per stage. The statistical analysis was performed by the Student’s t-test. A statistically significant difference is indicated by **p<0.01 or ***p<0.001, when expression in melanoma was compared with compound nevus. B). Representative immunohistochemical staining for IFNβ1 expression for two dysplastic compound nevus and two distant metastasis tumours is shown.

185

Figure 3.22. IFNβ1 is decreased in some melanoma cell lines compared to normal melanocyte NHEM cells, but does not correlate with TRIM16 expression. Immunoblotting for IFNβ1 protein expression was performed in proliferating normal human epidermal melanocytes (NHEM) and 10 melanoma cell lines. Whole cell lysates were prepared and immunoblots were probed with a specific anti-IFNβ1 antibody. An anti-Actin antibody served as a loading control. *p<0.05, **p<0.01

186

3.2.7.1 TRIM16 protein expression strongly correlates with IFNβ1 expression in distant metastasis melanoma

In this section, we sought to determine the correlation between TRIM16 and IFNβ1 expression at the distant metastasis stage in immunohistochemistry analysed melanoma patient samples. Nine patients were analysed and we determined that there is a strong positive correlation between TRIM16 and IFNβ1 expression at the distant metastasis stage (Figure 3.23A). We performed TRIM16 over-expression studies combined with both IFNβ1 and c-JUN gene silencing and measured cell proliferation by BrdU incorporation. This demonstrated that IFNβ1 is required in both Mel-JD (***p<0.001) and G361 (*p<0.05) cell lines for TRIM16 mediated anti-proliferative action (Figure 3.23B). Similarly, c-JUN is required in both Mel-JD (*p<0.05) and G361 (*p<0.05) cell lines for TRIM16 mediated anti-proliferative action (Figure 3.24). As c-JUN is required for IFNβ1 transactivation, these data support that a relationship exists between TRIM16, c-JUN, and IFNβ1 in reducing melanoma cell proliferation.

187

Figure 3.23A

188

Figure 3.23B

Figure 3.23. IFNβ1 expression correlates with TRIM16 expression in melanoma distant metastasis and is required for TRIM16-mediated inhibition in cell proliferation. A) A correlation analysis between IFNβ1 and TRIM16 expression at the distant melanoma metastasis stage from immunohistochemistry analysis of human melanoma patient sections (n = 9). B) Mel- JD and G361 cells were transiently transfected with either control siRNA, IFNβ1 siRNA, EV + siIFNβ1, TRIM16 expression vector + siIFNβ1, EV, or TRIM16 expression vector. Cell proliferation was measured by BrdU incorporation into DNA in three independent experiments, N=3. Comparison is between Empty Vector and TRIM16 with or without IFNβ1 siRNA. ***p<0.001, *p<0.05.

189

190

Figure 3.24. c-JUN is required for TRIM16-mediated inhibition of cell proliferation. Mel-JD and G361 cells were transiently transfected with either control siRNA, c-Jun siRNA, EV + c-Jun, TRIM16 expression vector + c-Jun, EV, or TRIM16 expression vector. Cell proliferation was measured by BrdU incorporation into DNA in three independent experiments. *p<0.05 comparison between Empty Vector and TRIM16 with or without c-JUN siRNA. .

191

3.2.7.2 Gene silencing of IFNβ1 decreases melanoma cell proliferation

To investigate the functional implications of IFNβ1 loss in melanoma, we used siRNA for gene silencing and measured cell proliferation by BrdU incorporation into the DNA. We determined that knockdown of IFNβ1 in CHL-1, SKMel-2 and IPC-298 which expressed the highest levels of IFNβ1 of all tested melanoma cell lines, resulted in a reduction in cell proliferation (Figure 3.25A), which is the same result achieved with gene silencing of TRIM16 (Figure 3.16A).

Knockdown of IFNβ1 results in a decrease in TRIM16 protein expression as determined by immunoblot with Mel-JD and G361 cells transiently transfected with siRNA to IFNβ1 (Figure 3.25B). This result suggests that there is a feedback regulation between TRIM16 and IFNβ1 whereby TRIM16 positively regulates IFNβ1 expression and IFNβ1 may be required for TRIM16 transcription or protein stability (Figure 3.25C), however further experiments are required to determine the interaction between the two proteins. An immunoprecipitation study of TRIM16 and IFNβ1 proteins may determine if they bind to each other with the hypothesis that IFNβ1 binds to and stabilises TRIM16 protein in melanoma.

192

Figure 3.25A

Figure 3.25B

193

Figure 3.25C

Figure 3.25. Knockdown of IFNβ1 reduces cell viability and TRIM16 protein expression. A) Gene silencing of IFNβ1 was mediated by a targeting pool of four siRNA’s against IFNβ1. Cell viability in CHL-1, SK-Mel-2 and IPC-298 cell lines (expressing relatively high levels of IFNβ1) was determined by Alamar blue after siRNA silencing of IFNβ1 ***p<0.001, *p<0.05 compared to siRNA control. B) Immunoblots were performed in G361 and Mel-JD cell lines using specific anti-TRIM16 and anti-IFNβ1 antibodies. C) Relationship between TRIM16 and IFNβ1 show that TRIM16 over-expression (o/e) increased IFNβ1 mRNA and gene silencing of IFNβ1 mRNA results in a decrease of TRIM16 protein.

194

Altogether, IFNβ1 gene silencing data suggests that IFNβ1 drives melanoma cell proliferation. This result is curious as it contrasts to the previous finding that IFNβ1 reduces cell proliferation (378). Gene silencing of IFNβ1 in reducing cell proliferation reflects that of TRIM16 where both the over-expression and gene silencing of TRIM16 reduces cell proliferation. It is possible that as gene silencing of IFNβ1 results in a decrease in TRIM16 protein expression that these proteins closely regulate each other. It is possible that IFNβ1 gene silencing may reduce TRIM16 protein and contribute to an EMT state. Given we have shown that IFNβ1 gene silencing results in a decrease in TRIM16 protein expression, the reduction in cell proliferation may be due to the loss of TRIM16, though further investigation is required to determine the relationship between the two proteins.

3.2.8 Exogenous recombinant human IFNβ1 protein does not increase TRIM16 protein in vitro

We tested the responsiveness of melanoma cells to exogenous recombinant human (rh)IFNβ1 in

Mel-JD and CHL-1 melanoma cells and found that the IC50 was not reached at 5000 international units (IU) (Figure 3.26A) with rhIFNβ1 treatment. A significant reduction in cell proliferation evidenced by reduced BrdU incorporation was observed at 1000 IU in Mel-JD cells (**p<0.01) and 2000IU in CHL-1 cells (***p<0.001) compared to no treatment controls. We also investigated whether rhIFNβ1 treatment could increase TRIM16 and IFNβ1 protein expression. The cell lines with low levels of both IFNβ1 and TRIM16 protein were selected for this study. We treated cells with up to 5000 IU of rhIFNβ1 and an apparent decrease in both TRIM16 and IFNβ1 protein expression was observed in Mel-JD or CHL-1 cell lines at treatment with 5000 IU (Mel-JD cells) and 4000 IU (CHL-1 cells) (Figure 3.26B). However, an increase in TRIM16 protein expression is observed at an earlier concentration of 3000 IU in Mel-JD cells (Figure 3.26B) and it remains unclear if rhIFNβ1 protein increases TRIM16 protein expression.

195

Figure 3.26A

Figure 3.26B

196

Figure 3.26. Recombinant human IFNβ1 treatment reduces melanoma cell proliferation but does not increase TRIM16 protein. Mel-JD and CHL-1 cells were incubated with recombinant human IFNβ1 (rhIFNβ1) for 72 hours and cell proliferation measured by BrdU incorporation into DNA. Statistical significance is indicated by **p<0.001, ***p<0.0001 compared to no treatment controls. Western blotting was performed after 72 hours from whole cell lysates using specific antibodies to TRIM16 and IFNβ1 and anti-Actin antibody as a loading control.

197

Results from this section suggest that rhIFNβ1 has some anti-proliferative effect on melanoma cells. However, the effectiveness of rhIFNβ1 in reducing melanoma cell proliferation plateaus at 3000 IU treatment and increased doses up to 5000 IU do not offer additional reduction in cell proliferation. We also demonstrate that treatment with rhIFNβ1 reduces TRIM16 and IFNβ1 protein expression in a dose dependant manner. As a single agent, it is unlikely that rhIFNβ1 has a marked effect on reducing melanoma proliferation, but may offer benefit in combination with additional anti-melanoma agents.

3.2.9 TRIM16 protein is increased and stabilized with vemurafenib treatment and is partially required for vemurafenib drug action

As exogenous IFNβ1 treatment did not appear to be an effective enhancer of TRIM16 protein expression, we tested the ability of current melanoma treatments as inducers of TRIM16 protein. Negative regulator of the MAPK pathway, Allyl isothiocyanate (AITC) in human colorectal adenocarcinoma induces a positive transcriptional regulation of TRIM16 (421). Vemurafenib is a heralded melanoma therapeutic that was recently approved for the treatment of BRAFV600 positive melanomas. Vemurafenib is a targeted inhibitor of the mutant BRAFV600 protein and, like AITC, blocks the activation of the MAPK pathway. Thus we sought to determine if vemurafenib treatment of melanoma cells would result in an increase of TRIM16 expression.

3.2.9.1 TRIM16 protein expression is increased in melanoma cell lines with vemurafenib in a dose dependant manner

MAPK pathway proteins, Jun N-terminus kinase (JNK), p38 and ERK have critical roles in cell migration (422). ERK regulates cell movement by phosphorylation of myosin light chain kinase (MLCK), calpain or focal adhesion kinase (FAK). ERK phosphorylation is a key mediator of cell proliferation and migration in the progression of metastatic melanoma (423). For this reason, we sought to determine whether treatment of melanoma cells with the BRAF inhibitor, vemurafenib,

198

has an effect on TRIM16 protein expression, which we have demonstrated plays a role in melanoma cell migration in this chapter.

To evaluate the possible therapeutic application of TRIM16 expression patterns on the treatment of melanoma, we examined the effect of vemurafenib on TRIM16 expression levels. After 72 hours of vemurafenib treatment at doses of 0-0.5 µM, TRIM16 protein expression increased in a dose- dependent manner in A375, G361 and Mel-CV melanoma cells (Figure 3.27). To understand the mechanism of action we performed the cycloheximide chase assay to inhibit protein synthesis and determine if vemurafenib increases TRIM16 protein half-life. We found that TRIM16 half-life increased from 12 to more than 24 hours in A375 cells in the presence of vemurafenib, demonstrating increased protein stability (Figure 3.28A). Similarly, the half-life of TRIM16 in Mel-CV cell increased from 6 to more than 24 hours in the presence of vemurafenib (Figure 3.28A). To determine if TRIM16 is required for vemurafenib mechanism of action, we used specific siRNA’s against TRIM16 (TRIM16-1 & TRIM16-2) and control siRNA to target TRIM16 in the presence of 0.5 µM vemurafenib in A375 cells. Cell viability was increased in the TRIM16 gene knockdown samples compared to control with both siTRIM16-1 (***p<0.001) and siTRIM16-2 (***p<0.001) in comparison to siRNA control, indicating TRIM16 is partially required for vemurafenib mechanism of action (Figure 3.28B).

199

Figure 3.27. Vemurafenib treatment increases TRIM16 protein in melanoma cells in vitro Melanoma cells, A375, G361 and Mel-CV were treated with vemurafenib at 0, 0.125, 0.25 or 0.5 µM for 72 hours and whole cell lysates were used for immunoblotting against an anti-TRIM16 antibody. An anti-β-Actin or Anti-GAPDH antibody was used as a loading control.Corresponding densitometry plots of TRIM16 expression as percentage of loading control. *p<0.05, **p<0.01, ***p<0.001.

200

Figure 3.28A.

201

Figure 3.28B.

Figure 3.28. TRIM16 protein stability is increased by vemurafenib and is required for vemurafenib-induced loss of melanoma cell viability. A). TRIM16 protein stability was assessed in melanoma (A375 and Mel-CV) cells following treatment with vemurafenib at 0.5 and 1.5 μM respectively, and CHX at 100 µg/mL. At the specified time points, the cells were harvested and the protein was extracted for analysis by Western blotting against an anti-TRIM16 or anti-Actin antibodies. B). Melanoma (A375) cells were transiently transfected with two different TRIM16 siRNA’s or control siRNA, and then treated with 0.5 µM vemurafenib (VEM) for 72 hours. Cell viability was measured using the Alamar blue assay. A statistically significant difference is indicated by ***p<0.001.

These data demonstrate that TRIM16 was partially required for vemurafenib drug mechanism of action.

202

3.10 TRIM16 is increased with vemurafenib treatment in human patient samples

To understand the clinical relevance of the increase and stabilization of TRIM16 protein in melanoma cells with vemurafenib treatment, we assessed clinical samples of melanoma patients receiving either vemurafenib or dabrafenib BRAF inhibitor therapy. Analysis of samples from patients on BRAF inhibitor treatment compared to matched pre-treatment samples shows a statistically significant increase in TRIM16 protein expression (Figure 3.29), indicating the potential for a significant role of TRIM16 in mediating vemurafenib drug action in vivo. We analysed 10 melanoma patients pre-treatment and while receiving BRAF inhibitor treatment. For 5 patients we also had access to samples after relapse on BRAF inhibitor. Our immunohistochemistry data demonstrate a significant increase of protein expression while receiving BRAF inhibitor treatment TRIM16 (**p=0.0018) compared to pre-treatment matched patient samples (Figure 3.29A&B). TRIM16 expression also decreased in patients that had relapsed while on BRAF inhibitor treatment, though this did not reach statistical significance (p=0.36). This data confirms an in vivo relevance of the relationship between TRIM16 and BRAF inhibitor treatment. It is yet to be determined if TRIM16 is required to mediate BRAF inhibitor activity in vivo.

203

Figure 3.29A.

Figure 3.29B.

204

Figure 3.29. TRIM16 protein expression is increased with vemurafenib treatment in patient clinical samples. A). 5 matched pre and treated patient samples and 5 matched pre, treated and progressed patient samples on BRAF inhibitor therapy were obtained. Immunohistochemistry using TRIM16 specific antibody was performed and 4 fields (averaging 300 cells) were independently and blindly analysed by two researchers assigning a TRIM16 score based on an arbitrary scale. A statistically significant p value is indicated by compared to matched pre- treatment patient controls. **p<0.01 B). Representative immunohistochemistry for two patients is shown, with TRIM16 protein expression in red and nuclear staining in blue.

205

3.11 Proposed model of TRIM16 action in response to vemurafenib treatment and modulation of IFNβ1 expression.

Taken together, we put forward the following proposed a model (Figure 3.30) of the relationship between vemurafenib, TRIM16, c-JUN and IFNβ1 for the suppression of melanoma proliferation.

In the absence of BRAF inhibitor, TRIM16 protein expression is repressed and target gene c-JUN is not up-regulated. This results in a decrease in the enhanceosome complex activity and reduced transactivation of IFNβ1, resulting in the consequence of melanoma proliferation and metastasis. In the presence of BRAF inhibitor, TRIM16 protein and TRIM16 target gene, c-JUN, are increased. This results in transactivation of IFNβ1 and inhibition of melanoma proliferation and metastasis.

206

Figure 3.30. A proposed model for the role of TRIM16 in melanoma. TRIM16 expression is significantly lost during progression from localized to metastatic melanoma. BRAF inhibitor increases and stabilizes TRIM16 protein levels. In the absence of BRAF inhibitor, TRIM16 and its target gene, c-JUN, are repressed, and results in loss of IFNβ1 promoter binding and reduced enhanceosome complex activity and IFNβ1 transcription. BRAF inhibitor increases TRIM16 protein stability and consequently, IFNβ1 transcription.

207

3.3 Discussion

We have previously shown that TRIM16 reduced cell proliferation and migration of cancer cells in two malignant diseases, neuroblastoma and SCC (350, 354). In the present study, we found that TRIM16 is strongly expressed in normal melanocytes and compound nevi, but during the evolution from localized to metastatic melanoma, TRIM16 expression is markedly repressed due to TRIM16 gene promoter methylation and reduced protein stability. As TRIM16 expression was markedly repressed in metastases compared with localized primary invasive melanomas (> 1mm thick), this suggests that low TRIM16 expression level in AJCC stage II and III melanoma may predict a high subsequent risk of distant metastases. Our data indicate that high TRIM16 protein expression is a predictor of favourable patient prognosis in lymph node metastasis. There are currently few good markers which can accurately predict the risk of metastases in patients with deeply invasive, but apparently localized, melanoma. We propose that TRIM16 expression is a good candidate for further evaluation as a prognostic marker in stage II or III disease and thus identify a patient cohort who may benefit from systemic therapy aimed at increasing TRIM16 levels and preventing metastasis. Further investigation into TRIM16 as a candidate tumour suppressor in melanomagenesis is warranted. However, it does not follow that all tumour suppressor proteins are prognostic markers of favourable patient outcome and conversely, not all prognostic markers are tumour suppressor proteins, though this is frequently the case (424-427). Prognostic markers are valuable clinically as they are informative of expected survival and can indicate a treatment strategy for molecular targeted therapy (428, 429). Identification of treatment responding subgroups using biomarkers or prognostic markers is the present focus of molecular targeted therapy (429, 430).

An important future study to test the role of TRIM16 in melanomagenesis would be the development of a melanocyte-specific TRIM16 knockout mouse crossed with a conditional melanocyte specific BRAFV600E expressing mice that develop melanocytic lesions (91). This would determine if TRIM16 loss in addition to mutant BRAF expression is sufficient to induce tumorigenesis. Alternatively, using a BRAFV600E melanocyte expressing/PTEN tumour suppressor gene silenced mouse model would ensure the development of melanomas (91), and crossed with a TRIM16 melanocyte knockout mouse will allow the investigation of metastasis with TRIM16 loss.

208

TRIM16 is significantly decreased at the dermal invasive melanoma stage. This is the point where the pathology of melanoma gains metastatic and invasive potential, transitioning from the radial to the vertical growth phase and penetrating the basement epidermis (10, 36, 48). These data suggest that TRIM16 loss may be a significant event in the progression of melanoma metastasis. Researchers have shown TRIM16 has nuclear localization in differentiating ganglia cells, but is reduced in neuroblastoma tumours (402). Similarly, this body of work has shown that normal melanocytes demonstrate high nuclear TRIM16 expression which is later lost upon progression to melanoma. From immunohistochemistry studies, TRIM16 nuclear expression is low in compound and dysplastic nevi, suggesting loss of TRIM16 nuclear protein expression may be an early event in nevi proliferation as nuclear TRIM16 is only observed in normal nevi and not compound or dysplastic nevi, which have undergone cell proliferation.

In this chapter we have shown that TRIM16 and c-JUN directly bound the IFNβ1 promoter, induced IFNβ1 transcription, and consequently cell growth arrest. IFNβ1 was significantly lost in melanomagenesis in a correlative manner to TRIM16. Furthermore, either IFNβ1 or c-JUN gene silencing was sufficient to abrogate the anti-proliferative effects from TRIM16 overexpression in melanoma cells. Taken together, our data demonstrated that repression of TRIM16 and consequently IFNβ1 expression, are markers of metastatic behavior in melanoma and TRIM16 may be an important novel treatment target to prevent distant metastasis. We are the first to show that IFNβ1 expression is significantly decreased during melanomagenesis and correlates with TRIM16 loss at the distant metastasis stage. We have demonstrated that siRNA mediated silencing of IFNβ1 results in TRIM16 protein decrease and TRIM16 over-expression modulates IFNβ1 mRNA expression. These data lead towards the possibility of a feedback loop between TRIM16 and IFNβ1. IRF-3 and IRF-7 are components of the IFNβ1 enhanceosome, binding the promoter region and inducing IFNβ1 transcription. Intriguingly, IRF-3&7 are both involved in positive feedback loops with IFN-β expression in murine fibroblasts (431). We propose that TRIM16, which we have demonstrated binds to the IFNβ1 promoter in the same region as enhanceosome proteins, is likely to form part of a positive feedback loop with IFNβ1. To gain insight into this mechanism, forced over-expression of IFNβ1 is required to determine if this protein increases TRIM16 mRNA expression, or stabilises TRIM16 protein. In addition, immunoprecipitation studies would be informative as to whether TRIM16 binds to IFNβ1, or if the protein stabilization 209

is via an indirect method. Given we have shown that IFNβ1 gene silencing results in TRIM16 protein decrease, loss of IFNβ1 in melanomagenesis may be the first event to occur and corresponding TRIM16 loss may be due to this event. Here we propose an important role for TRIM16 expression in the early stages of melanomagenesis as an inhibitor of cell growth and migration through its effects on the inflammatory cytokine, IFN1. These findings have significant diagnostic and therapeutic implications.

Primary melanoma tumour ulceration is a strong indicator of poor patient outcome (42, 404). Melanoma ulceration is also a predictive marker for responsiveness to adjuvant interferon (IFN- α2a) therapy in melanoma (432). In the analysis of 87 patients for TRIM16 score and ulceration presence or absence, our data shows a trend towards ulceration and lower TRIM16 score, though this did not reach statistical significance. However, a larger patient cohort may have the statistical power to demonstrate a significant difference. We feel that this is a possibility as TRIM family members are heavily implicated in immune regulation and we hypothesize that TRIM16 loss may result in increased aberrant immune signalling and tumour ulceration (307, 320, 344). In addition, interferon treatment could modulate or stabilize TRIM16 protein expression and thereby promote a favourable mechanism of reduced cell proliferation and reduced cell migration.

Previous studies have indicated that the MAPK pathway plays a key role in melanoma cell migration (422). ERK phosphorylation is a key mediator of cell proliferation and migration in metastatic melanoma. Interestingly, we found that treatment of melanoma cells with the BRAF inhibitor, vemurafenib, resulted in increased levels of TRIM16 protein in vitro and in vivo. As TRIM16 is stabilized by vemurafenib at the protein level, it is possible that vemurafenib treatment may alter TRIM16 phosphorylation states or induce other post-translational modifications of TRIM16 which affect TRIM16 protein stability. This mechanism of TRIM16 regulation has been shown previously after retinoic acid treatment, where TRIM16 undergoes serine threonine phosphorylation and enhanced protein stability (355). The cytopathic effects of the successful BRAF inhibitor, vemurafenib, in part required induction of TRIM16 expression. This pattern was mirrored in melanoma tissue samples from patients treated with BRAF inhibitors, vemurafenib and dabrafenib, indicating that de-repression of TRIM16 expression in melanoma cells is a novel therapeutic approach and provides new insights into the mechanism of action of BRAF inhibitors. Our data implicate TRIM16 as an effector protein of vemurafenib. Furthermore, the increase in

210

TRIM16 in the melanoma of patients receiving vemurafenib therapy, and the subsequent decrease in TRIM16 protein in relapsed samples, further strengthen the role of TRIM16 in mediating the anti-proliferative action of vemurafenib. This data clearly highlights a clinical relationship between vemurafenib treatment and TRIM16 protein expression in melanoma patients.

Our data has for the first time demonstrated a novel action of vemurafenib in suppressing melanoma cell growth in a TRIM16-dependent manner. We propose that TRIM16 suppresses melanoma cell growth via the up-regulation of c-JUN and the formation of the enhanceosome promoter complex that is subsequently required for IFNβ1 transactivation (Figure 3.30). As exogenous expression of IFNβ1 reduces cell proliferation and induces apoptosis in melanoma cells (373), restoration of IFNβ1 signaling by TRIM16 may enhance the anticancer activities of IFN1. Furthermore, combination treatment of recombinant human IFNβ and vemurafenib may reinforce each other and promote a co-operative anticancer signal.

3.4 Conclusion

Low TRIM16 expression was associated with metastasis in melanoma patients, and Stage III patients with low expression of TRIM16 were associated with poor prognosis. Stage III patients with low TRIM16 expression may represent a subgroup of patients that may benefit from BRAF inhibitor therapy.

211

Chapter 4

Modelling TRIM16 as a novel tumour suppressor protein in melanoma in vivo

212

In this chapter, we seek to test the hypothesis that TRIM16 is a novel tumour suppressor protein in melanoma by chemically inducing skin lesions and cancers and determining tumour number and latency. We propose that loss of TRIM16 will result in reduced tumour latency and a greater tumour size and number.

4.1 Introduction

In Chapter 3 we have shown that TRIM16 protein is significantly decreased in human primary and metastatic melanoma compared to dysplastic compound nevus, and that this loss corresponds to an increased migratory phenotype. We have demonstrated that TRIM16 is a prognostic factor in a cohort of human lymph node metastasis patients, with high TRIM16 indicating favourable patient outcome. We have also shown that TRIM16 gene silencing in normal human melanocytes results in increased cell migration. The purpose of this chapter was to determine the in vivo significance of TRIM16 as a candidate tumour suppressor in melanoma development by mouse modelling. Generally, for a protein to be termed ‘tumour suppressor’ this requires in vivo modelling typically using a genetically engineered mouse (GEM) model (275, 433, 434). Mouse modelling gives insight into molecular mechanisms of cancer pathogenesis (433). GEM models are used to determine loss of function (knockout, knockdown or dominant-negative) or gain of function (knock-in, transgenic) (433). There is also scope to cross GEM mice with other tumour suppressor knockout mice and determine tumour development if a single gene knockout is insufficient to develop tumours alone, or with carcinogen challenge (91, 125). One mouse model system that established CDK4 as a melanoma oncogene used a knock-in of CDK4, which alone was not sufficient to induce tumorigenesis, but required carcinogen challenge before the phenotype was revealed (125). Mice that harbour a conditional melanocyte-specific expression of oncoprotein BRAFV600E develop melanocytic hyperplasia that is benign due to the induction of senescence (91, 118). Combination of BRAFV600E with PTEN tumour suppressor gene silencing resulted in the development of melanomas with 100% penetrance (91).

We have produced two mouse systems to understand the role of TRIM16 in cancer development. The first is a tissue specific mouse model in which TRIM16 is knocked out of the keratinocytes

213

only, and is termed the skin-specific TRIM16 knockout mouse for study the role of TRIM16 in squamous cell carcinoma (SCC). The second is a full-tissue TRIM16 knockout mouse. The reason for the development of skin specific mice model, firstly, was to against the possibility of a full- tissue knock out mouse being embryonic lethal and secondly, allows a greater depth of insight into the role of TRIM16 in skin tumour biology. TRIM16 has been reported to play a role in keratinocyte differentiation (305). TRIM16 was secreted by keratinocytes in a caspase-1 dependent manner, and this process is enhanced by interleukin-1β (IL-1β). Furthermore, TRIM16 has been reported to be secreted in response to UV exposure (304), suggesting TRIM16 may mediate stromal activity on keratinocytes and melanocytes in the skin. TRIM16 was highly expressed in epidermal growth factor-responsive basal keratinocytes and down-regulation of the TRIM16 protein was observed in the hyper-thickened epithelium of skin wounds. TRIM16 has been shown to increase differentiation markers keratins 6, 10 and involucrin in keratinocytes (305). Intriguingly, TRIM16 expression was significantly reduced in vivo during the histological progression of SCC from normal skin, and inhibited cell migration in vitro in SCC (354). In addition to the role in keratinocyte biology, TRIM16 has demonstrated some of the features of a tumour suppressor protein in neuroblastoma through down-regulation of protein binding partners: cytoplasmic vimentin and nuclear E2F1 (350). Taken together, these data suggest that TRIM16 acts to repress cancer cell replication and migration. However, the role of TRIM16 in melanoma is presently undefined. In Chapter 3 we have determined that TRIM16 protein is significantly decreased as melanomagenesis progresses. Furthermore, we have found that high TRIM16 expression is prognostic of favourable patient outcome in a cohort of lymph node metastasis patients. We have demonstrated that gene silencing of TRIM16 in melanocytes results in increased cell migration. These results support the hypothesis that TRIM16 may be important in the progression of melanoma and melanoma cell migration. A definitive way to test this hypothesis is by inducing tumours in the TRIM16 skin knockout mice and to assess the tumour latency, size and numbers to determine if TRIM16 is a driver of tumour development in vivo.

In our skin-specific knock-out model we induced mutation and inflammation in the skin by chemical carcinogen application and assessed development of papilloma, squamous cell carcinomas and melanocytic lesions. However, in this system, melanocytes retain TRIM16 expression as the promoter of Keratin14, is not expressed in melanocytes (394, 395). This allows assessment of the specific loss of TRIM16 in keratinocytes and the development of SCC in the

214

skin. In the full-tissue knockout model, TRIM16 is lost from all tissues including the keratinocyte and melanocytes of the skin. This model allows the assessment of the loss of TRIM16 simultaneously in keratinocytes and melanocytes to determine the contribution of the TRIM16 protein in the development of melanocytic lesions and melanoma.

4.2 Results

4.2.1 Characterization of TRIM16 skin-specific knockout mice

TRIM16 skin-specific knockout mice are viable and expected mendelian birth ratios are observed. The genotyping is performed as described in Material and Methods (Chapter 2, Section 2.2) for the skin-specific knockout and full-tissue knockout mice. Analysis of mRNA and protein data by PCR and western blot confirms the successful knockout of the TRIM16 gene (Figure 4.1A&B). Characterization of general features, such as food consumption, body and tail length was performed to determine differences between genotypes. Food consumption was measured by specific genotypes and the average amount of food consumed divided between the numbers of mice in the cage. These were based on individual litters. The wild-type and heterozygous mice were assessed for males (Figure 4.2, left), and wild-type, heterozygous and homozygous mice were assessed for females (Figure 4.2, right).

215

Figure 4.1A

Figure 4.1B.

Figure 4.1 Confirmation of TRIM16 knockout mice. A). Targeted disruption of the TRIM16 gene, is confirmed by PCR for TRIM16 genomic DNA and mRNA, and immunoblot for mouse

TRIM16 protein of keratinocyte primary culture from TRIM16 wild-type (+/+), TRIM16 heterozygous (+/-) and TRIM16 homozygous (-/-) mice. B). Targeted disruption of the TRIM16 gene, is confirmed by PCR for TRIM16 genomic DNA and mRNA in full-tissue TRIM16 knockout mice. 216

In additional we analysed food consumption in the skin-specific TRIM16 knockout mice. This was performed due to the finding that skin-specific deletion of stearoyl-CoA desaturase-1 alters skin lipid composition and protects mice from obesity from a high fat diet due to increased energy expenditure (435). This establishes a relationship between a skin-specific deletion of a gene and relative mouse weight. In our study, we examined males (n=8) and females (n=9) for food consumption per mouse based on genotype and averaging food consumption between mice in the cage weekly over a period of 6 weeks. Statistical analysis was performed using two-way ANOVA. We found a slight increase in food consumption in males in comparison between TRIM16+/+ wild- type and TRIM16+/- heterozygous mice observed at week 14 (*p<0.01) (Figure 4.2, left). No statistically significant difference in food consumption between all three genotypes in females (Figure 4.2, right).

217

Figure 4.2. TRIM16 skin-specific knockout mice do not have difference in food consumption. Male (n=8, Left) and female (n=9, Right) mice TRIM16+/+ wild-type, TRIM16+/- heterozygous and TRIM16-/- homozygous mice were monitored for food consumption/mouse for 5 weeks. Two-way ANOVA was performed to determine statistical significance.

218

4.2.1.1 Analysis of hair morphology in TRIM16 skin-specific knockout mice

As Keratin 14 is expressed by keratinocytes, tissues in addition to the keratinocytes of the skin may be affected by TRIM16 gene knockout. Keratinocytes are components of the tongue, cornea, hair, oocytes, and skin (393-395, 436). Therefore, Cre recombinase deletion under the control of a Keratin 14 promoter will result in the knockout of TRIM16 in all Keratin 14 expressing tissues. A study showed that skin-specific expression of Dkk4 under the control of a Keratin 14 promoter resulted in mice that had severely malformed secondary hair structure (436). Similarly, disruption of SMAD4 via a keratinocyte Cre/LoxP system resulted in hair follicle differentiation and increased cycling leading to hair loss in knockout mice (437). Therefore, if TRIM16 were required for hair formation or structure, knockout of the TRIM16 gene under a keratin 14 promoter may show a change in hair morphology. We investigated the morphology of the hair for changes with TRIM16 deletion to determine if TRIM16 plays a role in the differentiation of the hair shaft. The structure of the predominant hair type (Awl) is shown in Figure 4.3. Nine hair samples were analysed giving three samples per genotype. There was no difference in the hair structure between genotypes (Figure 4.3). In addition, primary keratinocytes were cultured from the dorsal skin of 1 day old pups (three per genotype) and cultured to determine cell morphology between genotypes. No difference in cell morphology was observed between genotypes (Figure 4.4).

219

Figure 4.3. TRIM16 skin-specific knockout mice do not have different morphology in the Awl hair. Three hair samples from TRIM16 wild-type(+/+), TRIM16 heterozygous(+/-), and TRIM16 homozygous(-/-) mice were plucked post-euthanasia and examined under the microscope to determine macroscopic differences in hair morphology.

220

Figure 4.4. Primary keratinocytes from TRIM16 knockout mice do not have different cell morphology. Primary keratinocytes we cultured from 1 day old pups. Dorsal skin tissue was isolated and trypsinized overnight before culture in keratinocyte culture media containing dispase. Genotypes were pre-determined by tail tip PCR.

221

In this section, we determined a modest increase in food consumption in male TRIM16 heterozygous(+/-) compared to TRIM16 wild-type(+/+) skin-specific mice, and no change in food consumption in corresponding female mice. No significant differences were observed in hair morphology or keratinocyte morphology.

4.2.2 Characterization of TRIM16 full-tissue knockout mice

Blinded histopathology was performed by the Institute of Medical and Veterinary Science (IMVS) of South Australia, on TRIM16 wild-type(+/+) and two TRIM16 homozygous(-/-) full-tissue knockout mice for all major tissues. No significant histological findings were found. TRIM16 full tissue knockout mice are viable and do not deviate from mendelian ratios of offspring.

Analysis of food consumption for full-tissue knockout mice was conducted in the same manner as described for the skin-specific knockout mice. All three genotypes were analysed in littermates with males (n=12) and females (n=10). No statistically significant difference in food consumption was observed between genotypes for males (left). Female wild-type(+/+) mice had a higher food consumption compared to heterozygous(+/-) or homozygous(-/-) female mice (Figure 4.5, right). Analysis of body weight and length was measured over a period of 60 days at the indicated time points. For males (n=12), the TRIM16 heterozygous(+/-) and TRIM16 homozygous(+/+) mice were analysed and for females all genotypes were analysed. (Males, n=12; Females, n=10) did not show a significant difference between genotypes for the food consumption (Figure 4.6).

222

Figure 4.5. TRIM16 full-tissue knockout mice do not have difference in food consumption. Male (n=12) and female (n=10) mice TRIM16+/+ wild-type, TRIM16+/- heterozygous and TRIM16-/- homozygous mice were monitored for food consumption/mouse for 4 weeks. Food consumed per cage was divided by the number of mice to determine average food consumption based on genotype. Two-way ANOVA was performed to determine statistical significance.

223

Figure 4.6. TRIM16 full-tissue knockout mice do not show differences in body weight or length. TRIM16 skin-specific mice male (n=12, Left) and female (n=10, Right) TRIM16+/+ wild- type, TRIM16+/- heterozygous and TRIM16-/- homozygous were monitored for weight gain (g) from birth to 60 days. Two-way ANOVA was performed to determine statistical significance.

224

These data indicate that neither TRIM16 skin-specific or full-tissue knockout mice demonstrate an obvious phenotype in tissue morphology, food comsumption, body or tail length. This demonstrates that TRIM16 skin-specific or full-tissue mice showed no discernable phenotype, but may require challenge to exhibit a phenotype.

4.2.3 Skin carcinogenesis study on skin-specific TRIM16 knockout mice

It is possible that the lack of phenotype in knockout mice may be a result of a gene compensation effect, in which a gene is up-regulated in response to the absence of the knocked out gene (438). In addition, the effect of the knockout of a gene may only become evident with the inactivation of another gene, demonstrating biological robustness (439, 440). In addition, external challenge, in this case chemical carcinogen, may be required to reveal the phenotype with gene knockout. Mice may experience protection from disease despite exhibiting no phenotype, such as the case of Atrigin1 and MuRF1 knockout mice that display no dicernable phenotype, but experience protection from denervation-induced skeletal muscle atrophy (441).

As the TRIM16 keratinocyte knockout mice displayed no discernable phenotype, a chemical carcinogen challenge was applied to determine the potential role of TRIM16 in papilloma and squamous cell carcinoma development. This is based on previous research performed by Cheung and colleagues, which determined that TRIM16 is significantly down-regulated in SCC development in the analysis of human patient samples, TRIM16 overexpression reduced SCC cell migration (354). Chemically induced skin tumours using the 7,12-Dimethylbenz(a)anthracene (DMBA)/12-O-tetradecanoylphorbol-13-acetate (TPA) system that is well established and has been used in many studies to induce both squamous cell carcinoma (442) and melanoma (443). The genetic target of DMBA is Hras1 and Kras which are mutated by the agent. TPA targets and amplifies the expression of protein kinase C (442). In a comprehensive study, skin carcinogenesis using DMBA/TPA was performed in at least 18 mouse stocks and strains and the development of papilloma and squamous cell carcinomas were determined. All mice strains tested were susceptible to tumour development. The C57BL/6 strain were among the more resistant mice but at higher doses of DMBA/TPA did develop skin lesions, albeit at a lower rate than more sensitive strains (442). At a concentration of 97.5 nmol DMBA and 6.5 nmol TPA, it was found that C57BL/6 mice

225

developed papilloma’s at an incidence of 42% with an average of 2.2 papilloma/mouse (444). C57BL/6 mice were knocked-out for DNA topoisomerase II and induced melanocytic lesion development at a similar concentration (443). To determine the incidence of papilloma, squamous cell carcinoma and melanoma in TRIM16 knock-out mice, we employed the two-stage skin carcinogenesis model in both keratinocyte and full-tissue knockout mice (Figure 4.7).

226

Figure 4.7. Two mouse models are used to understand the role of TRIM16 in skin carcinogenesis. Model A is a skin-specific knock-out targeting the keratinocytes of the skin. Model B is a full tissue knock-out in which keratinocytes and melanocytes are knocked out for TRIM16 expression.

227

4.2.3.1 Two-stage skin carcinogenesis on TRIM16 conditional skin-specific knock-out mice

The development of papilloma and squamous cell carcinoma was investigated in the skin-specific TRIM knockout mice. This involved the treatment of shaved mouse skin with a sub-carcinogenic dose of DMBA of 97.5 nmol in 200 µL (Chapter 2, Section 2.2.2). DMBA targets Hras in the keratinocytes for mutation (442). After a two week rest period, twice weekly doses of TPA at 6.8 nmol in 200 µL are applied to the skin, which targets protein kinase c to promote inflammation and increased cell proliferation (442). This facilitates keratinocyte proliferation, papilloma promotion and the progression phase, resulting in a percentage of papilloma converted to squamous cell carcinoma (Figure 4.8). Representative images of normal skin, macroscopic papilloma and squamous cell carcinoma are shown (Figure 4.9). The upper panel shows H&E sections of the skin and were used for cytological diagnosis of normal skin, papilloma, or squamous cell carcinoma (Figure 4.9). The treatment duration for the skin-specific TRIM16 mice is 21 weeks from the DMBA initiation (Figure 4.8) (Chapter 2, Section 2.2.2).

228

Figure 4.8. Schematic of the progression of squamous cell carcinoma with two-stage skin carcinogenesis. The skin of 7 week old mice is shaved and treated with a single sub-carcinogenic dose of DMBA. Mice are given a two week rest period and twice weekly doses of TPA are applied. Mice are treated for a 21 week period (skin-specific TRIM16 knockout mice), or 28 week period (full-tissue TRIM16 knockout mice).

229

Figure 4.9. Representative examples of normal mouse skin, papilloma development and squamous cell carcinoma development. Lower panel, Normal skin reveals no skin lesions. Papilloma show a proliferation of the keratinocytes of the skin, and squamous cell carcinomas exhibit the characteristic circular raised structure. Upper panel, H&E sections showing the pathology of normal skin, papilloma, and squamous cell carcinoma.

230

4.2.3.1.1 TRIM16 heterozygous skin-specific knockout mice have reduced latency for papilloma development

From topical application of skin carcinogens, we counted the numbers and weeks of the development in macroscopic papilloma on the skin of all mice. The three genotype are represented as TRIM16 wild-type(+/+) (n=22), TRIM16 heterozygous(+/-) (n=30), and TRIM16 homozygous (-/-) (n=8). We showed that skin-specific TRIM16 heterozygous(+/-) mice had reduced papilloma latency at weeks 13 and 15 (p=0.031 both weeks) compared to wild-type mice (Figure 4.10). This was not significant at 18 and 21 weeks. No significant difference was observed in comparing TRIM16 homozygous(+/+) and TRIM16 wild-type(+/+) mice, or TRIM16 homozygous(+/+) and TRIM16 heterozygous(+/-) mice. This may be due to the low homozygous mouse numbers (n=8).

231

Figure 4.10. TRIM16 heterozygous skin knockout mice have reduced latency in papilloma development. TRIM16 knockout mice were scored for the number of papilloma from week 13 of treatment to week 21. Statistical analysis was performed by the Student’s t- test between TRIM16 wild-type(+/+) and TRIM16 heterozygous(+/-) mice and TRIM16 heterozygous(+/-) and TRIM16 homozygous(+/+) mice. Comparison between TRIM16 wild-type(+/+) and TRIM16 homozygous(-/-) mice was not significant.

232

4.2.3.1.2 TRIM16 heterozygous keratinocyte knockout mice have increased squamous cell carcinoma development

Numerous proteins have been identified as tumour suppressors in squamous cell carcinoma. Notably, a keratinocyte Cre/loxP knockout model of SMAD4 demonstrated spontaneous SCC tumours in the skin at between 3 and 13 month. These tumours were also accompanied with inactivation of PTEN and activation of Akt, due to the function of SMAD4 in repressing skin carcinogenesis through the TGFβ/BMP pathway (437). In an RNA interference (RNAi) screen using shRNA deliver in utero to identify driver mutations in SCC, nonmuscle myosin IIa heavy chain (Myh9) resulted in increased SCC formation (445). Loss of Myh9 resulted in SCC development when accompanied by TGF-β or Hras mutation (445). In this chapter we hypothesize that the loss of TRIM16 accompanied by Hras mutation induced by DMBA will result in increased SCC development.

In addition to the reduced papilloma latency observed in heterozygous skin-specific TRIM16 mice we observed the development of two squamous cell carcinomas converted from papilloma, observed exclusively in the heterozygous mice in this study. Figure 4.11 demonstrates the progression from papilloma (week 12) to potential squamous cell carcinoma (week 18) and characteristic SCC by week 22 of treatment. Euthanasia and resection of the SCC revealed a clear blood supply to the tumour (Figure 4.12) and this was observed for both SCC tumours that developed in this study. The number of SCC tumours based on mouse genotype is given in Table 4.1, where 0/22, 2/30 and 0/8 SCC tumours developed in TRIM16+/+ wild-type, TRIM16+/- heterozygous, and TRIM16-/- homozygous mice, respectively. The papilloma conversion based on genotype was 0/14, 2/21 (9%) and 0/1 in TRIM16+/+ wild-type, TRIM16+/- heterozygous, and TRIM16-/- homozygous mice, respectively. This study reveals that although skin-specific TRIM16 heterozygous mice developed SCC tumours, they also developed a larger number of papilloma able to undergo conversion to SCC. The lack of SCC development in the homozygous mice may be a factor of the small cohort size (n=8) and a larger cohort may reveal a propensity towards SCC development with complete loss of TRIM16 expression.

233

Figure 4.11. Squamous cell carcinoma were developed in the TRIM16 heterozygous mice with two-stage skin carcinogenesis treatment. Squamous cell carcinoma develops from papilloma of the skin (week 12) and shows the typical raised, round, crusted lesion as the papilloma converts to squamous cell carcinoma.

234

Figure 4.12. Evidence of blood supply to the squamous cell carcinoma tumour. In addition to histology (Figure 4.9), confirmation of SCC development is evidenced by a clear blood supply to the tumour.

Table 4.1. Summary of development of squamous cell carcinoma at 21 weeks in skin-specific knockout mice

235

These data demonstrate the successful induction of papilloma and SCC in mouse skin with the application of skin carcinogens. Though the rate of papilloma conversion to SCC is low, this data indicates the potential for TRIM16 loss observed in TRIM16 heterozygous(+/-) mice to reduce the latency of papilloma development and increase the likelihood of SCC development. Given the low rate of SCC development, a larger cohort is required to determine if TRIM16 loss contributes to SCC development in vivo. In particular, it is unclear if TRIM16 homozygous(-/-) mice exhibit an increased susceptibility to SCC development.

4.2.3.4 Development of melanocytic lesions with two-stage carcinogen treatment

Melanocytic hyperplasia with the histological appearance of nevi, have been observed in the mouse skin with BRAFV600E amplification targeted to melanocytes. This resulted in benign lesions which progressed to melanomas dependant on BRAF expression levels and CDKN2A loss (446). This model recapitulates the development of nevi in humans, which often harbour the BRAFV600 mutation, but undergo oncogene induced senescence (70, 118). The occurrence of melanocytic lesions has also been in a topoisomerase IIβ (TopIIβ) skin-specific model with the induction of melanoma development by etoposide (VP-16) in the skin of DMBA treated mice (443). In this model, knockout mice for TopIIβ were protected from melanocytic lesions development compared to TopIIβ expressing mice, which developed numerous melanocytic lesions that were positive for Melan-A (443). In this section, we characterize the development of melanocytic lesions in skin- specific TRIM16 knockout mice in size, incidence and latency, and potential conversion to melanoma evidenced by metastasis to regional lymph nodes.

Interestingly, in addition to the development of papilloma and squamous cell carcinoma, melanocytic lesions of the skin develop with carcinogen treatment at 13 weeks of TPA treatment. Melanocytes of the mouse skin reside in the base of the hair follicle (442). The progression of melanoma growth in the skin follows a similar pattern to the development of SCC, involving a promotion and progression phase (Figure 4.13). The melanocytes in mouse skin are clustered around the hair follicle, which is different from human melanocytes, which reside on the dermal- epidermal junction (36). Figure 4.14 (lower panel) shows representative examples of normal skin,

236

melanocytic lesions and potential melanomas from a mouse later found to have pigmented lymph nodes. The upper panel shows the corresponding H&E stain of a cross section of skin (Figure 4.14). The melanocytic lesion is the dark cluster of pigmented cells.

237

Figure 4.13. Schematic of melanocytic lesion development with two-stage skin carcinogen treatment. 7-9 week old mice are shaved on the dorsal flank and a sub-carcinogenic dose of DMBA is applied to the skin the following day. After a two week rest period mice are treated with twice weekly doses of TPA, applied to the skin for an extended period to allow promotion and progression of tumour development.

238

Figure 4.14. Model of the development of murine melanoma by two-stage skin carcinogenesis. Representative images are given of normal skin, melanocytic proliferation and candidate melanoma development where melanocytic cells are breaking away from the pigmented cell mass.

239

4.2.3.4.1 Skin-specific TRIM16 heterozygous mice have reduced latency of melanocytic lesion development

We measured the development of melanocytic tumours from weeks 13 to 21 in the skin-specific TRIM16 knockout mice with carcinogen treatment. Interestingly, TRIM16 heterozygous skin- specific knockout mice developed melanocytic lesions with reduced latency compared to wild-type mice at 13 weeks (p=0.035) and 15 weeks (p=0.011) (Figure 4.15). This was not significant in the later 18 and 21 weeks treatment (Figure 4.15). No significant difference is observed between wild- type and homozygous mice. Again, this may be due to the low numbers (n=8) of homozygous mice that may negate a significant finding.

240

Figure 4.15. TRIM16 heterozygous skin knockout mice have reduced latency in melanocytic lesion development. Pigmented lesions were counted at weeks, 13, 15, 18 and 21 post skin carcinogen treatment. Statistical analysis was performed between TRIM16 wild-type(+/+) and TRIM16 heterozygous(+/-) mice and TRIM16 homozygous(-/-) and TRIM16 heterozygous(+/-) mice. Comparison between wild-type and homozygous mice was not significant. Statistical analysis was performed using the Student’s t-test.

241

4.2.3.4.2 TRIM16 homozygous knockout mice develop larger melanocytic lesions that heterozygous or wild-type mice

Development of larger tumours, or tumours with reduced latency, in knockout mice for a given gene are in vivo evidence of the tumour suppressor function of the gene (447). Our initial study hypothesized that TRIM16 may be a tumour suppressor in squamous cell carcinoma and used the keratinocyte model to determine the difference in papilloma and SCC development with knockout of TRIM16 in keratinocytes on the skin. In addition to papilloma development we observed the advent of melanocytic lesions in the skin. Surprisingly, when we measured the size of the lesions between the genotypes, we found that homozygous TRIM16 knockout mice developed larger lesions (>1 mm) compared to heterozygous or wild-type mice (Figure 4.16 A, B&C). This result was striking as the melanocytes in the skin of the mice harbour wild-type TRIM16 and the TRIM16 gene is not deleted in melanocytes in these mice due to the lack of Keratin 14 expression (448). This led us to investigate the relationship between keratinocytes and melanocytes and the cross-talk that occurs between these two cells within the skin involving TRIM16 expression. TRIM16 is secreted by keratinocytes (304) and may influence the adjacent melanocytes in the skin. Therefore, we hypothesize that the loss of TRIM16 in keratinocytes may result in the inability to suppress the proliferation of melanocytic lesion development.

Measurement of the size (mm) of the papilloma to develop in the skin of knockout mice reveals that TRIM16 homozygous(-/-) mice develop smaller papilloma (<2 mm) whereas TRIM16 wild- type(+/+) and TRIM16 heterozygous(+/-) mice develop a similar proportion of small and larger papilloma (Figure 4.17).

242

Figure 4.16A

Figure 4.16B

243

Figure 4.16C

Figure 4.16. TRIM16 homozygous mice develop larger melanocytic lesions than wild-type and heterozygous mice. A) Melanocytic lesions were measured (mm) in TRIM16 wild-type(+/+) (N=22), TRIM16 heterozygous(+/-) (N=33) and TRIM16 homozygous(+/+) mice (N=8) at 21 weeks skin carcinogen treatment. Representative images are shown. B). The number of melanocytic lesions per mouse are measures and represented proportionally based on size (<1 or >1 mm). C). The number of melanocytic lesions are represented as a percentage for each genotype based on size.

244

Figure 4.17. TRIM16 heterozygous skin-specific mice display larger papilloma with carcinogen challenge. The size (mm) of papilloma in the skin were measured at 21 weeks of treatment in groups <2 mm, 2-5 mm and >5 mm between genotypes. TRIM16 wild- type(+/+) mice (N=22, blue), TRIM16 heterozygous(+/-) mice (N=33, red) and TRIM16 homozygous(-/-) mice (N=8, grey).

245

4.2.3.7 Genotype has no influence on inguinal lymph node size

Enlarged regional lymph nodes can suggest the presence of melanoma within the node (449). To understand the potential relationship for regional lymph node inflammation & metastasis with genotype, we resected the inguinal lymph nodes from 7 wild-type and 6 heterozygous mice that displayed melanocytic lesions in the skin (Figure 4.18A). We determined no significant difference in the size of the lymph nodes between the genotypes despite the presence of pigmentation in three heterozygous mice (Figure 4.18B). This indicates that although there is pigmentation present, the lymph nodes are not enlarged overall.

246

Figure 4.18A

Figure 4.18B

247

Figure 4.18. No significant difference is found in lymph node size between wild-type and heterozygous TRIM16 mice. A) At 21 weeks skin carcinogen treatment, 7 wild-type mice and six heterozygous TRIM16 skin-specific mice were euthanized and the right inguinal lymph node resected and measured in mm. B) The average lymph node size was determined and statistical significance measured by the Student’s t-test.

248

4.2.3.7.1 Melanocytic lesions suggest metastatic potential by pigmentation of regional lymph nodes in TRIM16 heterozygous mice

Melanoma is a highly migratory type of skin cancer and metastasis is the major cause of patient death (44). Identification of the genes that are involved in metastasis is important as it can serve as therapeutic targets to reduce the metastatic potential of disease spread.

On taking a cross-section of a melanocytic lesion in the mouse skin, we have identified a pigmented mass, with cells that have migrated away from the main mass and may indicate disease metastasis (Figure 4.19i). The presence of melanocytic tissue was confirmed by positive Melan-A staining (Figure 4.19ii), which is observed in multiple lesions (Figure 4.19iii). Encouragingly, the pigmentation of the inguinal lymph nodes was observed in 3/30 TRIM16 heterozygous mice, a representative example is shown (Figure 4.19iv) and resected (Figure 4.19v). This further supports the potential for TRIM16 heterozygous mice to have developed metastatic melanoma, though further investigation of the presence of Melan-A in the lymph nodes would corroborate the presence of melanocytic cells in the nodes.

249

Figure 4.19. Melanocytic lesions demonstrate metastatic potential by pigmentation of regional lymph node The proliferation of melanocytic tissue is confirmed by histological analysis using H&E staining (i) and immunofluorescence using Melan-A specific antibody (red) and DAPI nuclear stain (blue) (ii). Multiple lesions are observed in the skin (iii) and melanoma development is evidenced by metastatic spread of pigment to the inguinal lymph nodes (iv-v). Representative image of a heterozygous mouse is shown.

250

These data suggest the potential for the melanocytic lesions to be melanomas of the skin in TRIM16 heterozygous mice. Evidence of pigmentation in the regional lymph nodes of two heterozygous mice suggests the possibility that melanoma cells have migrated from the skin tumour (Figure 4.19i) and travelled to the regional lymph node. The presence of the heavily pigmented entities in the adipose tissue away from the melanocytic mass may indicate the presence of melanophages (450). Positive immunofluorescence staining of the lymph nodes for melanoma marker, melan-A or S-100, would identify if the pigmentation observed in the regional lymph nodes (Figure 4.19iv-v) was due to the presence of melanoma cells, or melanophages (450). However, Azarova and colleagues have reported the melanocytic lesions, with identical histology, induced in topoisomerase II knockout mice to be melanomas by the presence of positive Melan-A staining of the lesion (443). In this thesis, we have termed the growths “melanocytic lesions” as it is presently unclear if these are melanomas of the skin.

251

4.2.4 Study the role of TRIM16 in skin carcinogenesis in the TRIM16 full-tissue knock-out mice

Development of skin melanocytic lesions, papilloma and squamous cell carcinoma in the TRIM16 full-tissue knockout mice allows the determination of the impact of complete loss of TRIM16 in skin cancer development. As TRIM16 protein can be secreted (304), the full tissue TRIM16 knockout model removes the influence of stromal TRIM16 protein signalling and the potential for anti-proliferative effect of stromal protein signalling, that we hypothesized may be a factor in the skin-specific knockout model where larger melanocytic lesions were observed in the TRIM16 homozygous mice (Figure 4.16).

Importantly, the melanocytic lesion/melanoma development to be assessed in the full-tissue knockout mice is critical, where TRIM16 is deleted from the melanocytes as well as keratinocytes. In addition, the development of papilloma and squamous cell carcinoma will also be assessed. The treatment protocol was the same as described for the skin-specific TRIM16 knockout mice (Chapter 2, Section 2.2.2) and treatment duration was extended to 40 weeks to allow for increased numbers of papilloma to convert to SCC.

We sought to determine if full-tissue TRIM16 homozygous(+/+) mice developed larger melanocytic lesions compared to TRIM16 wild-type(+/+)mice. After 28 weeks of skin carcinogen treatment, mice were scored for melanocytic lesions >1 mm or <1 mm in size. TRIM16 heterozygous(+/-) mice developed fewer melanocytic lesions than either TRIM16 wild-type(+/+) or TRIM16 homozygous(+/+) mice (Figure 4.20). Both TRIM16 wild-type(+/+)and TRIM16 homozygous(+/+) mice developed predominantly larger lesions. Similarly, both TRIM16 wild-type(+/+)and TRIM16 homozygous(+/+) mice developed 2-5 mm papilloma and TRIM16 homozygous(+/+)mice developed the fewest larger (>5 mm) lesions (Figure 4.21).

252

4.2.4.1 TRIM16 full-tissue knockout mice did not increased melanocytic lesions size with skin carcinogen challenge

Figure 4.20. TRIM16 wild-type and heterozygous mice develop larger melanocytic lesions in full-tissue knockout mice. Melanocytic lesions were measured (mm) in TRIM16 wild-type(+/+) (N=15), TRIM16 heterozygous(+/-) (N=33) and TRIM16 homozygous(-/-) mice (N=26) at 28 weeks skin carcinogen treatment.

253

Figure 4.21. TRIM16 homozygous mice develop smaller papilloma compared to wild-type and heterozygous mice. Papilloma were measured (mm) in TRIM16 wild-type(+/+) (N=15), TRIM16 heterozygous(+/-) (N=33) and TRIM16 homozygous(-/-) mice (N=26) at 28 weeks skin carcinogen treatment.

254

In this section, we determined that TRIM16 wild-type(+/+) and TRIM16 homozygous(-/-) full tissue knockout mice develop both large and small melanocytic lesions in a similar proportion. TRIM16 heterozygous(+/-) mice, develop fewer papilloma and melanocytic lesions than the other genotypes. These results contradict with the findings for the skin-specific mice, which show that TRIM16 heterozygous(+/-) mice develop more numerous papilloma and melanocytic lesions. Importantly, the larger melanocytic lesions in the TRIM16 homozygous(-/-) mice that developed in the skin-specific model (Figure 4.16) were not observed in the full-tissue knockout model (Figure 4.20) and may reflect the loss of secreted keratinocyte TRIM16 signaling in the melanocytes no longer inhibiting lesion growth. In the full-tissue knockout model, TRIM16 is knocked out in the melanocytes, and the possible disruption of intercellular signaling between melanocytes and keratinocytes may explain this result. The dynamic of cell-cell communication is complex and requires further study to understand the role of TRIM16 in the extracellular signaling from keratinocytes to melanocytes.

4.2.4.2 TRIM16 full-tissue knockout mice did not reduced tumour latency with skin carcinogen treatment

In contrast to the melanocytic lesion development in skin specific TRIM16 knockout mice, full- tissue TRIM16 knockout mice do not display reduced latency to melanocytic lesion (Figure 4.22). There was a slight increase in papilloma latency between TRIM16+/+ wild-type and TRIM16+/- heterozygous mice at week 25 (p=0.023) (Figure 4.23). This result contrasts the observation in the skin-specific knockout mice, which shows reduced latency in both melanocytic lesion development (Figure 4.15) and papilloma development (Figure 4.10). Overall, the differences in melanocytic lesion and papilloma latency were minimal between genotypes in full-tissue knockout mice.

255

Figure 4.22. TRIM16 homozygous full-tissue mice did not have reduced latency for melanocytic lesion development. Melanocytic lesions were measured for TRIM16 wild-type(+/+),TRIM16 heterozygous(+/-), and TRIM16 homozygous(-/-) mice from 15 to 28 weeks. Statistical significance was determined by the Student’s t-test between all genotypes within a given week. No significant results were determined from this analysis.

256

Figure 4.23. TRIM16 homozygous full-tissue mice did not reduced latency for papilloma development. Papilloma number were measured for TRIM16 wild-type(+/+),TRIM16 heterozygous(+/-), and TRIM16 homozygous(-/-) mice from 15 to 28 weeks. Statistical analysis was performed by the Student’s t-test between all genotypes within each given week.

257

In contrast to the skin-specific mice, full-tissue TRIM16 knockout mice did not exhibit a change in latency of papilloma or melanocytic lesions across genotypes. This result may be due to the complete loss of TRIM16 from all tissues, which negates the potential for TRIM16 mediated extracellular signaling that may influence papilloma/lesion development. This could also reflect the difference in genetic background of the two delete mice, KRT14Cre for the skin-specific deletion and B6-Tg (Cmv-Cre)1Cgn for full-tissue TRIM16 deletion. The time to development of melanocytic lesions was delayed overall in the full-tissue knockout mice with multiple lesions developing around week 21 (Figure 4.22), compared to week 13 in the skin-specific model (Figure 4.15) with the same concentration of DMBA/TPA used in both systems.

258

4.3.2.3 TRIM16 full-tissue knockout mice do not develop squamous cell carcinomas compared to wild-type and heterozygous mice

C57BL/6 are expected to develop papilloma at an incidence of 42% and approximately 2.2 papilloma per mouse at the present dosing schedule (Chapter 2, Section 2.2.2) (442). Papilloma can be expected to convert of SCC after a treatment period of minimum 20 weeks (442). These must be determined histologically, though the identification of SCC can be determined macroscopically by the characteristic flattened, downward appearing growth (442) (Figure 4.11). The SCC’s that develop from malignant conversion from papilloma are similar in histology to human SCC (442). The conversion rate of papilloma to squamous cell carcinomas in C57BL6 mice receiving DMBA/TPA dosing in line with this study is unknown, though a similar study using the same DMBA concentration (25µg) and half the dose of TPA (2µg) demonstrated a papilloma conversion rate of 0.2% (451). This low conversion is due to the resistant nature of C57BL/6 mice to tumour development (451) (442).

Our study demonstrates that at 28 weeks treatment, TRIM16 wild-type(+/+) mice develop 3/15 (SCC/number) of mice and that 3/15 (20%) papilloma develop into SCC (Table 4.2). In TRIM16 heterozygous(+/-) mice, 4 SCC out of 33 mice treated were observed and 4/18 (22%) papilloma develop into SCC. For the TRIM16 homozygous(-/-) mice, no SCC were observed despite 23 mice receiving treatment and 26 papilloma developing, none were converted to SCC in TRIM16 homozygous(-/-) mice.

259

Table 4.2. Summary of development of squamous cell carcinoma at 28 weeks in full-tissue knockout mice

260

In this section, we have shown that TRIM16 homozygous(-/-) mice do not develop SCC tumours compared to TRIM16 wild-type(+/+) and TRIM16 heterozygous(+/-) mice. This is the opposite result to our hypothesis according to our in vitro data, in which TRIM16 over-expression reduces cell proliferation, demonstrated in both melanoma (Figure 3.13A) and neuroblastoma (350, 351). Therefore, we hypothesized that TRIM16 loss would result in an increase in cell proliferation and tumour growth. However, gene silencing of TRIM16 in melanoma cells demonstrated a decrease in cell proliferation (Figure 3.16A) and possible increase in epithelial- mesenchymal transition (EMT) and markers for of EMT (Figure 3.17). This suggests that the relationship between TRIM16 expression level and cell proliferation is complex and possibly cancer type specific.

4.3 Discussion

Ablation of a gene in a GEM model can result in an increase or decrease in tumour latency (452), or prevention of tumour development (453). This observation can lead to understanding of the underlying mechanism of how a given gene contributes to tumour development and progression. An example of this is the loss of integrin α3 (Itga3) which prevents skin tumour formation (453). Itga3 mice were investigated for their role in skin carcinogenesis after the observation that skin tumorigenesis correlated with altered expression of Itga3, suggesting it may play a role in tumour progression. Itga3 KO mice were treated with DMBA/TPA to induce papilloma, keratoacanthoma and SCC tumours. Tumour initiation in the absence of Itga3 was significantly decreased and epidermal proliferation was increased (453). It was further determined that due to the increase in epithelial cell turnover, the mutant Hras cells from DMBA treatment were lost with epidermal turnover and initiation of tumorigenesis was impaired (453). Similarly, Cd151 KO mice demonstrated increased epidermal proliferation in response to DMBA/TPA skin carcinogen treatment and a reduction in skin tumour number and size. The mechanism of action involved the increase in differentiation of keratinocytes resulting in an increase loss of tumour promoting cells (454). These data highlight the complexity of the skin carcinogenesis model and the requirement to understand underlying mechanisms that pertain to a phenotypic result.

An important measure of a gene contribution to tumorigenesis aside from tumour number and size is tumour latency (452, 455). PTEN is a known tumour suppressor protein in numerous malignancies. Temporal PTEN loss in mice from a Cre/LoxP system targeting Pten exon 5,

261

inducible by 4-OHT, resulted in multiple malignancies, the most prominent of which was lymphoma. The average tumour latency was 17 weeks in the Pten-deficient mice. Of interest was the observation that female mice developed tumours with reduced latency (10-11 weeks) compared to males (21 weeks) (452). This data demonstrates the rapid development of malignancy with the loss of a key tumour suppressor protein alongside the potential for gender to influence tumour latency.

In the human epidermis, one melanocyte interacts with approximately 36 keratinocytes to supply UV protective melanin (456, 457). Furthermore, melanocytes are intricately regulated by keratinocytes and stromal factors in the skin (13). These can be regulated by paracrine growth factors and cell-cell adhesion molecules (53). Melanocytes escape keratinocyte regulated growth control by down-regulating adhesion molecules such as E-cadherin (53), increasing melanoma-to-melanoma and melanoma-to-fibroblast cell adhesion molecule, N- cadherin, and, loss of cell anchorage due to changes in expression of integrin protein family members (53, 458). Our data demonstrated that loss of TRIM16 in a keratinocyte-specific knockout mouse model resulted in the development of larger melanocytic lesions in homozygous TRIM16 deleted mice after skin carcinogen challenge. The development of larger melanocytic lesions may indicate an increase in either radial migration of cells and/or increased melanocytic cell proliferation in TRIM16 keratinocyte knockout mice. Our result may also be due to a paracrine loss of TRIM16 signaling to melanocytes from the adjacent keratinocytes, since both TRIM16 and its target gene, IFN1, are known to be secreted into the extracellular environment (304, 459, 460). Based on our data in Chapter 3, showing that TRIM16 gene silencing reduced cell proliferation in melanocytes (Figure 3.15A) and melanoma cells (Figure 3.16A), it is possible that tumours may develop at a reduced rate in vivo as Table 4.2 suggests, but tumours that do arise may have a more aggressive disposition due to an increase in EMT phenotype given the increase in EMT markers, that may occur with TRIM16 silencing (Figure 3.17). This requires more study and in particular, the assessment of TRIM16 gene silencing in SCC and the evaluation of EMT markers. In addition, evaluation of EMT markers in SCC tumours comparing wild-type, heterozygous and homozygous TRIM16 mice may provide insight into the molecular pathology of tumour development in vivo.

Flower (Fwe) deficient mice have a reduced susceptibility to skin papilloma formation (461). Like TRIM16 knockout mice, Fwe mice have no dicernable phenotype but display a significantly lower number of papilloma after DMBA/TPA carcinogen treatment compare to wild-type and heterozygous mice (461). In the skin-specific TRIM16 heterozygous knockout 262

mice, it is observed that a reduced latency and increased number of papilloma develop in TRIM16 heterozygous mice (Figure 4.10 and Table 4.1). However, loss of both copies of TRIM16 results in fewer papilloma, and in the full-tissue TRIM16 knockout mice, no SCC are observed in homozygous mice but develop in both wild-type and heterozygous mice. This suggests that loss of a single copy of TRIM16 increases tumour development and reduces latency, but loss of both copies abrogates this result. It is possible that a gene compensation effect is occuring by which loss of TRIM16 results in an increased expression of other genes (possibly members of the TRIM family) to compensate the lost of TRIM16. It would be interseting to test this hypothesis by performing a PCR or microarray of tissue from the full- tissue TRIM16 homozygous mice and determining the expression levels of TRIM family members compared to the heterozygous mice. Candidate TRIM compensations genes could be validated in vitro by knockdown of TRIM16 in SCC cells and over-expression of the candidate compensation gene to determine if TRIM16 effects on cell proliferation are ablated with gene compensation. Alternatively, the ablation of SCC development in TRIM16 homozygous knockout mice may indicate TRIM16 functions in vivo as an oncogene. An example of this is TRIM27, which exhibits complex function in cancer being characterized as both a tumour suppressor and oncogene (297). TRIM27 regulates RARα through PML, and colocalizes with the PML-RARα fusion in acute promyelocytic leukaemia (APL). TRIM27 also translocates with the RET tyrosine kinase giving higher catalytic activity than RET alone in lymphoma, and resulting in increased cell proliferation and tumorigenesis (297). To contrast the seemingly oncogenic activity of TRIM27, the protein has also been shown to induce apoptosis through a mechanism dependant on JUN N-terminal kinase (332). This profile of TRIM27 function in cancer suggests that the activity of TRIM proteins can be pleiotropic and hence, TRIM16 may suppress cell proliferation and migration in melanoma, but has the potential to increase tumorigenesis in SCC. Further investigation is required to determine if TRIM16 is a tumour suppressor protein in melanoma and this requires developing a melanocyte specific TRIM16 knockout mouse model that ascertains the function of TRIM16 in contributing to melanomagenesis.

To specifically address the potential for TRIM16 to act as a tumour suppressor protein in melanoma, a melanocyte specific TRIM16 knockout model would be a more suitable model to use. In addition, a model that has amplification of a known melanoma oncogene e.g. CDK4 may be a way to ensure tumour development and assess the contribution of TRIM16 loss in facilitating this process. A limitation of using a mouse model system for human melanoma is

263

the presence of murine melanocytes in the hair follicles and not the dermal-epidermal junction in hairy parts of the skin, as is present in human disease (462). To effectively address this, mutation in NRASQ61R on an INK4a deficient background results in epidermal melanocyte presence, more closely approximating human disease (463). It would be interesting to cross the NRASQ61R/INK4a deficient background mice with TRIM16 melanocyte specific knockout mice and assess the loss of TRIM16 in the contribution to melanoma development. This is important as the effect of the knockout of a gene may only become evident with the inactivation of another gene (439, 440). Additional knockdown of a known tumour suppressor gene important in melanoma pathogenesis e.g. p16 (145, 424) or PTEN (91, 144, 408) (discussed below) may demonstrate an enhanced melanoma phenotype in conjunction with the loss of TRIM16.

Mouse models of melanoma development have shown that multiple genetic aberrations are required for the induction of carcinogenesis (91, 125). Further investigations of the role of TRIM16 in melanomagenesis in melanocyte-specific TRIM16 knockout mice are needed to explore this complex interaction between keratinocytes and melanoma cells. One mouse model involves the melanocyte-specific expression of BRAFV600E crossed with the Cre inducible loss of Pten, resulting in melanomas with 100% frequency (91). The importance of the MAPK and PI3K pathways was confirmed by the prevention of melanoma development by MEK1/2 (PD325901) or mTORC1 inhibitors, respectively. A cross of the BRAF/PTEN expression/knockout model with a TRIM16 melanocyte specific knockout mouse would definitively address the contribution that loss of TRIM16 has to melanoma development. Our present data would suggest that this BRAF/PTEN model crossed with loss of TRIM16 knockout mice may result in a higher rate of melanoma cell migration and metastasis. As melanomas develop at 100% frequency with the BRAF/PTEN system, it would be interesting to cross the TRIM16 melanocyte knockout mice with BRAF melanocyte amplified mice to determine if TRIM16 loss could induce benign nevi proliferation (from the BRAF amplified mice alone) to a malignant phenotype.

4.4 Conclusions

Homozygous knockout of TRIM16 in skin-specific mice result in the development of larger melanocytic lesions, possibly due to the stromal crosstalk between keratinocytes and melanocytes in the skin. Homozygous knockout of TRIM16 in full-tissue mice results in the

264

ablation of SCC development and may reflect TRIM16 as a candidate oncogene or a gene compensation effect (possibly from another TRIM family member) with TRIM16 gene deletion.

265

Chapter 5

Identification, characterization, and mechanism of action of compound 012 as an enhancer of vemurafenib

266

5.1 Introduction

Melanoma has the highest mutation load of any type of cancer (464). This is thought to be due to the main causative factor of UV induced DNA damage sustained by melanocytes in the skin that go on to develop melanoma (464), (11). Interestingly, NRAS/BRAF melanomas carry a higher mutation load than melanomas that harbour either mutation, though the reason for this is unknown (11). Melanoma is a heterogeneous cancer with a high propensity towards metastasis due to its neural crest origins (4, 10). Median overall survival of patients with melanoma metastasis is less than 1 year (45). Recent clinical advances in the treatment of metastatic melanoma harbouring the BRAFV600E mutation involve the use of targeted BRAF inhibitors, notably vemurafenib which results in dramatic tumour regression and significantly improved patient quality of life. However, vemurafenib resistance typically develops at around 7-9 months of treatment (63, 210, 221), highlighting the need for combination treatments that can work in synergy with vemurafenib and further reduce tumour burden. There are conflicting studies as to the prognostic value of BRAF mutation in melanoma, potentially due to the influence of additional genetic changes. Association between the presence of BRAF mutation and poor patient survival does not mean BRAF mutation is a direct cause of poor prognosis (100). The role of an oncogene is to be considered within the context of additional changes promoting carcinogenesis. For example, thyroid cancers harbouring BRAF mutations are more likely to invade regional lymph nodes then wild type BRAF cells. Conversely, the BRAF mutation associates with a favourable micro-satellite unstable phenotype in colorectal cancer (100). Hence, the BRAF oncogene has the ability to act as a biomarker of poor prognosis in one context, and good prognosis in another. For the case of melanoma, the associated mutations alongside the BRAF mutation may provide further context to determine the major drivers of melanoma associated with poor prognosis (100). As it stands, most of the evidence suggests BRAF is a predictive marker of poor patient outcome and may be further sub-typed depending on associated aberrations.

Vemurafenib is mostly specific to melanomas harbouring the BRAFV600E mutation, though some BRAFWT melanomas are also inhibited by vemurafenib, demonstrating a similar drug response profile to the mutant tumours (210, 249, 465). In one study determining the differential response of mutant and wild-type melanomas to vemurafenib, multiple kinase substrates in addition to BRAF are significantly affected by vemurafenib in Mel-JD BRAFWT cells indicating extensive off-target effects in these cells. Surprisingly, reduction of overall

267

phosphorylation was most marked in Mel-JD cell compared to BRAFV600E MM200 cells after vemurafenib treatment of cell lysates. Researchers concluded that a presently unknown subset of BRAFWT patients may benefit from vemurafenib treatment clinically (249). Furthermore, in a study on BRAFWT SCC that were treated with PLX4720, researchers discovered suppression of apoptosis through the inhibition of multiple off-target kinases giving further evidence for additional kinase targeting by vemurafenib (465). Presently, NRAS mutated melanoma is associated with poor prognosis and remains without targeted therapy (211). Directly targeting the NRAS protein has been ineffective thus far, and research targeting upstream farnesyl transferases has also not proven successful (65, 211). Identification of patients that are NRASWT/BRAFWT and NRASMT/BRAFWT who may benefit from vemurafenib therapy could provide a novel therapeutic option for these patients. Furthermore, identification of compounds that work in synergy with vemurafenib could contribute towards a mechanistic understanding of pathway inhibition required for effective BRAF inhibitor synergy, and, a means to further reduce tumour burden for NRAS mutant patients.

It is clear that due to multiple pathway deregulation in melanoma, targeting of tumours requires a combination strategy to effectively reduce tumour burden. Pathways that are deregulated in melanoma include invasion and metastasis, growth signalling, evasion of apoptosis, angiogenesis, replicative potential and others (49, 50). In targeting melanomas a multi-faceted approach is required based on known genetic aberrations to effectively block the many mechanisms in place promoting cancer survival.

In this chapter, we sought to identify a compound that is synergistic with the BRAF inhibitor, vemurafenib, and delineate the mechanism of action for the novel compound. We have characterized this compound in vitro for its effects on cell proliferation, cell viability, colony forming ability and migration. Lastly, we sought to determine the in vivo efficacy of the combination of vemurafenib and compound 012 by employing a melanoma xenograft model.

5.2 Results

5.2.1 Identification and characterization of three anti-melanoma compounds

Anti-melanoma compounds were identified from a 10,560 compound library, initially investigated for enhancing activity with the histone deacetylase inhibitor,

268

suberanilohydroxamic acid (SAHA/Vorinostat). Twenty-four hit compounds were tested as enhancers of SAHA in breast cancer (MDA-MB-231 cells) and melanoma (G361 cells) (Figure 5.1). Seven of the 24 hit compounds were validated as enhancers the cytotoxic effects of SAHA in MDA-MB-231 breast cancer cells, but no compounds were found as SAHA enhancers in G361 melanoma cells. Compounds, 012, 236, and 759 were identified as having anti-melanoma activity as single agents at 10 μM with limited toxicity to normal embryonal fibroblasts (Table 5.1 & Figure 5.2). For this reason, these compounds were investigated as anti-melanoma agents, as a separate project of the SAHA screening study. These compounds were investigated as potential enhancers of the BRAF inhibitor, vemurafenib.

269

Figure 5.1. Identification of three anti-melanoma compounds from compound SAHA enhancer screening project

270

5.2.1.1 Determination of the IC50 for three anti-melanoma compounds

When screened against a panel of melanoma cell lines, compounds 236, 759 and 012 demonstrated single agent anti-melanoma activities at 10 µM, and limited toxicity to normal fibroblast cells (MRC-5 and WI-38) and normal human epidermal melanocytes (NHEM) (Table 5.1 and Figure 5.2). These three agents were further investigated for medicinal chemistry properties, amenability to structural modification and drug-likeness and potential to work in synergy with current clinically used therapies treating advanced cutaneous melanoma.

271

Table 5.1 IC50 of compounds against a melanoma cell line panel and normal human embryonal fibroblast lines

IC50 (μM) Cell line #012 #236 #759 Mel-JD 3.4 3.7 3.6 Mel-RM 4.3 7.8 6.3 Mel-CV 5.9 3.4 3.9 MM200 8.2 8.5 10 M4405 11.6 10 12 A375 6.8 2.8 10 G361 3.4 3.5 3.6 MRC-5 10 10 12.2 WI38 32.7 ND 40

272

5.2.1.2 Compounds demonstrate greater efficacy against melanoma compared to normal fibroblasts

An important aspect of selecting a compound for further development is establishing a therapeutic window in which the compound exhibits maximal toxicity against the target tissue and minimal toxicity against normal tissues (430, 466). A therapeutic index is considered to be the ratio of the highest exposure to the drug that results in no toxicity to the exposure that produces the desired efficacy (pharmacology) (467). In vitro, the assessment of therapeutic index is based on the on-target vs. off-target effect and an IC50 is used to determine efficacy against the target tissue (melanoma) compared to the normal tissues (melanocytes and fibroblasts) to gain understanding of the specificity of a drug to the desired target. The IC50 is the amount of drug required for 50% inhibition in vitro (467). Cell lines with an IC50 greater than 10 µM to an agent are considered to be resistant to the test compound (105). Melanoma cell lines displayed a range of effective IC50’s in response to all 3 compounds (from 2.8 to 10 µM) with only compound 012 being inefficient on the M4405 cell line (IC50 11.6 µM). All 3 compounds displayed limited toxicity (≥10 µM) against control fibroblast cell lines (Table 5.1).

273

Figure 5.2. Anti-melanoma compounds reduce cell viability in melanoma cells, but have limited toxicity to normal fibroblasts. Melanoma cell lines, A375, G361, MM200, Mel-CV, Mel-RM and Mel-JD and normal fibroblast cells, MRC-5 and WI-38 were treated with DMSO vehicle control or compounds 759, 012 or 236 at 10 µM for 72 hours. Cell viability was determined by the Alamar Blue assay and expressed as % of control.

274

5.2.1.3 Compounds decrease melanoma cell colony forming ability

Anti-melanoma compounds were tested for their ability to form colonies from a single cell. Cell lines, MM200 (BRAFV600E) and Mel-RM (BRAFWT) were seeded into 6 well plates at a final density of 100 cells/well. Cells were treated with either DMSO control, or compounds

012, 236 or 759 at the indicated IC50 concentration (Figure 5.3). Colonies were allowed to form over 14 days before quantitation. A colony was determined to be a cluster of 50 or more cells.

We found that colony forming ability was ablated with treatment of the IC50 dose for compound 236, and was reduced with compounds 012 and 759.

275

Figure 5.3. Colony forming ability is reduced in the presence of all three anti-melanoma V600E compounds at IC50 doses. Melanoma cell lines, MM200 (BRAF ) and Mel-RM (BRAFWT), were seeded into 6-well plates at 100 cells/well in duplicate. Cells were treated with the IC50 doses indicated and colonies allowed to form over 14 days. Cells were stained with crystal violet and colonies counted.

276

5.2.1.4 Lipinski’s rule of 5

We gained advice from a qualified medicinal chemist (Walter and Eliza Hall Institute) on the nature of the three candidate structures. All three compounds obeyed Lipinski’s rule of five for orally active drugs. The small molecule must have no more than 5 hydrogen bond donors, no more than 10 hydrogen bond acceptors, a molecular mass of less than 500 daltons, and an octanol-water partition coefficient log P of no greater than 5 (468). Compound 012 has a molecular weight less than 500 daltons and a log P value of 4.69. Within these parameters, the compound 012 is anticipated to be bioavailable (468). Furthermore, the core pyrin ring system of compound 012 and substituents are found in many bioactive molecules and in this way is thought to be stable under physiologic conditions (468). In addition, compound 012 is amenable to structural development by modification of side groups (Figure 5.4). Compound 759 is also highly amenable to structural modification and is within Lipinski’s criteria for orally available drug-likeness. As compound 236 is almost completely unamenable to chemical modification (Figure 5.4). Compound 236 is also under patent, unlike compound 012 and 759 and this compound was the least desirable to progress in drug development. Within 6 melanoma cell lines were tested, the two melanoma cell lines A375 and MM200, were found to be resistant to compound 759 at 10 µM (Figure 5.2), this was a less desirable compound to develop in comparison with the most effective compound. Compound 012 displayed consistent anti-melanoma activity over a range of melanoma cell lines (Figure 5.2), is amenable to structural modification (Figure 5.4), and displayed synergism with dacarbazine and vemurafenib, and hence we chose to develop this compound further.

277

Figure 5.4. Structure of the three anti-melanoma compounds and comment on the suitability as a lead compound for drug development. The medicinal chemistry assessment from Walter and Eliza Hall (WEHI) on compound amenability to development and the compound patent status was determined by Scifinder.

278

5.2.1.5 The three anti-melanoma compounds demonstrate activity in combination with dacarbazine and vemurafenib

Combination therapy for the treatment of metastatic melanoma is essential due to the highly heterogeneous nature of the disease (101, 250, 469). For this reason, it was of interest to test the three anti-melanoma compounds in combination with the previous standard of care, dacarbazine, and the present treatment for BRAFV600 melanomas, vemurafenib. This was performed by combination studies in A375 cells using the IC50 dose of each compound in combination with the IC50 dose of either dacarbazine or vemurafenib in 1.5 fold increasing or decreasing concentrations (Figure 5.5). This data demonstrates significant potency with vemurafenib for all 3 compounds (Figure 5.7, top panel). Compound 236 was not tested in combination with dacarbazine due to time constraints, and compound 759 showed limited efficacy with dacarbazine. Compound 012 demonstrated a further reduction in cell viability with vemurafenib in a 1:12 ratio and also with dacarbazine in a 1.5:1 ratio and hence we were interested in developing the combination of compound 012 and vemurafenib further.

279

Figure 5.5. Anti-melanoma compounds demonstrate increased potency with vemurafenib over dacarbazine. BRAF mutant melanoma cells, A375, were treated with vemurafenib

(VEM) or dacarbazine using a dose range centred on the respective compound IC50 with a 1.5 fold increase or decrease for each compound tested and a 1.5 fold increase or decrease of VEM or dacarbazine. Cell viability was measured using the Alamar Blue assay and expressed as % of control. Statistical analysis between the lowest and highest dose combination is shown. **p<0.01, ***p<0.001.

280

Taken together, these data indicate that compound 012 is the most suitable candidate to investigate for synergistic anti-cancer activity with vemurafenib in melanoma. Compound 759 also demonstrates potential as an enhancer of both dacarbazine and vemurafenib, however, since melanoma cell lines, A375 and MM200 (both BRAFV600E lines) are resistant to compound 759 (Figure 5.2) and hence, compound 012 is a more desirable compound for synergy testing. As vemurafenib specifically targets mutant BRAFV600 melanoma cells, we proposed that compound 012 may be synergistic specifically with mutant BRAF melanoma cells.

5.2.2 Characterization of compound 012 synergy with vemurafenib

5.2.2.1 Compound 012 demonstrates minimal synergy with vemurafenib in BRAF mutant cells

As vemurafenib preferentially targets mutant BRAFV600 melanoma cells, we proposed that compound 012 may be synergistic with vemurafenib specifically in mutant BRAF melanoma cells. To test this hypothesis we initially performed cell viability assays on one BRAF mutant cell (A375) line and one BRAF wild type cell line (Mel-JD) using (IC50 dose range) for 72 hours, and calculated synergism using the commercially available program, Calcusyn. Calcusyn uses an algorithm to calculate the combination index (CI) where in general a CI of <1 indicates synergy (for more details see Figure 5.6). The CI is determined at a range of effective doses (ED) where an ED50 represents 50% cell death, ED75 represents 75% cell death, and ED90 represents 90% cell death, and it is important to note that synergism at high levels e.g.

ED90 is more therapeutically relevant than synergism at low levels e.g. ED50 (470). A determined range of synergism based on CI is as follows; 0.1 is very strong synergism, 0.1–0.3 is strong synergism, 0.3–0.7 is synergism, 0.7–0.85 is moderate synergism, 0.85–0.90 is slight synergism, 0.90–1.10 is nearly additive, 1.10–1.20 is slight antagonism. Values greater than this are antagonistic (470). The combination index for A375 (BRAFV600E) cells was determined by performing a synergy assay of compound 012 and vemurafenib in a 12:1 ratio based around the IC50 values. The results shown in Figure 5.6 demonstrate moderate synergism at the ED75, and ED90 values (CI of 0.76 and 0.72 respectively), but not at ED50 (CI = 2.03). (Figure

5.6B&C). This indicates antagonism at the ED50 value, and moderate synergism at the ED75, WT and ED90 values. Similarly, Mel-JD (BRAF ) cells were treated with compound 012 and

281

vemurafenib in a 1:1.25 ratio. The CI values were determined to be 0.40, 0.51, and 0.65, at

ED50, ED75, and ED90, respectively (Figure 5.7B&C) indicating synergism at all levels of cell death. Additional cell lines, Mel-RM, Mel-CV, MM200, and G361 were also tested for synergy between compound 012 and vemurafenib. Both BRAFWT cell lines exhibited synergy at all EDs, whereas only 2/4 BRAFV600E cell lines showed some degree of synergy at the higher EDs (summarised in Table 5.2).

282

Figure 5.6A

Figure 5.6B

283

Figure 5.6C

Figure 5.6. Compound 012 demonstrates synergy with vemurafenib in BRAFV600E mutant cells. A) A375 cells were treated with compound 012 and vemurafenib (VEM) in a 12:1 ratio for 72 hours and cell viability performed by the Alamar Blue assay. B) Synergy was determined by the Calcusyn algorithm. The fraction affected by dose (Fa) is plotted against the combination index (CI). A combination index <1 was determined as synergistic and indicated by a red x. C) Effective dose (ED) values were used to determine level of synergy between compound 012 and VEM.

284

5.2.2.2 Compound 012 demonstrates strong synergy with vemurafenib in BRAF wild-type cells

As vemurafenib specifically targets the mutant BRAFV600 protein and has limited efficacy against the BRAFWT protein (206), BRAFWT cells were initially used as a negative control for synergy between BRAFV600E cells and compound 012. However, the combination between vemurafenib and compound 012 unexpectedly demonstrated greater synergy in the BRAFWT cells than the BRAFV600E cells (Figure 5.7A). At a concentration of 4.9µM, the compound 012 combined with 6.0µM vemurafenib, the single agent’s demonstrated minimal activity, while the combination showed an 80% reduction in cell viability (Figure 5.7A).

285

Figure 5.7A

Figure 5.7B

286

Figure 5.7C

Figure 5.7. Compound 012 demonstrates synergy with vemurafenib in BRAFWT mutant cells. A) Mel-JD cells were treated with compound 012 and vemurafenib (VEM) in a 1:1.25 ratio for 72 hours and cell viability performed by the Alamar Blue assay. B) Synergy was determined by the Calcusyn algorithm. The fraction affected by dose (Fa) is plotted against the combination index (CI) A combination index <1 was determined as synergistic and indicated by a red x. C) Effective dose (ED) values were used to determine level of synergy between compound 012 and VEM

Table 5.2. Summary of cell line and synergy between compound 012 and vemurafenib

287

These data highlight the difference in synergy between the BRAFV600E cells (A375) and BRAFWT cells (Mel-JD) for compound 012 and vemurafenib. While the A375 cells demonstrate some moderate synergy in the combination of the two agents, the two compounds demonstrate a strong synergy in the Mel-JD cells. This is unexpected, as vemurafenib preferentially targets the mutant form of BRAF and has low efficacy against the wild-type V600E BRAF cells. However, though vemurafenib has a 10-fold lower IC50 towards the BRAF melanoma cells, in combination with compound 012, vemurafenib demonstrates an increased efficacy for the BRAFWT cells. In addition, although BRAFWT cells are preferentially targeted by compound 012/vemurafenib combination, they did not need to harbour the NRAS mutation to demonstrate synergy with vemurafenib. The compound 012 and vemurafenib demonstrate synergy in the cell lines, CHL-1 and MeWo, even though both lines lack BRAF or NRAS mutation. This allows the possibility of application of vemurafenib with a novel a small molecule compound to BRAFWT tissues, which has not been investigated to date.

Encouragingly, both compound 012 and vemurafenib have low single agent activity in BRAFWT melanoma cells (Figure 5.7A) which indicates the agents may target specific anomalies (430). However, vemurafenib has been shown to specifically target the BRAFV600 melanomas, resulting in decreased ERK phosphorylation, and paradoxically, transactivates CRAF in NRAS signalling in mutant melanomas and promote ERK phosphorylation (200, 221, 471). Known off-target effects of vemurafenib include the inhibition of ACK1, SRMS and MAP4K5 (88). It is possible that these off target inhibitions are reversed or minimised by the combination of 012 and vemurafenib, and hence contribute to the synergy observed between them. Testing of the new BRAF inhibitors known as ‘paradox breakers’ would also be of interest to determine if a similar mechanism of action is observed with the combination of compound 012 and vemurafenib. Paradox breakers (PB) such as PLX7904 (PB04) effectively attenuate ERK phosphorylation in BRAFV600 melanoma cells, but do not transactivate CRAF in BRAFWT cells. PLX7904, like vemurafenib, has the off target effect of up-regulating FOXD3, an inhibitor of G1 to S phase transition (472), and it would be interesting to determine if a similar mechanism contributes to the effectiveness of the combination of compound 012/vemurafenib in BRAFWT melanomas. Combination of compound 012 with PLX7904 would provide insight as to whether the transactivation of CRAF, and subsequent ERK phosphorylation, was required for the mechanism of action in reducing cell viability with drug combination.

288

5.2.2.3 Compound 012 demonstrates synergy with vemurafenib across a panel of BRAF wild-type melanoma cells

From the data obtained in the synergy studies, a standardized dose of compound 012 and vemurafenib was estimated for the BRAFV600E and BRAFWT cell lines. We determined this to be 4 µM of compound 012 for both BRAFV600E and BRAFWT cell lines, with 5 µM vemurafenib for the BRAFWT lines, and 0.5 µM vemurafenib for the BRAFV600E cell lines. We applied this dose to a panel of BRAFWT melanoma cells and determined a significant combination effect in Mel-JD, Mel-RM, SKMel2, and IPC-298 cell lines (Figure 5.8). As the dose is standardized across cell lines, it is sub-optimal for some of the cell lines. This standardized dose was applied to normal melanocytes (NHEM) and normal fibroblasts (WI-38 and MRC-5) and displayed no toxicity towards the normal fibroblasts, with limited activity against the normal melanocytes in combination (Figure 5.8). The morphology and evidence of cell death can be seen in Figure 5.11 for CHL-1, Mel-JD, Mel-RM BRAFWT cell lines. The BRAFWT cell line, M4405 appears resistant to the compound 012/vemurafenib combination (Figure 5.9 & 5.13B).

289

Figure 5.8. Combination 012/vemurafenib is effective against a panel of BRAFWT melanoma cell line with minimal toxicity to normal fibroblasts. Normal melanocytes (NHEM), normal fibroblasts (MRC-5 and WI-38) and BRAFWT melanoma cells, Mel-JD, Mel-RM, SK-Mel2 and IPC-298, were treated with DMSO control, compound 012 (4 µM), vemurafenib (VEM, 5 µM) or combination compound 012/VEM for 72 hours. Cell viability was determined by the Alamar blue assay as a % of control.

290

Figure 5.9. Combination 012/vemurafenib effectively induces cell death in BRAFWT cell lines. Melanoma cell lines, CHL-1, Mel-JD, Mel-RM and M4405, were treated with DMSO control, compound 012 (4 µM), vemurafenib (VEM, 5 µM) or combination for 72 hours. Cell morphology was observed by microscopy and representative images are shown.

291

These data indicate that the 012/vemurafenib combination is effective across a panel of BRAFWT melanoma cells with 6/7 BRAFWT melanoma cell lines tested showing significant sensitivity. One cell line, M4405, demonstrates resistance to the combination. This is the only line that is null for p53 which may contribute to resistance to the drug combination.

5.2.2.4 Combination of compound 012 and vemurafenib induces cell senescence

The mechanism of the reduction in cell viability (Figure 5.8) was investigated by looking at senescence. The primary mechanism of vemurafenib is the induction of apoptosis (206, 207), but there is also evidence to show that vemurafenib can induce senescence (209). Senescence is a stress response that is used by cells to block proliferation (473). Staining for the senescence biomarker, β-galactosidase (474) was employed in Mel-JD and MM200 cells treated with DMSO control, compound 012, vemurafenib, or the combination at the indicated concentrations for 5 days. Compound 012 alone induced senescence in Mel-JD cells with 95% of cells staining positive for β-galactosidase at 4µM treatment. This suggests senescence induced by compound 012 may contribute to reduced cell viability when used in combination with vemurafenib. In MM200 cells, the combination of compound 012/vemurafenib increased the proportion of senescent cells to 45% (Figure 5.10), further supporting the involvement of senescence as the mechanism of action for this combination.

292

Figure 5.10. Compound 012 induces senescence in BRAFWT cells and combination compound 012/vemurafenib induces senescence in both BRAFWT and BRAFV600E cells. Mel-JD and MM200 melanoma cells were treated with DMSO control, compound 012 (4 µM), vemurafenib (VEM, 5 µM and 0.5 µM respectively) or the combination for 5 days. Senescence was determined by positive β-galactosidase (blue) staining. 50-100 cells were counted from triplicate experiments and the number of β-galactosidase positive cells expressed as a percentage.

The induction of apoptosis and/or senescence is a desirable outcome of an anti-melanoma agent. Here we show evidence that compound 012 induces senescence in Mel-JD cells. Furthermore, we show an induction of apoptosis in the Mel-JD cells with the combination of 012/vemurafenib. It is possible that these two mechanisms are occurring simultaneously in cells to significantly reduce cell viability. In MM200 cells we show an increase in senescence with compound 012/vemurafenib combination. Compound 012 alone demonstrates limited induction of senescence. This further highlights the potential for a different mechanism of action between BRAFWT and BRAFV600E cells in response to both compound 012 and vemurafenib.

293

5.2.2.5 Combination of compound 012 and vemurafenib reduces clonogenicity in both BRAF mutant and wild-type melanoma cells

Combinatorial synergy can be corroborated by long-term clonogenic assays (475). To this end, we performed clonogenicity assays on both BRAFV600E and BRAFWT cell lines with compound 012 and vemurafenib. Cell lines, Mel-JD, Mel-RM, and M4405 were seeded to single cell density and treated with DMSO control, compound 012, vemurafenib, or combination. Colony formation was assessed between the treatment groups after 14 days. Colonies were defined as having a cluster of 50 cells or more (476). Colony forming was reduced in Mel-JD cells and the smaller colonies were formed in the presence of 4 µM compound 012, approximately 50% of control (Figure 5.11A&B). Most interestingly, combination of compound 012/vemurafenib ablated colony forming ability in Mel-JD cells (Figure 5.11A&B). This was also evident in Mel-RM cells that showed few colonies forming in the presence of both compound 012 and vemurafenib (Figure 5.12B). BRAFWT cell line, M4405, appeared resistant to drug combination and the drug combination did not further impair colony forming ability over single agents (Figure 5.12B). Combination compound 012 and vemurafenib also impaired BRAFV600E cell line colony forming ability (Figure 5.12A), though to a lesser degree than BRAFWT cell lines (Figure 5.12B). BRAFV600E cell line, Mel-CV, appeared resistant to the combination (Figure 5.12A).

294

Figure 5.11A

Figure 5.11B

295

Figure 5.11. Combination compound 012/vemurafeib ablates colony formation in BRAFWT melanoma cells. A) Mel-JD cells were seeded at 100 cells/well and allowed to adhere. Cells were treated with DMSO control, compound 012 (4 µM), vemurafenib (VEM, 5 µM) or combination in duplicate. Colonies were formed over 14 days and stained with crystal violet for counting. Colonies were determined as having 50 cells or more. Representative colony forming assay for Mel-JD cells is shown. B) Number of colonies with treatment is represented in a histogram.

296

Figure 5.12A

Figure 5.12B

297

Figure 5.12. Combination of the compound 012/vemurafenib reduced colony forming ability in 2/3 of both BRAFV600E, and BRAFWT cells. A) BRAFV600E melanoma cells, A375, Mel-CV and MM200 were seeded at 100 cells/well and treated with DMSO control, compound 012 (4 µM), vemurafenib (0.5 µM) or combination for 14 days. Colonies were fixed and stained with crystal violet for counting. B) BRAFWT cells, Mel-RM, Mel-JD and M4405 were treated with DMSO control, compound 012 (4 µM), vemurafenib (5 µM), or combination for 14 days. Colonies were fixed and stained with crystal violet for counting.

298

The clonogenicity data corroborates the reduced cell viability data obtained with the combination of compound 012 and vemurafenib showed preferential to the BRAFWT melanoma cell lines. The significant reduction of colony forming ability in the presence both agents in Mel-JD and Mel-RM cell lines further suggests a role for vemurafenib treatment of BRAFWT melanoma in combination with an enhancing agent such as compound 012. Understanding the molecular mechanisms of enhancing drug action of vemurafenib in the BRAFWT cell lines is the key to applying this drug treatment, as is determining a subset of BRAFWT patients may benefit from this treatment.

5.2.3 Development of the compound 012 substructure library

5.2.3.1 Similarity search

Development of a focus library with a high degree of structural similarity to elucidate a more effective compound that demonstrates a higher degree of synergy may be a way forward in developing compound 012. In parallel to this, understanding the underlying mechanism and using a mechanistic understanding to drive development of more efficacious agents may be used to enhance the synergistic relationship in targeting melanoma cells and further increase the therapeutic window for drug combination.

Marketed drugs are frequently structurally similar to the lead compound from which they were derived (477). For this reason, a similarity library was developed with >90% structural similarity to the parent compound 012 using the ‘Hit2lead’ software. From this point, compounds were tested to determine the IC50 (Figure 5.13) in two melanoma cell lines, Mel- JDWT and MM200V600E, and the normal embryonal fibroblasts, MRC-5 and WI-38, to determine if a structurally similar compound demonstrates increased anti-melanoma activity measured by cell viability.

299

IC50 (μM) Compound Mel-JD MM200 Parent 012 3.4 8.2 012-424 2 30 012-770 1.8 30 012-803 2 18 012-903 2.5 30 012-984 3 35 012-714 4 50 012-400 2.5 3 012-149 2.5 3 012-031 >100 >100 012-009 7.5 30 012-042 8 35 012-302 13 40 012-315 12.5 40 012-378 2 7.5 012-382 15 40 012-443 22 40 012-452 4.5 35 012-482 10 35 012-536 25 40 012-551 28 70 012-699 2.5 8 012-894 20 30

Figure 5.13. IC50 of focus library from parental compound 012. The IC50 values were determined for the 22 compound focus library derived from the parent compound, 012 in BRAFWT (Mel-JD), and BRAFV600E (MM200) cell lines. These were determined by applying an escalating dose concentration of compound for 72 hours and cell viability was determined with Alamar blue assay.

300

5.2.3.2 Compound 012’s structure and activity relationship

From this focus library of compound 012, a structure-function activity relationship (SAR) was determined, based on the increase or decrease in IC50 in Mel-JD and MM200 cells, while maintaining a therapeutic window with limited toxicity to the normal fibroblasts, MRC-5 and WI-38. All compounds were also tested in combination with vemurafenib to determine if compounds demonstrated an increased anti-melanoma combination effect compared to the parental compound, 012.

Compounds that are proliferative as a single agent (Figure 5.15, lower panel) or show increased toxicity to MRC-5 and WI-38 cells are not selected for further study. A structure activity relationship (Figure 5.16) established in this focus library demonstrates that a 4-ethoxyphenyl group containing the ethoxy substituent (Figure 5.14A) represents a key functional group that is essential for compound activity. Addition to this functional group (Figure 5.16) promotes increased activity, but addition of the phenyl grouping in 012-149 and 012-400 result in a reduced therapeutic window between melanoma cells and fibroblasts (Figure 5.16). From the structural similarity of 22 compounds, compound 012-149 demonstrated activity that was more efficacious than parent compound 012 in combination with vemurafenib (Figure 5.15, lower panel). However, the IC50 of compound 012-149 for normal fibroblasts MRC-5 and WI-38 were determined to be 3 and 3.25 µM, respectively (Figure 5.16). This indicates a loss of therapeutic window in compound 012-149 and an increase in general toxicity that is non- specific and this compound was not investigated further. Overall, replacing the ethoxy group with a shorter substituent (Figure 5.14A) promotes increased activity in BRAFWT melanoma cells, whereas replacing the ethoxy group with a larger component (phenyl ring – Figure 5.16) promotes increased activity in melanoma cells regardless of BRAF status however has the negative effect of resulting in a reduced therapeutic window between melanoma cells and fibroblasts. Interestingly, removal of the entire 4-ethoxyphenyl ring (containing the ethoxy group -Compound 012-031) resulted in an IC50 of greater than 100 µM indicating a complete loss of activity (Figure 5.17), which highlights the importance of the phenyl ring component of the 4-ethoxyphenyl group (Figure 5.14). Compound 012-714 (Figure 5.16) also demonstrated an alteration in the diethylamino group as well as the ethoxy group; however the results were similar to other compounds in which the ethoxy group only was changed. This suggests that the diethylamino group may not be essential for activity of compound 012.

301

Figure 5.14. Compound 012 structure activity relationship (SAR) A) is a 4-ethoxyphenyl group, B) is a diethylamino group and C) is an ethoxy group (including the oxygen attaching it to the ring). Modification of these three groupings was performed using the Hit2lead program selecting 90% structural similarity to the parent compound 012 to produce a focus library of 24 compounds.

302

Figure 5.15. The twenty-two compounds from the focus library demonstrate variable cell viability to compound 012/vemurafenib combination. Mel-JD cells were screened with DMSO control, test compounds alone (4 µM), vemurafenib alone (5 µM) or the combination for 72 hours and cell viability determined by the Alamar Blue assay. Compounds were screened using parent compound 012 as a control to determine increased or decreased efficacy to the parental compound.

303

304

Figure 5.16. Enhanced anti-melanoma activity in Mel-JD cells but not MM200 cells. Structural modification of the diethylamino group and the ethoxy group was performed using the Hit2lead software to produce compounds 012-424 and 012-714 (diethylamino group modification) and compounds, 012-770, 012-803, 012-903, and 012-984 (ethoxy group modification). The IC50 was determined in Mel-JD (BRAF WT) and MM200 (BRAF MT) melanoma cells and MRC-5 and WI-38 normal fibroblasts.

305

Figure 5.17. Anti-menaloma activities in Mel-JD and MM200 cells with 3 compounds from the focus library. Structural modification of the ethoxy group and 4-ethoxyphenyl group was performed using the Hit2lead software to produce compounds, 012-400 and 012- 149 (ethoxy group modification), which have an addition of a phenyl group, and compound 012-031 (4-ethoxyphenyl group modification), which has the removal of a phenyl group. The

IC50 of these three compounds was determined in Mel-JD and MM200 melanoma cells, and MRC-5 and WI-38 normal fibroblasts by the Alamar blue assay. Loss of compound activity is showed in compound 012-031.

306

In this focus library, we have identified some structure-function activity that provides insight into the essential functional groups on small molecule compound 012. The molecular target of compound 012 is presently unknown and structural modelling based on a protein binding pocket is not possible presently. However, we have identified that loss of the 4-ethoxyphenyl group causes a complete loss of functional activity as determined by assessing IC50, whereas modification to the ethoxy group alone causes a gain of functional activity, indicating that this functional group is important for the drug action. Addition of a short substituent results in a gain of function in the BRAFWT cells, whereas a larger substituent resulted in a non-specific gain in toxicity. Of the 22 compounds in the focus library, none demonstrated increased activity towards melanoma cells without also increasing activity towards normal fibroblast controls. Hence, parental compound 012 was selected to progress into in vivo characterization.

307

5.2.4 The mechanism of action for the combination of compound 012/vemurafenib

5.2.4.1 Compound 012 decreases ERK phosphorylation after prolonged treatment

We sought to characterize key components of the MAPK pathway on treatment with combination compound 012/vemurafenib to determine if ERK phosphorylation was attenuated with combination drug treatment. We performed immunoblotting on whole cell lysate to determine the expression levels of BRAF, and COT, and phosphorylation states of MEK and ERK (Figure 5.18) after 72 hours of drug treatment in Mel-JD cells. We confirmed that vemurafenib increases MEK and ERK phosphorylation in these cells indicating the paradoxical CRAF activation by vemurafenib in BRAFWT cells. COT levels are also increased with all three treatment conditions (Figure 5.18). Interestingly, long term culture (10 days) of Mel-JD cells with compound 012 alone resulted in a decrease in ERK phosphorylation (Figure 5.18), indicating a potential for compound 012 to act as a MEK inhibitor.

308

Figure 5.18 Compound 012 decreases ERK phosphorylation with long term treatment. Mel-JD cells were treated with DMSO control, compound 012 (4 µM), vemurafenib (5 µM) or combination for 72 hours or 10 days. Immunoblotting was performed for key proteins of the MAPK pathway using anti-BRAF, anti-COT, anti-phos-MEK, anti-phos-ERK antibodies. Anti- GAPDH was used as a loading control.

309

5.2.4.2 Compounds demonstrate an increase in TRIM16 protein expression

In Chapter 3, we have shown that over-expression of TRIM16 results in a decrease in cell proliferation and viability (Figure 3.13). To test the potential for the anti-melanoma compounds to increase TRIM16 protein, we treated MM200 cells with compounds 759, 236 and 012 at IC50 concentrations with DMSO as a control. Using whole cell lysate, we determined that compound 236 significantly increases TRIM16 protein expression, while compounds 759 and 012 slightly increase TRIM16 protein expression (Figure 5.19).

310

Figure 5.19. Anti-melanoma compounds increase TRIM16 at IC50 concentrations. Melanoma cell line, MM200, was treated with anti-melanoma compounds 759, 236 and 012 at

IC50 concentrations 10 µM, 8.5 µM and 8.2 µM respectively for 72 hours. An immunoblot performed from whole cell lysate probing for anti-TRIM16 antibody and anti-β-Actin antibody, which is used as a loading control. Corresponding densitometry plot of TRIM16 protein expression as a percentage of β-Actin control. **p<0.01.

311

5.2.4.3 Compound 236 increases cytoplasmic TRIM16 protein, but decreases the level of nuclear TRIM16

As compound 236 markedly increases TRIM16 protein expression compared to DMSO control, this was a desirable compound to characterize further. We sought to determine if this increase in TRIM16 protein was in the cell nucleus, where TRIM16 is expressed in normal melanocytes and is subsequently lost in melanoma (Chapter 3, Figure 3.7). TRIM16 has also been demonstrated to inhibit cell cycle progression by nuclear localization, and has effects on cyclin D1, p27 (351) and E2F1 (350) in neuroblastoma. We treated MM200 cells with increasing doses of compound 236 up to 10 µM for 24 hours and performed extraction of the cytoplasmic and the nuclear fractionations to determine the level of TRIM16 protein in each compartment. We showed by immunoblot, that TRIM16 protein expression is increased in a dose dependant manner in the cytoplasm, but is decreased in the nucleus. This indicates that while compound 236 increases TRIM16 protein overall (Figure 5.20), it decreases nuclear TRIM16 protein, which is an undesirable action of the compound.

312

Figure 5.20. TRIM16 protein is increased in the cytoplasm and is decreased in the nucleus in a dose dependent manner with compound 236. Melanoma cells, MM200, were treated with compound 236 at 0, 0.001, 0.01, 0.1, 1 and 10 µM for 24 hours and a subcellular fractionation performed to separate the cytoplasm and nucleus. Immunoblotting was performed determining TRIM16 protein expression using a specific anti- TRIM16 antibody. Anti-GAPDH and anti-Histone-H3 were used as loading controls for the cytoplasm and nucleus, respectively.

313

5.2.4.4 TRIM16 is partially required for compound 012/vemurafenib mechanism of action

As TRIM16 reduces cell viability, and is increased by vemurafnib in a number of BRAF mutant melanoma cells (Chapter 3, Figure 3.27), we tested the expression of TRIM16 on treatment with compound 012/vemurafenib. Most interestingly, treatment with combination compound 012/vemurafenib in Mel-JD, Mel-RM and CHL-1 cells resulted in an increase in TRIM16 protein expression (Figure 5.21). To determine the significance of this increase in TRIM16 for drug action, we used TRIM16 specific siRNA’s (TRIM16-1 & TRIM16-2) and control siRNA to target TRIM16 in the presence of drug treatment (Figure 5.22A). We show a loss of reduction of cell viability induced by compound 012/vemurafenib when in the presence of targeted siRNAs. This indicates a partial requirement of TRIM16 for compound 012/vemurafenib mediated reduction of cell viability. Figure 5.22B shows the corresponding cell morphology demonstrating the expected cell death with combination treatment in the siControl cultures, and the partial reversal of cell death in siTRIM16 treated cultures (Figure 5.22B).

314

Figure 5.21. Combination compound 012/vemurafenib increases TRIM16 in BRAFWT cells. Mel-JD, Mel-RM and CHL-1 cells were treated with the above agents for 72 hours and the level of TRIM16 protein was measured using a specific anti-TRIM16 antibody. Anti- GAPDH was used as a loading control.

315

Figure 5.22A.

316

Figure 5.22B.

Figure 5.22. TRIM16 is partially required for the compound 012/VEM mechanism of action. A) Mel-JD cells were transfected with siControl, siTRIM16-1 or siTRIM16-2 for 24 hours. After 24 hours, cells were treated with DMSO control, compound 012 (4 µM), vemurafenib (VEM, 5 µM) or combination for 48 hours. Cell viability was assessed by Alamar Blue assay and expressed as % of DMSO control for each siRNA. Statistical analysis was performed between siControl combination 012/VEM and each siTRIM16 combination 012/VEM. ***p<0.001. Immunoblotting was performed for TRIM16 expression using a specific anti-TRIM16 antibody. Anti-β-Actin was used as a loading control. Corresponding densitometry plot of TRIM16 expression is shown, *p<0.05. B) Cell morphology of combination 012 + VEM treatment between siControl, siTRIM16-1 and siTRIM16-2 shows reduced cell death with TRIM16 gene silencing.

317

Activation of the RAF/MEK/ERK pathway in melanoma results in cell proliferation via phosphorylated-ERK acting on proliferative target genes. Growth factors and mitogens utilise this pathway to regulate gene expression and prevent apoptosis (478). However, induction of the MAPK pathway has also been associated with cell cycle arrest and apoptosis in some cancers (478). To strengthen this finding, hyperactivation of the MAPK pathway following induction of an oncogenic form of CRAF induced a senescence-like proliferation arrest in BRAF mutant melanoma cells (479). Corresponding reduction in the levels of phosphor-Rb and an increase in p21 expression is also seen (479). These data indicate that oncogene induced senescence is a possibility in melanoma (479) and is not only observed in benign melanocytic lesions. In addition, activation of the MAPK pathway at low levels led to cyclin D1-cdk4 and cyclin E-cdk2 complexes and subsequent cell cycle progression. High levels of Raf activation led to cell cycle arrest by the induction of p21 expression and inhibition of cyclin-cdk activity (480). This may be a rationale for the use of small molecules that induce MAPK pathway activation as a mechanism of cancer treatment. This is demonstrated in the use of Wentilactone B (WB) that is able to induce G2/M phase arrest and apoptosis in human hepatoma cells by up- regulation of the MAPK pathway (481). As our immunoblot data shows that ERK phosphorylation is increased with vemurafenib treatment, and the effect is not attenuated with compound 012 combination, vemurafenib is transactivating CRAF in these cells. An off-target effect of vemurafenib may be the mechanism of action by which the compound 012/vemurafenib combination is inducing cell death. Combination of compound 012 with paradox breaker, PLX7904, would address whether the transactivation of CRAF, and subsequent ERK phosphorylation, was required for the mechanism of action in reducing cell viability with drug combination, as PLX7904 does not transactivate CRAF in NRAS mutant cells.

318

Therapies to restore tumour suppressor gene expression are an attractive therapeutic option in the advancement of melanoma treatment (156). A prime example of this is the effort to restore p53 function to melanoma cells. Over 90% of melanomas harbour wild-type p53, which is inactivated. Efforts using small molecules to restore p53 by blocking MDM2 and cyclin B1/CDK-1 phosphorylated nuclear iASPP (412). This resulted in an additive suppression and increased apoptosis when combined with vemurafenib in vivo (412). Similarly, restoration of TRIM16 by the combination of compound 012/vemurafenib demonstrate a partial requirement of TRIM16 for compound 012/vemurafenib mechanism of action, suggesting induction of TRIM16 may enhance vemurafenib action.

5.2.4.5 Microarray analysis reveals WNT and Beta3 Integrin signalling may be required for combination compound 012/vemurafenib mechanism of action

To determine the mechanism of action of compound 012/vemurafenib, we used Mel-JD cells treated with DMSO control, compound 012 (4 µM), vemurafenib (5 µM) and combination for 6 hours, and isolated RNA for microarray analysis. Bioactivity of the agents is confirmed in a parallel culture for 72 hours with assessment of cell viability by Alamar blue (Figure 5.23A). Microarray was performed on the human Agilent 8x15K array at the Ramaciotti centre, UNSW. Candidate genes were determined using differential gene expression analysis, where significant genes has a p value <0.05 and a minimal fold change of 1.5x. Twenty-one genes were differentially regulated in the control vs. compound012/vemurafenib (CTRL-012PLX).

Determining the mechanism of action of control vs. compound 012 alone was also of interest, where 38 genes are differentially expressed. (Figure 5.23B). In the comparison of compound 012 vs. control, 22/53 significantly regulated genes are involved in cholesterol biosynthesis. Of these, 22/29 up-regulated genes are involved in cholesterol biosynthesis. Genes that are centrally involved in cholesterol biosynethesis are depicted in the cholesterol pathway (Figure 5.24). Additional up-regulated genes that support cholesterol biosynethesis include, LDLR, ABCA1, STARD4, PNPLA3, ACSS2, LPIN1, HSD17B7, PCYT2, and FDPS. TRIM16 is not significantly up-regulated in the microarray data. This may reflect the short treatment period of 6 hours before RNA was harvested for microarray analysis. TRIM16 changes may also be post- translational and not rely on a transcriptional increase with combination treatment.

319

Figure 5.23A

Figure 5.23B

320

Figure 5.23. Twenty-one genes differentially expressed in the combination of 012/vemurafenib treatment of Mel-JD cells. A) Mel-JD cells are treated with DMSO control, compound 012 (4 µM), vemurafenib (5 µM) or combination. Cell viability assay was performed to ensure agent activity against melanoma cells. RNA was harvested at 6 hours treatment for microarray analysis. Statistical analysis between single agent vemurafenib and DMSO control, and combination of compound 012/vemurafenib and DMSO control is shown. ***p<0.001. B) Differential gene expression analysis was used to determine candidate genes which might mediate compound 012/vemurafenib action. Candidate genes were determined as having a fold-change of at least 1.5 and a p value<0.05. 21 candidate genes are identified as being differentially expressed in compound 012/vemurafenib (PLX) vs. DMSO control (CTRL- 012PLX), but not by either agent alone (CTRL-PLX or CTRL-012).

321

Figure 5.24. Key components of cholesterol biosynthesis are up-regulated by compound 012. Differential gene expression was determined in Mel-JD BRAFWT cells treated with compound 012 for 6 hours and compared to control treated cells using microarray analysis. Key transcriptional changes involving cholesterol biosynthesis are highlighted with a white diamond.

322

Figure 5.25. Significantly regulated genes are involved in WNT signaling and Beta-3 integrin signaling. Pathway analysis was applied to microarray data to determine significant pathways regulated by compound 012/vemurafenib combination. Additional candidate genes, RGS4, RFC1, and NR4A2 were selected due to significant fold-change and known function.

323

Recently, cholesterol expression has been shown to play a role in melanoma development. Specifically, the precursor to cholesterol biosynethesis, 7-dehydrocholesterol (7-DHC), has a deleterious effect on cell viability and function (482). Researchers used A2058 melanoma cells to assess the effect of 7-DHC on cell viability. It was determined that reactive oxygen species (ROS) were significantly increased with 7-DHC entry into cells, and after 24 hours of treatment a significant decrease in cell viability and proliferation was observed. This was associated with an increase in pro-apoptotic Bax and a decreased in the Bcl-2/Bax ratio, a reduction in mitochondrial membrane potential, and an increase in apoptosis inducing factor (AIF) expression, suggesting apoptosis as the most likely mechanism of cell death (482). As our data shows a significant increase in the cholesterol biosynthesis pathway, including 7-DHC, with compound 012 treatment, this suggests cholesterol biosynethsis could be a mechanism by which compound 012 exerts anti-melanoma activity. To test this, assessment of ROS levels and apoptosis indicators, Bax and AIF, are to be investigated.

One of the significantly regulated pathways in the compound 012/vemurafenib treatment group compared to the control is the β-catenin/WNT signaling pathway. WNT signaling is a pro- survival pathway associated with oncogenesis. Two genes are identified as being differentially expressed, dikkopf-1 (DKK1, -2.3) and dishevelled-binding antagonist of beta-catenin 1 (DACT-1, -2.0). DKK1 is known to inhibit the invasive potential of melanoma cells and functions as a negative regulator of the β-catenin/WNT signaling pathway. DACT-1 is an inhibitor of the WNT signaling pathway (483). However, DACT1 signalling paradoxically has been identified as an oncogene in colon cancer (484). Down-regulation of these two genes is counterintuitive in this context as they function as negative regulators of β-catenin/WNT signaling. It is possible that the potential for increased WNT/β-catenin signalling in repsonse to treatment with 012/vemurafenib through down-regulation of DACT and DKK1 may contribute to oncogene-induced senescence, mediated by highly active WNT signaling.

Cystein-rich 61 (CYR61, -2.1) is highly expressed in invasive melanoma cells and is hypoxia- inducible in lowly aggressive melanomas (485) and has been identified as a pro-angiogenic factor in numerous cancers, including melanoma (486) (487). Thus, down-regulation of CYR61 may attenuate melanoma cell invasion and angiogenesis, which would be consistent with the results of our microarray analysis in treatment with the combination of 012/vemurafenib. Furthermore, CYR61 is regulated by the canonical WNT signaling pathway (488) which suggests a potential relationship between the WNT-signaling and Beta-3 integrin candidate

324

genes. Thrombospondin-1 (THBS1, -1.8) is frequently methylated in melanoma and de- repression using the de-methylating agent 5-aza-2-deoxycitidine results in the suppression of angiogenesis (489). This is thought to occur via a paracrine loop with BMP4 (490). In contrast to this, both THBS1 and CYR61 are overexpressed in esophageal squamous cell carcinoma and are all associated with poor prognosis (491). Hence, the down-regulation of THBS1 as an anti- angiogenic agent is counterintuitive to the induction of apoptosis and senescence observed functionally in Mel-JD cells. However, the role of THBS1 in tumourigenesis is unclear and to take an unbiased approach in selecting candidate genes for further characterization, THBS1 is selected as a significantly differentially expressed gene and it remains to be determined if either CYR61 or THBS1 contribute to the phenotype observed. Genes, RGS4, RFC1, and NR4A2 are selected for further investigation as genes that are significantly differentially expressed in the compound 012/vemurafenib treatment vs. control. To test the contribution of these genes, treatment of Mel-JD cells with combination compound 012/vemurafenib while performing gene silencing for RGS4, RFC1, and NR4A2 will need to be performed. Determination of a loss of decreased cell viability with combination treatment indicates the dependence on a given gene. To test the role of DKK1, DACT1, CYR61 and THBS1 in the combination, the treatment with combination of compound 012/vemurafenib and forced overexpression of the candidate gene will need to be performed to determine if the loss of decreased cell viability is seen, indicating dependence on the candidate gene.

325

5.2.4.6 Combination of compound 012 with other inhibitors for melanoma treatment

As compound 012 has demonstrated synergy with vemurafenib, one means of determining additional synergistic relationships is to combine compound 012 with other clinical anti- melanoma agents as they are known to target specific signalling pathways in melanoma. This also has potential to give insight into compound 012 mechanism of action as combinations of pathway inhibition are highly effective in melanoma treatment. For example, there is evidence to suggest that simultaneous inhibition of MEK and PI3K/AKT induce a high degree of synergy of anti-melanoma activity (104). If compound 012 is a MEK inhibitor, we may anticipate a high degree of synergy between compound 012 and PI-103, which blocks the AKT pathway.

We used Mel-JD BRAFWT and G361 BRAFMT melanoma cells to determine the potential for synergy between compound 012 and the inhibitor by treating with compound 012 at a standardized dose of 4 µM and the published IC50 dose of inhibitor. We investigate three inhibitors, sorafenib, PI-103, and rapamycin for synergy study with compound 012. This is indicated by reduced cell viability in the combination group compared to single agent alone. As sorafenib preferentially targets CRAF, when sorafenib is combined with compound 012, a reduction in cell viability is observed for Mel-JD (BRAFWT), but not G361 (BRAFV600E) (Figure 5.27), similar to vemurafenib. Rapamycin and PI-103 combination with compound 012 offered a modest additive effect with compound 012 for both cell lines. Combination with the JAK/STAT inhibitor, AZD-1480, did not offer a further reduction in cell viability when combined with compound 012 (Figure 5.27). Interestingly, combination of the mekinist, trametinib, and compound 012 does not further reduce cell viability than single agent alone, suggesting possible redundancy. As we have shown that compound 012 may target MEK (Figure 5.18), this data is in agreement with the hypothesis that compound 012 is a mekinist and addition of another MEK inhibitor does not further decrease cell viability.

326

Figure 5.26. Anti-melanoma compounds and their primary targets. Clinical agents, vemurafenib, trametinib, sorafenib, rapamycin and pre- clinical agents, PI-103, AZD-1480, and their respective target proteins and pathways.

327

Figure 5.27. Melanoma inhibitors display an additive effect in combination with compound 012. Inhibitors: PI3K inhibitor (PI-103), Sorafenib (Sor), Rapamycin (Rap), JAK1&2 inhibitor (AZD) and MEK inhibitor, Trametinib (Tra). Cell were treated for 72 hour and cell viability measured by the Alamar blue assay.

328

This brief study indicates the potential for compound 012 to act in synergy with other clinical or pre-clinical anti-melanoma agents. It also is informative in elucidating the target pathway of compound 012, with the potential that compound 012 is acting as a mekinist due to the lack of potential synergy when in combination with the MEK inhibitor, trametinib. Combination of compound 012 and the PI3K inhibitor, PI-103 may be a desirable combination. There is evidence to suggest that simultaneous inhibition of MEK and PI3K/AKT induce a degree of synergy of anti-melanoma activity (104). A full study of synergy needs to be performed to understand the potential for compound 012 to synergise with other anti-melanoma agents in addition to vemurafenib.

5.2.5 Development of three melanoma cell lines resistant to vemurafenib

We developed three BRAFV600E melanoma cell lines, Mel-CV, MM200, and A375, with resistance to vemurafenib at 10 µM. Cell lines with an IC50 greater than 10 µM to an agent are considered to be resistant to the test compound (105). All cell lines were developed using increasing doses of vemurafenib until no cell death was observed at 10 µM (Figure 5.28). Resistant cells showed altered morphology compared to parental cells (Figure 5.29A).

Confirmation of resistance to vemurafenib was determined by the reactivation of phosphor- ERK which is typical of vemurafenib resistant cells in a by-pass mechanism of vemurafenib BRAF protein targeting (258, 475). This can also be accompanied by an increase in BRAF protein expression (475). In addition, increased COT expression is detected in 2/3 melanomas that become resistant to Vemurafenib (68). In line with this, our resistant cell lines Mel-CV and MM200 show increased COT expression in resistant cells compared to parental cells (Figure 5.29B). Both of these lines also showed increase in BRAF protein expression. All lines showed a re-activation of ERK phosphorylation in resistant cells compared to parental. This was markedly different in the Mel-CV cells which showed ERK re-activation to a lesser degree than MM200 and A375 cells (Figure 5.29B). Figure 5.30 schematically represents the protein expression changes in the three resistant melanoma lines. Interestingly all lines show an increase in TRIM16 protein expression in the resistant lines. It remains to be determined if this indicates TRIM16 as a resistance factor, or if the cell populations expressing TRIM16 are abrogating TRIM16 effects on cell proliferation could due to additional mutations occurred in TRIM16 gene.

329

Figure 5.28. Development of three melanoma cell lines resistant to vemurafenib. Vemurafenib resistant cell lines, Mel-CV, and MM200 & A375, were developed over 1.5 and 3 months, respectively. Escalating doses of vemurafenib were used up to 10 µM.

330

Figure 5.29A

331

Figure 5.29B

Figure 5.29. Cell morphology and western blotting confirms ERK re-activation in vemurafenib resistant cells compared to parental cells. A) The morphology changes of resistant cells, Mel-CV, MM200, and A375 was observed by light microscopy. B) The whole cell lysate from parental and resistant cell lines, Mel-CV, MM200, and A375 was subjected to immunoblotting to investigate vemurafenib resistance mechanism. Antibodies against, BRAF, COT, phos-ERK, total-ERK and TRIM16 were used. Anti-GAPDH was used as a loading control.

332

Figure 5.30. Summary of resistance mechanism of three melanoma cell lines. In response to vemurafenib resistance, Mel-CV and MM200 exhibit increased BRAF, COT and phos-ERK protein levels. Line, A375 shows increased phos-ERK protein expression only.

333

We have successfully developed three melanoma cell lines resistant to vemurafenib and have confirmed this by the re-activation of ERK evidenced by ERK phosphorylation in the presence of 10 µM of vemurafenib. As vemurafenib resistance is a frequent clinical event, averaging 6-9 month treatment with the drug (253), development of agents that can synergise with vemurafenib and overcome resistance mechanisms are valuable. In the next section, we seek to treat resistant melanoma cells with compound 012 to determine the potential for the compound to demonstrate efficacy against vemurafenib resistant melanoma.

334

5.2.5.1 Compound 012 may demonstrate activity against vemurafenib resistant cell lines

As combination of compound 012/vemurafenib preferentially targets BRAFWT melanoma cells, we tested whether compound 012 could overcome vemurafenib resistance in MM200 cells. MM200 vemurafenib resistant cells demonstrate increased phos-ERK, increased COT and increased BRAF protein expression. If compound 012 has activity as a MEK or ERK inhibitor, it may demonstrate activity against BRAFi resistant melanoma. Interestingly, the IC50 of compound 012 in MM200 parental cells was 8.2 µM and in the resistant cells was 3.1 µM. This suggests an increased sensitivity to compound 012 in vemurafenib resistant cells. When comparing the difference between 0.5 µM vemurafenib (VEM) activity to VEM + 012 (4 µM) between parental (p=0.033) and resistant cells (***p<0.001), it appears that resistant cells are more sensitive to the presence of compound 012 than parental cells (Figure 5.31).

335

Figure 5.31. Vemurafenib resistant cells show increased sensitivity to compound 012 compared to parental cells. MM200 parental (blue) and resistant (red) cells were treated with 0.5µ M of vemurafenib (VEM) or combination 0.5 µM VEM/4 µM compound 012 (012) for 72 hours. Cell viability was measured using Alamar blue.

336

Clinical resistance to vemurafenib frequently occurs at 6-9 months of treatment. The resistance mechanisms are varied (Chapter 1, Section 1.9.1), and the heterogeneity of melanoma suggests multiple resistance mechanisms may develop within a single tumour. Testing agents that can overcome these resistance mechanisms is valuable in developing treatment options for patients who are no longer responding to frontline therapy. Compound 012 demonstrates a degree of efficacy in overcoming resistance in MM200 vemurafenib resistant cells. However, it remains to be determined if this is applicable to multiple resistant cell lines, or if compound 012 may target cell lines that acquire mutations in NRAS. As compound 012 targets BRAFWT cell lines preferentially, these typically harbour the NRAS mutation and this may be a melanoma subtype that compound 012 demonstrates greater efficacy against.

Secondary NRAS mutation is a frequent event in vemurafenib resistance with 4/19 NRAS mutations identified in patients progressing on vemurafenib (68). The paradox breaker BRAF inhibitor, PLX7904, can attenuate ERK phosphorylation in vemurafenib resistant cells that have gained the NRAS mutation as a resistance mechanism (471). As compound 012 demonstrates increased efficacy against vemurafenib resistant cells, testing of combination compound 012 and the paradox breaker PLX7904 may be effective in overcoming vemurafenib resistance.

5.2.6 Testing combination vemurafenib/compound 012 in vivo

As we have shown that compound 012 acts in synergy with vemurafenib in BRAFWT melanoma cells, we sought to determine the stability and toxicity of compound 012 in vivo in preparation for a xenograft model to study the efficacy of the compound 012/vemurafenib combination in vivo. At each stage of drug development it is important to carefully assess how amenable a compound is for further biological evaluation and potential as a cancer therapeutic. The study of compound stability in a biological context provides information on the most appropriate route of administration and whether a compound may need structural modification and re-testing before application to a biological system is attempted. In vivo toxicity studies are important to determine the potential for drug efficacy at a specific target tissue and the therapeutic index that can be achieved. If a drug is too toxic as indicated by adverse events, the achievable dose may be too low to determine efficacy against the target tissue, and may rule out the compound in its present form from further evaluation as a cancer therapeutic.

337

5.2.6.1 Determination of compound 012 stability

In order to test compound 012 as an anti-melanoma and enhancer of vemurafenib in vivo, we sought to first determine the stability of compound 012 by the in vitro microsomal stability assay. We incubated our compound at set time points with human liver microsomes. This is useful as an in vivo indicator of stability if the compound passes through the liver (492) i.e. is administered orally. We determine that the half-life of compound 012 in the microsomal stability assay was 13.5 minutes (Figure 5.32). A half-life of less than 15 minutes indicates a low microsomal stability (492).

338

Figure 5.32. Compound 012 has a half-life if 13.5 minutes by the microsomal stability assay. Compound 012 was tested for stability in the presence of human liver microsomes at 5, 15, 30, 60 and 90 minutes of incubation. Compound degradation was measure in duplicate (sample 1&2) using mass spectrometry.

339

These data indicate that compound 012 has a low stability (<15 min) in the in vitro microsomal stability assay. This suggests that either an intraperitoneal (IP) or intravenous (IV) route of administration may be a preferable option to test drug efficacy in vivo as the drug is more likely to reach the target tissue (engrafted melanoma tumour) if it does not pass through the liver. For this reason, we selected an intravenous route of administration for compound 012 for in vivo maximum tolerated dose (MTD) studies.

5.2.6.2 Maximum tolerated dose study for compound 012

To determine the maximum dose tolerated in vivo of compound 012, a dose study was undertaken in BALB/c nude mice using increasing drug concentrations. We selected female BALB/c nude mice between 5-6 weeks of age for IV administration of compound 012. Four doses were administered 5 mg/kg, 10 mg/kg, 15 mg/kg, and 20 mg/kg with three mice in each treatment group (Table 5.4). The treatment period was 5 days on, 2 days off for a two week period. Mice were monitored for body weight, indicators of stress, seizures, or uncharacteristic behavior. The study specifics are outlined in Table 5.3. The results of animal monitoring on days 10-13 of the study indicate that some adverse events (slow movement, hunching) are experienced by mice at the highest dose of 20 mg/kg on days 10 and 11 and resolve on days 12 and 13 of the study. No adverse events are noted in any of the lower dose groups at any point during the study (Table 5.5).

340

Table 5.3. Specifics of the maximum tolerated dose study for compound 012

Mouse strain BALB/c nude Gender & age Female 5-6 weeks Weight 18+/- 2 g Housing 3 mice/cage/treatment group Food ad libitum Water ad libitum Treatment solution Compound 012 in 5% DMSO/95% saline solution (0.9%) Injection volume 200μL Route of administration Intravenous (IV) Treatment frequency 5 days on/2 days off for 2 weeks Days 0-13 Experimental Body weight and daily observation for adverse event (hunching, endpoints diarrhoea, seizures, uncharacteristic behaviour or death)

Table 5.4. Treatment groups of maximum tolerated dose study

Groups (n) Treatment Completion time

1 3 20 mg/kg, i.v, 5 days on 2 days off 2 weeks

2 3 15 mg/kg, i.v, 5 days on 2 days off 2 weeks

3 3 10 mg/kg, i.v, 5 days on 2 days off 2 weeks

4 3 5 mg/kg, i.v, 5 days on 2 days off 2 weeks

341

Table 5.5. Appearance and behavior of mice with compound 012 treatment at days 10-13 Day 10 Day 11 Day 12 Day 13 Appearance and Appearance and Appearance and Appearance and weight weight weight weight moved slowly, moved slowly, Normal, Normal, 18.7 g 18.8 g 19.1 g 19.0 g Group 1: 20 mg/kg, i.v, Normal, Normal, Normal, Normal, 5 days on 2 days off 20.2 g 20.3 g 20.5 g 20.7 g moved slowly, hunched slightly, Normal, Normal, 20.1 g 20.3 g 20.3 g 20.8 g Normal, Normal, Normal, Normal, 20.4 g 20.4 g 20.6 g 20.1 g Group 2: 15 mg/kg, i.v, Normal, Normal, Normal, Normal, 5 days on 2 days off 19.8 g 19.8 g 20.0 g 20.2 g Normal, Normal, Normal, Normal, 21.3 g 20.1 g 20.2 g 20.2 g Normal, Normal, Normal, Normal, 19.8 g 20.0 g 20.3 g 20.1 g Group 3: 10 mg/kg, i.v, Normal, Normal, Normal, Normal, 5 days on 2 days off 20.6 g 20.9 g 20.8 g 21.0 g Normal, Normal, Normal, Normal, 21.0 g 20.9 g 21.2 g 20.8 g Normal, Normal, Normal, Normal, 19.8 g 20.0 g 20.4 g 20.8 g Group 4: 5 mg/kg, i.v, Normal, Normal, Normal, Normal, 5 days on 2 days off 20.2 g 20.3 g 20.1 g 20.3 g Normal, Normal, Normal, Normal, 21.3 g 21.5 g 21.6 g 22.2 g

342

Our data suggests the maximum tolerated dose is 20 mg/kg with an acceptable tolerated dose of 15 mg/kg with intravenous administration in BALB/c nude mice. This did not result in any adverse appearance or behavior in BALB/c nude mice; therefore, this dose represents a safe dosage to use further in combination studies with vemurafenib in a melanoma xenograft model. We propose that the following model would be useful to determine the in vivo efficacy of compound 012/vemurafenib (Figure 5.33). This model involves the use of a sub-cutaneous xenograft of Mel-JD cells and an IV administration of compound 012 and oral administration of vemurafenib.

343

Figure 5.33. Schematic of melanoma xenograft and treatment. BRAFWT melanoma cells are engrafted into Balb-(c) nude mice and allowed to proliferate to a standard volume of 0.25 cm3. Mice are randomized into four treatment groups of DMSO vehicle control, compound 012 (15 mg/kg), vemurafenib (75 mg/kg) or combination and treated for a two week period 5 days on, two days off.

344

5.3 Discussion

Treatment of metastasized melanoma is complex due to the heterogeneous nature of disease. Molecular sub-typing has heralded an era of targeted melanoma treatment as the field moves from systemic chemotherapies to targeted inhibitors of mutant proteins. Vemurafenib represents one such inhibitor that has demonstrated an increased median progression free survival over dacarbazine and, for the first time, increased patient overall survival. However, patients experience drug resistance at 6-9 months of treatment, highlighting the need for combination therapy to target resistant sub-populations of cells. In this chapter we have characterized a small molecule, compound 012 that synergises with vemurafenib, most unexpectedly, in BRAFWT melanoma cells.

We have evidence to support the induction of both apoptosis and senescence in compound 012/vemurafenib treated BRAFWT melanoma cells, indicating a mechanism of action that makes vemurafenib treatment applicable to NRAS mutant melanoma patients. This is important as the treatment options for patients harbouring NRAS mutant melanomas are presently limited due to the lack of drug-ability of mutant NRAS proteins and the lack of melanoma-specificity of targeting upstream farnesyl-transferase or downstream MEK of the MAPK pathway, due to dose limiting toxicity. Vemurafenib is well tolerated, though is presently not given to NRAS mutation positive patients due to the paradoxical transactivation of CRAF that may facilitate tumour proliferation (221) and the lack of clinical benefit. Partnering vemurafenib with a small molecule that can induce apoptosis in BRAFWT melanomas represents a novel mechanism of drug treatment and opens up a new treatment strategy. For this to become clinically applicable, a clear mechanism of action of drug combination must be defined and a patient sub-group that is likely to benefit from drug combination to be identified. Here we have identified a panel of BRAFWT melanoma cell lines that are susceptible to compound 012/vemurafenib combination. All BRAFWT cells tested were sensitive to the combination, with the exception of the M4405 line, the point of difference being this line is p53 null. The majority of melanoma cells (81%) harbour wild-type p53 (134), that is functionally inactivated by aberrant MDM2 activity

345

(412). Restoration of p53 function by small molecules (409) is an applicable approach to melanoma treatment as p53 functional, but degraded in the disease. It is not known if p53 is required for compound 012/vemurafenib mechanism of action, but this demonstrates a possible subgroup (functional p53) that the drug combination may target. Given that apoptosis is induced by drug combination, the combination may stabilize p53 in melanoma cells, and requires further investigation.

As vemurafenib resistance is an important clinical problem, we sought to investigate the ability for compound 012 to overcome vemurafenib resistance. Interestingly, vemurafenib resistant cells displayed a lower IC50 to compound 012 and an increased sensitivity to compound 012/vemurafenib combination. This data is intriguing and requires further investigation; however, it would also be interesting to determine if other cancer types may be sensitive to compound 012/vemurafenib combination. Colon cancer cell lines harbouring BRAFV600E mutation are largely unresponsive to vemurafenib (493). The mechanism of action of compound 012 may overcome the resistance of these BRAFV600E colon cancer cells to vemurafenib and sensitise towards treatment. If the mechanism of action of combination compound 012/vemurafenib is dependent on an off-target effect of vemurafenib, combination treatment may be applicable to other cancer types that harbour wild-type BRAF and a cancer cell panel could be screened for responsive cancer types. Given long term treatment of cells with compound 012 can repress ERK signalling, it is possible that compound 012 is a MEK or ERK inhibitor. This may make compound 012 applicable to cancer types that have MAPK pathway activation and may also explain the efficacy observed against BRAFWT cells. Furthermore, compound 012 may be useful in reducing the risk of developing squamous cell carcinomas in BRAFV600E patients being treated with vemurafenib as combination of MEK inhibitors attenuate the toxicity of BRAF inhibitors by reducing the paradoxical RAF isoform switch observed in NRAS mutation harbouring cells (77, 200). BRAFi and MEKi pairings are an attractive treatment combination that much clinical research is being directed towards (183).

Secondary malignancies induced by vemurafenib treatment include squamous cell carcinomas, RAS-mutant colorectal cancer and gastric and colonic polyps, and RAS-mutant leukaemia’s (228). This is thought to be due to that paradoxical transactivation of CRAF in 346

NRAS mutant cell types that increases MAPK pathway signalling (228) (471). The development of pan-RAF inhibitors termed ‘paradox breakers’ BRAF inhibitors is an exciting development in which the BRAF protein is effectively inhibited, while NRAS mutant cells are subject to paradoxical increased CRAF signalling, hence the term ‘paradox breaker’. One such molecule is PLX7904 (PB04) (471). For the purpose of this chapter, as we are applying a BRAF inhibitor to a BRAFWT in combination with compound 012, a BRAF inhibitor that does not transactivate CRAF will provide valuable insight into the drug combination mechanism of action. This will determine if the hyperactivation of the MAPK pathway, evidenced by ERK phosphorylation, is required for the induction of apoptosis and corresponding reduced cell viability that is observed with combination compound 012/vemurafenib.

In this Chapter we have demonstrated the in vitro efficacy of combination of compound 012 with vemurafenib preferentially targeting BRAFWT cells. We have shown the significant reduction in colony forming ability in the presence of drug combination. We have determined a mechanism of drug action which partially requires TRIM16 and we have also identified additional putative target genes for further investigation. We have determined the maximum tolerated dose of compound 012 in preparation for in vivo testing of drug efficacy and proposed a xenograft model as a relevant method to further validate the combined efficacy of compound 012/vemurafenib as a cancer therapeutic (Figure 5.32). As the efficacious does of vemurafenib is already well defined in melanoma xenograft models (494) and our study has determined an MTD of compound 012, the proposed model directly addresses the in vivo capacity and potential therapeutic relevance for the two compounds to work in synergy. Further understanding of combination drug mechanism of action will allow the development of drug combinations with greater efficacy in combination with vemurafenib, or the new paradox breaker, PB04.

347

5.4 Conclusions

We have identified a unique role for the BRAF inhibitor, vemurafenib, in the treatment of BRAFWT melanoma in combination with the novel small molecule, compound 012. We have shown that combination compound 012/vemurafenib preferentially targets BRAFWT melanomas over BRAFV600 melanoma which is unexpected as vemurafenib is thought to specifically target the mutant BRAF protein. We have determined that TRIM16 is partially required for this drug combination mechanism of action and have identified further candidate genes for investigation.

348

Chapter 6

Concluding remarks

349

6.1 General discussion

Malignant melanoma is a heterogeneous disease (236, 250, 469) that is refractory to many treatments strategies and the incidence is increasing (18). Increased understanding of the molecular pathology of this disease is informative for the development of targeted therapies. Melanoma treatment is entering an era of personalized medicine, where disease biomarkers predictive of response to therapy are an essential way of characterizing melanoma on an individual basis and tailoring a treatment strategy predicted to be most successful. In addition, the personalized medicine approach allows for the development of combination therapies targeting the specific deregulated pathways that contribute to melanomagenesis.

Migration and metastasis is the main cause of death in melanoma (44). In Chapter 3 we have demonstrated the tripartite motif protein, TRIM16, is significantly decreased in melanomagenesis, and in an independent cohort of lymph node metastasis patients, that high TRIM16 is prognostic of favourable patient outcome. Furthermore, we have shown that gene silencing of TRIM16 in normal melanocytes resulted in an increase in cell migration. These data establish a basis for further investigation of the functional role of TRIM16 in suppressing cell migration, and determination of the mechanism of TRIM16 action. Key to reducing melanoma metastasis is the understanding of the loss of metastasis suppressors that occur in the progression of disease. Metastasis requires a dissemination of cancer cells from the primary tumour and the adaptation of these cells in a new tissue microenvironment (50). We have shown that gene silencing of TRIM16 can induce epithelial to mesenchymal marker, snail, in A375 melanoma cells and it is of interest to investigate a panel of melanoma cell lines for the expression of snail with TRIM16 gene silencing. EMT is also associated with decreased E-cadherin and increased N-cadherin (495), and it would be interesting to determine if gene silencing of TRIM16 also induced changes in these proteins. As we have shown that TRIM16 is significantly decreased at the dermal invasive melanoma >1mm stage, we hypothesize that loss of TRIM16 may facilitate metastasis at the dissemination stage.

350

We have shown that TRIM16 induced reduction in cell proliferation is dependent on IFNβ1 and c-JUN, both of which are target genes of TRIM16. Transcription factor protein, c-JUN is required to bind to the IFNβ1 promoter as part of the enhanceosome complex, and is essential for gene transactivation (390). We have shown that TRIM16 also binds to the IFNβ1 promoter in the same region as c-JUN in the enhanceosome complex. From these data, we hypothesize that TRIM16 may play a central role in enhanceosome activation in modulating the essential component, c-JUN, and itself binding to the complex binding site. It would be of interest to determine if TRIM16 were a binding partner of enhanceosome substituents and if TRIM16 were essential for IFNβ1 transactivation. To address this, gene silencing of TRIM16 could be performed alongside IFNβ transactivation by RIG1 or dsRNA (496). We have proposed a novel mechanism of action of TRIM16 in reducing melanoma migration and metastasis and that loss of TRIM16 contributes to poor patient prognosis and increased proliferation and metastasis. These data also suggest a potential for a novel combination of IFNβ1 therapy in conjunction with TRIM16 activation, such as vemurafenib, to reinforce the TRIM16/IFNβ1 anti-cancer signal. TRIM family members play an extensive role in innate immunity. As TRIM16 is known to bind to and enhance the secretion of IL-β1, and associates with the NALP1 inflammasome (304), understanding the general role of TRIM16 transactivation of IFNβ in response to viral infection is important as the mechanism elucidated in this thesis may have a broader application than the relevance to melanoma development and suppression. Furthermore, determining the relationship between the two proteins is of interest. We have shown a correlation between the loss of TRIM16 and IFNβ1, which reach significance at the same clinical stage of dermal invasive melanoma >1mm. In addition, gene silencing of IFNβ1 results in a decrease in TRIM16 protein. This suggests a close regulatory relationship between the proteins and a potential feedback loop of TRIM16 transactivation of IFNβ1, and IFNβ1 potentially stabilizing TRIM16. It is possible the IFNβ1 and TRIM16 are binding partners. It remains to be determined if loss of TRIM16 or IFNβ1 in melanomagenesis is the cause of the loss of the other protein. The mechanism of IFNβ1 loss requires further investigation since this study though the cell line panel did not reflect the significant loss observed in patient samples. However, using a number of normal melanocyte lines may show a higher

351

expression level of IFNβ1 in the normal melanocytes and a lower expression in the melanoma cell lines.

We have shown that TRIM16 protein increases in cell lines in response to vemurafenib treatment in a dose dependent manner and is partially required for vemurafenib mechanism of action. In the tissue samples of the patients on BRAF inhibitor therapy, we have shown a significant increase in TRIM16 protein expression and TRIM16 protein expression was decreased in patients that had progressed on therapy. Hence, we have confirmed in vitro and in vivo, that TRIM16 protein increases in response to BRAF inhibitor treatment and is partially required for drug mechanism of action in vitro. This raises the interesting possibility that vemurafenib may act with partial dependence on TRIM16 in vivo. Firstly, more patient samples in a larger cohort are required to establish the increase in TRIM16 protein expression with BRAF inhibitor treatment. As BRAF inhibitor treatment was only recently approved for BRAF mutant melanoma treatment, these samples are difficult to obtain at present. To address whether TRIM16 is required for vemurafenib action in vivo, an experiment using ex-vivo patient BRAFV600E tumours gene silenced for TRIM16 and treated with vemurafenib will determine if TRIM16 is required for vemurafenib reduction in cell viability in these samples. This also opens up the possibility for compounds that modulate TRIM16 expression to potentially work in synergy with vemurafenib as the presence of TRIM16 may facilitate vemurafenib activity and result in greater efficacy.

In Chapter 4 we sought to test the tumour suppressive potential of TRIM16 in the development of both squamous cell carcinoma, and melanoma. As our group has previously shown that TRIM16 is decreased in the progression from normal skin to SCC and that TRIM16 reduces SCC cell migration in vitro, we sought to deterime the function of TRIM16 as a tumour suppressor in SCC in vivo. To address this, we used both a conditional keratinocyte TRIM16 knockout mouse model and a full-tissue TRIM16 knockout model and topically applied chemical carcinogenesis, known to induce both SCC and melanoma. We determined the heterozygous skin knockout mice developed more numerous papilloma with reduced latency, whilst homozygous mice developed larger melanocytic lesions in the skin of the skin-specific knockout mice. This suggests that TRIM16 may play a role as a tumour suppressor protein in SCC development. However, 352

the small mouse number was a limitation in this study to determine if the homozygous TRIM16 knockout mice also displayed increased papilloma development and reduced latency. The development of large cutaneous melanocytic lesions in the homozygous knockout mice suggests an indirect effect of TRIM16 loss in signaling to the adjacent melanocytes. This increase in melanocytic lesion size may be indicative of an increase radial melanocyte migration with TRIM16 loss, or an increase in melanocyte proliferation. Further investigation is required to characterize the signaling between keratinocytes and melanocytes in the skin, particularly the effect of TRIM16 secretion by keratinocytes on melanocyte migration and proliferation.

In the full-tissue knockout mice, we determined that homozygous mice were protected from SCC development, suggesting TRIM16 may act as an oncogene in certain context, or, that a gene compensation effect is occurring with TRIM16 knockout that ablates the development of SCC with carcinogen treatment. This is an intriguing result as we hypothesized that loss of TRIM16 would result in reduced SCC latency and increase SCC development. To address the possibility of gene compensation with TRIM16 loss, changes in gene expression of TRIM family members could be determined by tissue sampling of both normal (baseline TRIM family gene expression) and carcinogen treated tissues from the three genotypes, followed by a TRIM family specific PCR array since a likelihood of affected genes could be from the genes from the TRIM family. Candidate TRIM genes that exhibit significant increase in gene expression in the TRIM16 knockout tissue samples can be validated in vitro by TRIM16 gene silencing and/or add-back of the candidate TRIM gene proposed to compensate for TRIM16 using cell proliferation assays as the functional determinant. An unbiased approach could involve the use of microarray to determine significantly altered pathways or genes with TRIM16 knockout, followed by functional validation.

In Chapter 5, we have proposed a novel compound that works in synergy with vemurafenib, surprisingly, in BRAFWT melanomas, a mechanism which is also partially dependent on TRIM16. This is important as the treatment options available to melanoma patients harbouring the NRAS mutation are limited at present. Further elucidation of the mechanism

353

of action of the novel compound may provide a means to overcome drug resistance to the patients that have gained the NRAS mutation during vemurafenib treatment.

Vemurafenib has off-target effects (249, 465), a property that may prove to be important in elucidating the compound 012/vemurafenib mechanism of action. It has been shown that an unidentified subset of BRAFWT ex-vivo tumour samples respond to vemurafenib in like- manner to the BRAFV600 counterparts (249), highlighting the potential for vemurafenib therapy to benefit a group of patients that lack the target therapy. A more detailed study of the combination of compound 012 with other inhibitors of relevant pathways that are deregulated in melanoma may provide further insight into the small molecular mechanism of action. Next-generation pan-RAF inhibitors and ERK inhibitors are currently being developed and may enhance the outcome of the patients treated with the current clinical BRAFi’s, vemurafenib and dabrafenib (497). BRAFi alone results in paradoxical activation of MAPK signaling in NRAS melanoma patients (66) (64). Pan-RAF inhibitors offer a ‘paradox breaker’ solution in targeting CRAF in conjunction to BRAF. Importantly, the development of ERK inhibitors offers a therapy that targets NRAS mutant melanomas. There is potential for the combination of ERK inhibitors and PI3K inhibitors to be highly effective for NRAS melanoma patients as these two inhibitors target the two major nodes that are deregulated in NRAS melanoma patients. ERK inhibitors are being developed in combination with BRAFi, which may eliminate paradoxical oncogenesis (497). Though the mechanism of compound 012 is presently unknown, there is potential that this compound is acting in synergy with BRAFi, vemurafenib, in a manner that negates the paradoxical oncogenesis that is anticipated with BRAFi treatment in NRAS melanoma. Though it is evident that ERK activation is increased with compound 012/vemurafenib combination, it is possibly that a downstream effector of this pathway activation is interrupted and subverted from inducing cell proliferation, as a reduction in cell viability is actually observed.

The next phase of targeted inhibitors for the treatment of metastatic melanoma is swiftly approaching and offer hope for melanoma patients by applying the most efficacious treatment strategy to individual profile. The rapid accumulation of patient information and sub-grouping from the identification of new tumour suppressors and oncogenes alongside 354

the key de-regulated nodes in melanoma are valuable in this process and offer insight into targeting this heterogeneous malignancy.

6.2 Future perspectives

Melanoma treatment is moving from chemotherapy based strategies towards targeted treatment and personalized medicine. Here we have established TRIM16 as a potential metastasis suppressor in melanoma and high TRIM16 expression as a prognostic marker of favourable patient outcome. Detailed understanding of the molecular pathology of melanoma and the increasing sub-groups is essential in establishing subsets of patients that respond to a given therapy. We have shown that the breakthough clinical BRAF targeted drugs, vemurafenib and dabrafenib, increases TRIM16 protein patients receiving BRAF inhibitor therapy. We propose that patients expressing low levels of TRIM16 may benefit for BRAF inhibitor therapy and this may reduce tumour metastasis and hence, improve overall survival. As BRAF inhibitor treated patient samples become more readily available, collection of a large cohort of these samples and grading of the level of TRIM16 expression at pre-treatment, during the treatment with BRAF inhibitor, and post-treatment with overall survival data will definitively determine whether low TRIM16 expressing patients can be more benefit from the inhibitor therapy.

For the in vivo studies of TRIM16 as a candidate tumour suppressor protein, ongoing research should focus on the completion of PCR or microarray studies of tissues from the two-stage skin carcinogenesis in TRIM16 knockout mice to determine if a gene compensation effect is occurring with TRIM16 knockout that abrogates SCC development. Future research should also include performing immunohistochemistry studies to further characterize the secretion of TRIM16 and IFNβ1 in the skin tissues from the skin-specific knockout mice to gain understanding of the signaling from the interaction between keratinocytes and melanocytes. To complete the elucidation of the role of loss TRIM16 in melanoma development, develop of a BRAF/PTEN overexpression/knockout model with a TRIM16 melanocyte specific knockout mouse will be required. Our present data suggest 355

that the melanomas arising from this animal model crossed with loss of TRIM16 may result in a higher rate of melanoma cell migration and metastasis. This experiment may establish the loss of TRIM16 as a facilitator of metastasis and this data may also be correlated with the human patient samples, which indicated that the loss of TRIM16 is associated with metastatic progression.

In this current study, we have identified and characterized a small molecule compound 012 and a novel drug combination, compound 012 and vemurafenib that surprisingly targets BRAF wild-type melanoma cells. To demonstrate the in vivo efficacy of the drug combination, further study by a xenograft model using BRAF wild-type melanoma cells, treated with single agent or vemurafenib and compound 012 combination, is required. This experiment will determine the effectiveness of this combination in vivo. Delineation of the molecular mechanism of drug combination will be also valuable for determination of the applicability of vemurafenib to BRAF wild-type melanomas. Presently vemurafenib treatment is not an option for BRAF wild-type patients due to RAF isoform switching. Future research should also focus on elucidating the mechanism of action based on the candidate genes obtained from microarray analysis of the drug combination. From this point, specific inhibitors of candidate pathways may be tested in combination with vemurafenib to determine a drug combination with greater efficacy than compound 012 and vemurafenib. The in vivo efficacy of this combination will need to be trialed in a xenograft model.

6.3 Conclusions

In this thesis and for the first time, we have characterized TRIM16 as a putative tumour suppressor that warrants further investigation as a metastasis suppressor and prognostic marker in melanoma. We have implicated TRIM16 in the mechanism of action of the frontline treatment, vemurafenib, which increases TRIM16 protein expression, suggesting vemurafenib may partially require TRIM16 to mediate anti-cancer activity in vivo. We have determined a small molecule, compound 012, as an enhancer of vemurafenib, and 356

surprisingly have strong activity in BRAF wild-type melanoma cells. This opens up the potential for vemurafenib therapy to be efficacious in BRAF wild-type melanoma patients in the presence of small molecule enhancers after the molecular mechanisms are elucidated, giving a novel treatment option to a patient subgroup presently lacking therapeutic options. We have also determined TRIM16 to be partially required for the drug combination mechanism of action. We have demonstrated that TRIM16 skin-specific knockout mice develop larger melanocytic lesions, but TRIM16 full-tissue knockout mice are protected from SCC development, indicating TRIM16 may act as tumour promoter in SCC, or a gene compensation effect has occurred with TRIM16 total knockout mice.

357

References

1. Thomas A, Erickson C. The making of a melanocyte: the specification of melanoblasts from the neural crest. Pigment Cell Melanoma Research 2008;21:598-610. 2. Sommer L. Generation of melanocytes from neural crest cells. Pigment Cell Melanoma Research 2011;24:411-421. 3. Kawakami A, Fisher D. Key discoveries in melanocyte development. Journal of Investigative Dermatology 2011;131:E2-E4. 4. Theveneau E, Mayor R. Neural crest delamination and migration: from epithelium-to- mesenchyme transition to collective cell migration. Developmental Biology 2012;388:34-54. 5. Thomas A, Erickson C. FOXD3 regulates the lineage switch between neural crest- derived glial cells and pigment cells be repressing MITF through a non-canonical mechanism. Development 2009;136:1849-1858. 6. Lindsay C, Lawn S, Campbell A, et al. P-rex1 is required for efficient melanoblast migration and melanoma metastasis. Nature Communications 2011;2:1-9. 7. Levy C, Khaled M, Fisher D. MITF: master regulator of melanocyte development and melanoma oncogene. Trends in Molecular Medicine 2006;12(9):406-414. 8. Bandarchi B, Jabbari C, Vedadi A, et al. Molecular biology of normal melanocytes and melanoma cells. Journal of Clinical Pathology 2013;66:644-648. 9. Kulesa P, Bailey C, Kasemeier-Kulesa J, et al. Cranial neural crest migration: new rules for an old road. Developmental Biology 2010;344(2):543-554. 10. Chudnovsky Y, Khavari P, Adams A. Melanoma genetics and the development of rational therapeutics. The Journal of Clinical Investigation 2005;115(4):813-824. 11. Mar V, Wong S, Li J, et al. BRAF/NRAS wild-type melanomas have a high mutation load correlating with histological and molecular signatures of UV damage. Clinical Cancer Research 2013;19(17):4589-4598. 12. Krauthammer M, Kong Y, Ha B, et al. Exome sequencing identifies recurrent somatic RAC1 mutations in melanoma. Nature Genetics 2012;44(9):1006-1017. 13. Santiago-Walker A, Li L, Haass N, et al. Melanocytes: from morphology to application. Skin pharmacology and physiology 2009;22:114-121. 14. Garbe C, Blum A. Epidemiology of cutaneous melanoma in Germany and worldwide. Skin pharmacology and applied skin physiology 2001;14(5):280-290.

358

15. Garbe C, Leiter U. Melanoma epidemiology and trends. Clinics in Dermatology 2009;27:3-9. 16. Thompson J, Scolyer R, Kefford R. Cutaneous melanoma in the era of molecular profiling. The Lancet 2009;374:362-365. 17. Emmert S, Kraemer K. Do not underestimate nucleotide excision repair: it predicts not only melanoma risk but also survival outcome. Journal of Investigative Dermatology 2013;133:1713-1717. 18. Reed K, Brewer J, Lohse C, et al. Increasing incidence of melanoma among young adults: an epidemiological study in Olmsted County, Minnesota. Mayo Clinic Proceedings 2012;87(4):328-334. 19. Marashi-Pour S, Morrell S, Cooke-Yarborough C, et al. Competing risk analysis of mortality from invasive cutaneous melanoma in New South Wales: a population- based study, 1988-2007. Australian and New Zealand Journal of Public Health 2012;36(5):441-445. 20. Gamba C, Clarke C, Keegan T, et al. Melanoma survival disadvantage in young, non- hispanic white males compared with females. JAMA Dermatology 2013;149(8):912- 920. 21. Fisher D, Geller A. Disproportionate burden of melanoma mortality in young US men. JAMA Dermatology 2013;149(8):903-904. 22. Green A, Baade P, Coory M, et al. Population-based 20-year survival among people diagnosed with thin melanomas in Queensland, Australia. The Journal of Clinical Oncology 2012;30(13):1462-1467. 23. Katalinic A, Kunze U, Schafer T. Epidemiology of cutaneous melanoma and non- melanoma skin cancer in Schleswig-Holstein, Germany: incidence, clinical subtypes, tumour stages and localization (epidemiology of skin cancer). British Journal of Dermatology 2003;149(6):1200-1206. 24. Curado M, Edwards B, Shin H, et al. Cancer incidence in five continents, vol. IX. edition. IARC Scientific Publications, Lyon 2007;160. 25. Smalley K, McArthur G. The current state of targeted therapy in melanoma: this time it's personal. Seminars in Oncology 2012;39(2):204-214. 26. Gershenwald J, Ross M. Sentinel-lymph-node biopsy for cutaneous melanoma. New England Journal of Medicine 2011;364(18):1738-1745. 27. Jemal A, Siegel R, Ward E, et al. Cancer statistics, 2008. CA: Cancer Journal for Clinicians 2008;58(2):71-96. 28. Chen J, Chi M, Chen C, et al. Obesity and melanoma: possible molecular links. Journal of Cellular Biochemistry 2013;114(9):1955-1961.

359

29. Sambandan D, Ratner D. Sunscreens: an overview and update. Journal of the American Academy or Dermatology 2011;64(4):748-758. 30. Okamoto M, Liu W, Luo Y, et al. Constitutively active inflammasome in human melanoma cells mediating autoinflammation via caspase-1 processing and secretion of interleukin-1beta. The Journal of Biological Chemistry 2010;285(9):6477-6488. 31. Feldmeyer L, Keller M, Niklaus G, et al. The inflammasome mediates UVB-induced activation and secretion of interleukin-1beta by keratinocytes. Current Biology 2007;17:1140-1145. 32. Faustin B, Reed J. Sunburned skin activates inflammasomes. TRENDS in Cell Biology 2008;18(1):4-8. 33. Ogura Y, Sutterwala F, Flavell R. The inflammasome; first line of the immune response to cell stress. Cell 2006;126:659-662. 34. Rathinam V, Vanaja S, Fitzgerald K. Regulation of inflammasome signaling. Nature Immunology 2012;13(4):333-342. 35. MacKie R, Hauschild A, Eggermont A. Epidemiology of invasive cutaneous melanoma. Annals of Oncology 2009;20:1-7. 36. Bandarchi B, Ma L, Navab R, et al. From melanocyte to metastatic malignant melanoma. Dermatology Research and Practice 2010;2010:1-8. 37. Weatherhead S, Haniffa M, Lawrence C. Melanomas arising from naevi and de novo melanomas- does origin matter? British Journal of Dermatology 2007;156(1):72-76. 38. Lee J, Li L, Brafford P, et al. PLX403, a potent inhibitor of the B-Raf V600E oncogene, selectively inhibits V600E-positive melanomas. Pigment Cell Melanoma Research 2010;23:820-827. 39. Balch C, Gershenwald J, Soong S, et al. Final version of 2009 AJCC melanoma staging and classification. Journal of Clinical Oncology 2009;27(36):6199-6206. 40. Breslow A. Thickness, cross-sectional areas and depth of invasion in the prognosis of cutaneous melanoma. Annuals of Surgery 1970;172(5):902-908. 41. Marghoob A, Koenig K, Bittencourt F, et al. Breslow thickness and clark level in melanoma. Cancer 2000;88(3):589-595. 42. Balch C, Soong S, Gershenwald J, et al. Prognostic factors analysis of 17,600 melanoma patients: validation of the american joint committee on cancer melanoma staging system. Journal of Clinical Oncology 2001;19(16):3622-3634. 43. Therasse P, Arbuck S, Eisenhauer E, et al. New Guidelines to evaluate the response to treatment in solid tumours. Journal of the National Cancer Institute 2000;92(3):205- 216.

360

44. Zbytek B, Carlson J, Granese J, et al. Current concepts of metastasis in melanoma. Expert Reviews in Dermatology 2008;3(5):569-585. 45. Finn L, Markovic S, Joseph R. Therapy for metastatic melanoma: the past, present , and future. BMC medicine 2012;10(23):1-10. 46. Morton D, Cochran A, Thompson J, et al. Sentinel node biopsy for early-stage melanoma. Annals of Surgery 2005;242(3):302-313. 47. Tsao H, Atkins M, Sober A. Management of cutaneous melanoma. New England Journal of Medicine 2004;351(10):998-1012. 48. Miller A, Mihm Jr M. Melanoma. New England Journal of Medicine 2006;355(1):51- 65. 49. Hanahan D, Weinberg R. The hallmarks of cancer. Cell 2000;100:57-70. 50. Hanahan D, Weinberg R. Hallmarks of cancer: the next generation. Cell 2011;144:646- 674. 51. Law M, MacGregor S, Hayward N. Melanoma genetics: recent findings take us beyond well-traveled pathways. Journal of Investigative Dermatology 2012;132:1763-1774. 52. Zimmer L, Hillen U, Livingstone E, et al. Atypical melanocytic proliferations and new primary melanomas in patients with advanced melanoma undergoing selective BRAF inhibition. Journal of Clinical Oncology 2012;30(19):2375-2383. 53. Haass N, Smalley K, Li L, et al. Adhesion, migration and communication in melanocytes and melanoma. Pigment Cell Research 2005;18:150-159. 54. Johansson C, Azimi A, Stolt M, et al. Association of MITF and other melanosome- related proteins with chemoresistance in melanoma tumors and cell lines. Melanoma Research 2013;23(5):360. 55. Mirmohammadsadegh A, Mota R, Gustrau A, et al. ERK1/2 is highly phosphorylated in melanoma metastases and protects melanoma cells from cisplatin-mediated apoptosis. Journal of Investigative Dermatology 2007;127:2207-2215. 56. Kyrgidis A, Tzellos T, Triaridis S. Melanoma: stem cells, sun exposure and hallmarks for carcinogenesis, molecular concepts and future clinical implications. Journal of Carcinogenesis 2010;9(3):1477-3163.62141. 57. Belden S, Flaherty K. MEK and RAF inhibitors for BRAF-mutated cancers. Expert Reviews in Molecular Medicine 2012;14:1-11. 58. Lister J, Capper A, Zeng Z, et al. A conditional zebrafish MITF mutation reveals MITF levels are critical for melanoma promotion versus regression in vivo. Journal of Investigative Dermatology 2013;134(1):133-140.

361

59. Yokoyama S, Woods S, Boyle G, et al. A novel recurrent mutation in MITF predisposes to familial and sporadic melanoma. Nature 2011;480:99-103. 60. Dhillon A, Hagan S, Rath O, et al. MAP kinase signalling pathways in cancer. Oncogene 2007;26:3279-3290. 61. Poulikakos P, Zhang C, Bollag G, et al. RAF inhibitors transactivate RAF dimers and ERK signalling in cells with wild-type BRAF. Nature 2010;464:427-431. 62. Kaplan F, Mastrangelo M, Aplin A. The wrath of RAFs: rogue behavior of B-RAF kinase inhibitors. Journal of Investigative Dermatology 2010;130:2669-2671. 63. Corcoran R, Settleman J, Engelman J. Potential therapeutic strategies to overcome acquired resistance to BRAF or MEK inhibitors in BRAF mutant cancers. Oncotarget 2011;2:336-346. 64. Marquette A, Andre J, Bagot M, et al. ERK and PDE4 cooperate to induce RAF isoform switching in melanoma. Nature Structural & Molecular Biology 2011;18(5):584-591. 65. Fedorenko I, Gibney G, Smalley K. NRAS mutant melanoma: biological behavior and future strategies for therapeutic management. Oncogene 2013;32(25):3009-3018. 66. Dumaz N. Mechanism of RAF isoform switching induced by oncogenic RAS in melanoma. Small GTPases 2011;2(5):289-292. 67. Wellbrock C, Rana S, Paterson H, et al. Oncogenic BRAF regulates melanoma proliferation through the lineage specific factor MITF. Plos One 2008;3(7):1-9. 68. Poulikakos P, Rosen N. Mutant BRAF melanoma- dependence and resistance. Cancer Cell 2011;19:11-15. 69. Tschandl P, Berghoff A, Preusser M, et al. NRAS and BRAF mutations in melanoma- associated nevi and uninvolved nevi. Plos One 2013;8(7):1-8. 70. Vredeveld L, Possik P, Smit M, et al. Abrogation of BRAFV600E-induced senescence by PI3K pathway activation contributes to melanomagenesis. Genes & Development 2012;26:1055-1069. 71. Haferkamp S, Rizos H. Oncogene-induced senescence pathways in melanocytes. Cell Cycle 2010;9(24):4778-4779. 72. Cichowski K, Janne P. Inhibitors that activate. Nature 2010;464:358-359. 73. Ma J, Han H, Liu D, et al. HER2 as a promising target for cytotoxic T cells in human melanoma therapy. Plos One 2013;8(8). 74. Solit D, Garraway L, Pratilas C, et al. BRAF mutation predicts sensitivity to MEK inhibition. Nature 2006;439:358-362.

362

75. Aplin A, Kaplan F, Shao Y. Mechanisms of resistance to RAF inhibitors in melanoma. Journal of Investigative Dermatology 2011;131:1817-1820. 76. Flaherty K, Hodi F, Fisher D. From genes to drugs: targeted strategies for melanoma. Nature Reviews Cancer 2012;12:349-361. 77. Dumaz N, Hayward R, Martin J, et al. In melanoma, RAS mutations are accompanied by switching signaling from BRAF to CRAF and disrupted cyclic AMP signaling. Cancer Research 2006;66(19):9483-9491. 78. Paraiso K, Fedorenko I, Cantini L, et al. Recovery of phospho-ERK activity allows melanoma cells to escape from BRAF inhibitor therapy. British Journal of Cancer 2010;102:1724-1730. 79. Smalley K, Xiao M, Villanueva J, et al. CRAF inhibition induces apoptosis in melanoma cells with non-V600E BRAF mutations. Oncogene 2009;28:85-94. 80. Orgaz J, Sanz-Moreno V. Emerging molecular targets in melanoma invasion and metastasis. Pigment Cell Melanoma Research 2012;26:39-57. 81. Arozarena I, Sanchez-Laorden B, Packer L, et al. Oncogenic BRAF induces melanoma cell invasion by downregulating the cGMP-specific phosphodiesterase PDE5A. Cancer Cell 2011;19:45-57. 82. Shi H, Kong X, Ribas A, et al. Combinatorial treatments that overcome PDGFRbeta- driven resistance of melanoma cells to V600EB-RAF inhibition. Cancer Research 2011;71(15):5067-5074. 83. Arkenau H, Kefford R, Long G. Targeting BRAF for patients with melanoma. British Journal of Cancer 2011;104:392-398. 84. Johnson D, Sosman J. Update on the targeted therapy of melanoma. Current Treatment Options in Oncology 2013;14:280-292. 85. Shull A, Latham-Schwark A, Ramasamy P, et al. Novel somatic mutations to PI3K pathway genes in metastatic melanoma. Plos One 2012;7(8):1-10. 86. Palmieri G, Capone M, Ascierto M, et al. Main roads to melanoma. Journal of Translational Medicine 2009;7(86):1479-5876-7-86. 87. Luke J, Hodi F. Vemurafenib and BRAF inhibition: a new class of treatment for metastatic melanoma. Clinical Cancer Research 2011;18(1):9-14. 88. Ji Z, Flaherty K, Tsao H. Targeting the RAS pathway in melanoma. Trends in molecular medicine 2012;18(1):27-35. 89. Bucheit A, Syklawer E, Jakob J, et al. Clinical characteristics and outcomes with specific BRAF and NRAS mutations in patients with metastatic melanoma. Cancer 2013;119(21):3821-3829.

363

90. Wan P, Garnett M, Roe S, et al. Mechanism of activation of the RAF-ERK signaling pathway by oncogenic mutations of B-RAF. Cell 2004;116:855-867. 91. Dankort D, Curley D, Cartlidge R, et al. BrafV600E cooperates with Pten loss to induce metastatic melanoma. Nature Genetics 2009;41(5):544-552. 92. Dhawan P, Singh A, Ellis D, et al. Constituitive activation of Akt/protein kinase B in melanoma leads to up-regulation of nuclear factor-kB and tumour progression. Cancer Research 2002;62:7335-7342. 93. Shao Y, Aplin A. Akt3-mediated resistance to apoptosis in BRAF-targeted melanoma cells. Cancer Research 2010;70(16):6670-6681. 94. Cagnol S, Chambard J. ERK and cell death: mechanisms of ERK-induced cell death- apoptosis, autophagy and senescence. FEBS Journal 2010;277:2-21. 95. Maddodi N, Huang W, Havighurst T, et al. Induction of autophagy and inhibition of melanoma growth in vitro and in vivo by hyperactivation of oncogenic BRAF. Journal of Investigative Dermatology 2010;130:1657-1667. 96. Long G, Menzies A, Nagrial A, et al. Prognostic and clinicopathologic associations of oncogenic BRAF in metastatic melanoma. Journal of Clinical Oncology 2011;29(10):1239-1246. 97. Liu W, Kelly J, Trivett M, et al. Distinct clinical and pathological features are associated with BRAFT1799A(V600E) mutation in primary melanoma. Journal of Investigative Dermatology 2007;127:900-905. 98. Hacker E, Nagore E, Cerroni L, et al. NRAS and BRAF mutations in cutaneous melanoma and the association with MC1R genotype: findings from spanish and austrian populations. Journal of Investigative Dermatology 2013;133:1027-1033. 99. Cust A, Goumas C, Vuong K, et al. MC1R genotype as a predictor of early-onset melanoma, compared with self-reported and physician-measured traditional risk factors: an Australian case-control-family study. BMC Cancer 2013;13(406). 100. Flaherty K. Is it good or bad to find a BRAF mutation? Journal of Clinical Oncology Editorials 2011;29(10):1229-1230. 101. Albino A, LeStrange R. Transforming ras genes from human melanoma: a manifestation of tumour heterogeneity? Nature 1984;308:69-72. 102. Devitt B, Liu W, Salemi R, et al. Clinical outcome and pathological features associated with NRAS mutation in cutaneous melanoma. Pigment Cell Research 2011;24:666-672. 103. Milagre C, Dhomen N, Geyer F, et al. A mouse model of melanoma driven by oncogenic KRAS. Cancer Research 2010;70(13):5549-5557.

364

104. Smalley K, Flaherty K. Integrating BRAK/MEK inhibitors into combination therapy for melanoma. British Journal of Cancer 2009;100:431-435. 105. Atefi M, Euw E, Attar N, et al. Reversing melanoma cross-resistance to BRAF and MEK inhibitors by co-targeting the AKT/mTOR pathway. Plos One 2011;6(12):1-12. 106. Jonsson A, Tuominen R, Grafstrom E, et al. High frequency of p16 (INK4A) promoter methylation in NRAS-mutated cutaneous melanoma. Journal of Investigative Dermatology 2010;130:2809-2817. 107. Birkeland E, Busch C, Berge E, et al. Low BRAF and NRAS expression levels are associated with clinical benefit from DTIC therapy and prognosis in metastatic melanoma. Clinical & Experimental Metastasis 2013;30(7):867-876. 108. Jakob J, Bassett R, Ng C, et al. NRAS mutation status is an independant prognostic factor in metastatic melanoma. Cancer 2012;118:4014-4023. 109. Mann G, Pupo G, Campain A, et al. BRAF mutation, NRAS mutation, and the absence of an immune-related expressed gene profile predict poor outcome in patients with stage III melanoma. Journal of Investigative Dermatology 2013;133(2):509-517. 110. Dai D, Martinka, M., Li, G. Prognostic significance of activated Akt expressions in melanoma: A clinicopathologic study of 292 cases. Journal of Clinical Oncology 2005;23(7):1473-1482. 111. Hennessy B, Smith D, Ram P, et al. Exploiting the PI3K/AKT pathway for cancer drug discovery. Nature Reviews Drug Discovery 2005;4(12):988-1004. 112. Rodon J, Dienstmann R, Serra V, et al. Development of PI3K inhibitors: lessons learned from early clinical trials. Nature reviews Clinical Oncology 2013;10:143-153. 113. Courtney K, Corcoran R, Engelman J. The PI3K pathway as a drug target in human cancer. Journal of Clinical Oncology 2010;28(6):1075-1083. 114. Kwong L, Davies M. Navigating the therapeutic complexity of PI3K pathway inhibition in melanoma. Clinical Cancer Research 2013;19(19):5310-5319. 115. Vidwans S, Flaherty K, Fisher D, et al. A melanoma molecular disease model. Plos One 2011;6(3):1-10. 116. Haluska F, Tsao H, Wu H, et al. Genetic alterations in signaling pathways in melanoma. Clinical Cancer Research 2006;12(7):2301-2307. 117. Hodis E, Watson I, Kryukov G, et al. A landscape of driver mutations in melanoma. Cell 2012;150:251-263. 118. Dhomen N, Reis-Filho J, da Rocha Dias S, et al. Oncogenic BRAF induces melanocyte senescence and melanoma in mice. Cancer Cell 2009;15:294-303.

365

119. Olenchock B, Vander Heiden M. Pyruvate as a pivot point for oncogene-induced senescence. Cell 2013;153:1429-1430. 120. Kaplon J, Zheng L, Meissl K, et al. A key role for mitochondrial gatekeeper pyruvate dehydrogenase in oncogene-induced senescence. Nature 2013;498:109-112. 121. Larsson L. Oncogene- and tumor suppressor gene-mediated suppression of cellular senescence. Seminars in Cancer Biology 2011;21(6):367-376. 122. Meyle K, Guldberg P. Genetic risk factors for melanoma. Human Genetics 2009;126:499-510. 123. Sheppard K, McArthur G. The cell-cycle regulator CDK4: an emerging therapeutic target in melanoma. Clinical Cancer Research 2013;19(19):5320-5328. 124. Puntervoll H, Yang X, Vetti H, et al. Melanoma prone families with CDK4 germline mutation: phenotypic profile and associations with MC1R varients. Journal of Medical Genetics 2012;0:1-7. 125. Sotillo R, Garcia J, Ortega S, et al. Invasive melanoma in CDK4-targeted mice. PNAS 2001;98(23):13312-13317. 126. Luo C, Sheng J, Hu M, et al. Loss of ARF sensitizes transgenic BRAFV600E mice to UV-induced melanoma via suppression of XPC. Cancer Research 2013;74(14):4337- 4348. 127. Avery-Kiejda K, Bowden N, Croft A, et al. P53 in human melanoma fails to regulate target genes associated with apoptosis and the cell cycle and may contribute to proliferation. BMC Cancer 2011;11(203):1-17. 128. Gembarska A, Luciani F, Fedele C, et al. MDM4 is a key therapeutic target in cutaneous melanoma. Nature Medicine 2012;18(8):1239-1247. 129. Dai D, Martinka M, Bush J, et al. Reduced apaf-1 expression in human cutaneous melanomas. British Journal of Cancer 2004;91:1089-1095. 130. Tai E, Benchimol S. TRIMming p53 for ubiquitination. PNAS 2009;106(28):11431- 11432. 131. Flaherty K, Hodi F, Bastian B. Mutation-driven drug development in melanoma. Current Opinions in Oncology 2010;22(3):178-183. 132. Handolias D, Hamilton A, Salemi R, et al. Clinical responses observed with imatinib or sorafenib in melanoma patients expressing mutations in KIT. British Journal of Cancer 2010;102:1219-1223. 133. McArthur G, Ribas A. Targeting oncogenic drivers and the immune system in melanoma. Journal of Clinical Oncology 2013;31(4):499-506. 134. Vultur A, Herlyn M. Snapshot: Melanoma. Cancer Cell 2013;23:706-707.

366

135. Gupta P, Kuperwasser C, Brunet J, et al. The melanocyte differentiation program predisposes to metastasis after neoplastic transformation. Nature Genetics 2005;37(10):1047-1054. 136. Kutzner H, Metzler G, Argenyi Z, et al. Histological and genetic evidence for a variant of superficial spreading melanoma composed predominantly of large nests. Modern Pathology 2012;25:838-845. 137. Lin K, Baritaki S, Militello L, et al. The role of B-Raf mutations in melanoma and the induction of EMT via dysregulation of the NF-KB/Snail/RKIP/PTEN circuit. Genes & Cancer 2010;1(5):409-420. 138. Liu S, Kumar S, Martin J, et al. Snail1 mediates hypoxia-induced melanoma progression. American Journal of Pathology 2011;179(6):3020-3031. 139. Sinnberg T, Menzel M, Ewerth D, et al. Beta-catenin signaling increases during melanoma progression and promotes tumour cell survival and chemoresistance. Plos One 2011;6(8):1-13. 140. Biechele T, Kulikauskas R, Toroni R, et al. Wnt/beta-catenin signaling and AXIN1 regulate apoptosis triggered by inhibition of the mutant kinase BRAFV600E in human melanoma. Science Signaling 2012;5(206):1-12. 141. Pawlikowski J, McBryan T, van Tuyn J, et al. Wnt signaling potentiates nevogenesis. PNAS 2013;110(40):16009-16014. 142. Grossmann A, Yoo J, Clancy J, et al. The small GTPase ARF6 stimulates beta-catenin transcriptional activity during WNT5A-mediated melanoma invasion and metastasis. Science Signaling 2013;6(265):1-13. 143. Gallagher S, Rambow F, Kumasaka M, et al. Beta-catenin inhibits melanocyte migration but induces melanoma metastasis. Oncogene 2013;32(17):2230-2238. 144. Damsky W, Curley D, Santhanakrishnan M, et al. Beta-Catenin signaling controls metastasis in braf-activated pten-deficient melanomas. Cancer Cell 2011;20:741-754. 145. Delmas V, Beermann F, Martinozzi S, et al. Beta-Catenin induces immortalization of melanocytes by suppressing p16INK4a expression and cooperates with N-Ras in melanoma development. Genes & Development 2007;21:2923-2935. 146. Webster M, Weeraratna A. A Wnt-er migration: the confusing role of beta-catenin in melanoma metastasis. Science Signaling 2013;6(268):pe11. 147. Weeraratna A, Jiang Y, Hostetter G, et al. Wnt5a signaling directly affects cell motility and invasion of metastatic melanoma. Cancer Cell 2002;1:279-288. 148. Alexaki V, Javelaud D, Van Kempen L, et al. GLI2-mediated melanoma invasion and metastasis. Journal of the National Cancer Institute 2010;102(15):1148-1159.

367

149. Scott K, Nogueira C, Haffernan T, et al. Proinvasion metastasis drivers in early-stage melanoma are oncogenes. Cancer Cell 2011;20:92-103. 150. Caramel J, Papadogeogakis E, Hill L, et al. A switch in the expression of embryonic EMT-inducers drives the development of malignant melanoma. Cancer Cell 2013;24(4):466-480. 151. Wels C, Joshi S, Koefinger P, et al. Transcriptional activation of ZEB1 by slug leads to cooperative regulation of the epithelial-mesenchymal transition-like phenotype in melanoma. Journal of Investigative Dermatology 2011;131:1877-1885. 152. Fenouille N, Tichet M, Dufies M, et al. The epithelial-mesenchymal transition (EMT) regulatory factor SLUG (SNAI2) is a downstream target of SPARC and AKT in promoting melanoma cell invasion. Plos One 2012;7(7):1-15. 153. Kumar S, Sharma P, Kumar D, et al. Functional characterization of stromal osteopontin in melanoma progression and metastasis. Plos One 2013;8(7):1-12. 154. Rangel J, Nosrati M, Torabian S, et al. Osteopontin as a molecular prognostic marker for melanoma. Cancer 2008;112(1):144-150. 155. Horst B, Gruvberger-Saal S, Hopkins B, et al. Gab2-mediated signaling promotes melanoma metastasis. American Journal of Pathology 2009;174(4):1524-33. 156. Sigalotti L, Covre A, Fratta E, et al. Epigenetics of human cutaneous melanoma: setting the stage for new therapeutic strategies. Journal of Translational Medicine 2010;8(56):1-22. 157. Mori T, O'Day S, Umetani N, et al. Predictive utility of circulating methylated DNA in serum of melanoma patients recieving biochemotherapy. Journal of Clinical Oncology 2005;23(36):9351-9358. 158. Liu S, Ren S, Howell P, et al. Identification of novel epigenetically modified genes in human melanoma via promoter methylation gene profiling. Pigment Cell Melanoma Research 2008;21:545-558. 159. Hoon D, Spugnardi M, Kuo C, et al. Profiling epigenetic inactivation of tumour suppressor genes in tumours and plasma from cutaneous melanoma patients. Oncogene 2004;23:4014-4022. 160. Muthusamy V, Duraisamy S, Bradbury. CM., et al. Epigenetic silencing of novel tumor suppressors in malignant melanoma. Cancer Research 2006;66(23):11187- 11193. 161. Lian C, Xu Y, Ceol. C., et al. Loss of 5-hydroxymethylcytosine is an epigenetic hallmark of melanoma. Cell 2012;150:1135-1146. 162. Fratta E, Sigalotti L, Covre A, et al. Epigenetics of melanoma: implications for immune-based therapies. Immunotherapy 2013;5(10):1103-1116.

368

163. Ongenaert M, Van Neste L, De Meyer T, et al. Pubmeth: a cancer methylation database combining text-mining and expert annotation. Nucleic Acids Research 2008;36(842-846):842. 164. Villanueva J, Vultur A, Lee J, et al. Acquired resistance to BRAF inhibitors mediated by a RAF kinase switch in melanoma can be overcome by cotargeting MEK and IGF- 1R/PI3K. Cancer Cell 2010;18:683-695. 165. Bhatia S, Tykodi S, Thompson J. Treatment of metastatic melanoma: an overview. Oncology 2009;23(6):488-496. 166. Kenchappa R, Tran N, Rao N, et al. Novel treatments for melanoma brain metastases. Cancer Control 2013;20(4):298-306. 167. Minniti G, Clarke E, Lanzetta G, et al. Stereotactic radiosurgery for brain metastases: analysis of outcome and risk of brain radionecrosis. Radiation Oncology 2011;6(48):1-9. 168. Forschner A, Heinrich V, Pflugfelder A, et al. The role of radiotherapy in the overall treatment of melanoma. Clinics in Dermatology 2013;31:282-289. 169. Rompoti N, Schilling B, Livingstone E, et al. Combination of BRAF inhibitors and brain radiotherapy in patients with metastatic melanoma shows minimal acute toxicity. Journal of Clinical Oncology 2013. 170. Roos W, Jost E, Belohlavek C, et al. Intrinsic anticancer drug resistance of malignant melanoma cells is abrogated by IFN-beta and valporic acid. Cancer Research 2011;71(12):4150-4160. 171. Middleton M, Grob J, Aaronson N, et al. Randomized phase II study of temozolomide versus dacarbazine in the treatment of patients with advanced metastatic malignant melanoma. Journal of Clinical Oncology 2000;18(1):158-166. 172. Agarwala S, Kirkwood J. Temozolomide, a novel alkylating agent with activity in the central nervous system, may improve the treatment of advanced metastatic melanoma. The Oncologist 2000;5:144-151. 173. Alcala A, Flaherty K. BRAF inhibitors for the treatment of metastatic melanoma: clinical trials and mechanisms of resistance. American Association for Cancer Research 2012;18(1):33-39. 174. Rochet N, Dronca R, Kottschade L, et al. Melanoma brain metastases and vemurafenib: need for further investigation. Mayo Clinic Proceedings 2012;87(10):976-981. 175. Flaherty K, Yasothan U, Kirkpatrick P. Vemurafenib. Nature Reviews Drug Discovery 2011;10:811-812.

369

176. Gogas H, Ioannovich J, Dafni U, et al. Prognostic significance of autoimmunity during treatment of melanoma with interferon. New England Journal of Medicine 2006;354(7):709-718. 177. Ribas A. Tumor immunotherapy directed at PD-1. New England Journal of Medicine 2012;366(26):2517-2519. 178. Smith F, Downey S, Klapper J, et al. Treatment of metastatic melanoma using interleukin-2 alone or in conjunction with vaccines. Clinical Cancer Research 2008;14(17):5610-5618. 179. Weinreich D, Rosenberg S. Response rates of patients with metastatic melanoma to high-dose intravenous interleukin-2 after prior exposure to alpha-interferon or low- dose interleukin-2. Journal of Immunotherapy 2008;25(2):185-187. 180. Sullivan R, LoRusso P, Flaherty K. The intersection of immune-directed and molecularly targeted therapy in advanced melanoma: where we have been, are, and will be. Clinical Cancer Research 2013;19(19):5283-5291. 181. Joseph R, Sullivan R, Harrell R, et al. Correlation of NRAS mutations with clinical response to high-dose IL-2 in patients with advanced melanoma. Journal of Immunotherapy 2012;35:66-72. 182. Garber K. Industry makes strides in melanoma. Nature Biotechnology 2010;28(8):763- 764. 183. Mullard A. Dual fronts poised to transform melanoma therapy. Nature Reviews Drug Discovery 2011;10:325-326. 184. Salama A, Hodi S. Cytotoxtic T-lymphocyte-associated antigen-4. Clinical Cancer Research 2011;17(14):4622-4628. 185. Pardoll D. Immunology beats cancer: a blueprint for successful translation. Nature Immunology 2012;13(12):1129-1132. 186. Pardoll D. The blockade of immune checkpoints in cancer immunotherapy. Nature Reviews Cancer 2012;12:252-264. 187. Waitz R, Solomon S, Petre E, et al. Potent induction of tumor immunity by combining tumor cryoablation with anti-CTLA-4 therapy. Cancer Research 2012;72(2):430-439. 188. Hodi F, O'Day S, McDermott D, et al. Improved survival with ipilimumab in patients with metastatic melanoma. New England Journal of Medicine 2010;363(8):711-723. 189. Ribas A, Kefford R, Marshall M, et al. Phase III randomized clinical trial comparing tremelimumab with standard-of-care chemotherapy in patients with advanced melanoma. The Journal of Clinical Oncology 2013;30:1-7.

370

190. Ott P, Hodi F, Robert C. CTLA-4 and PD-1/PD-L1 blockade: New immunotherapeutic modalities with durable clinical benefit in melanoma patients. Clinical Cancer Research 2013;19(19):5300-5309. 191. Wolchok J, Kluger H, Callahan M, et al. Nivolumab plus ipilimumab in advanced melanoma. New England Journal of Medicine 2013;369(2):122-133. 192. Hamid O, Robert C, Daud A, et al. Safety and tumour responses with lambrolizumab (anti-PD-1) in melanoma. New England Journal of Medicine 2013;369(2):134-144. 193. Garber K. Melanoma combination therapies ward off tumor resistance. Nature Biotechnology 2013;31(8):666-668. 194. Topalian S, Hodi F, Brahmer J, et al. Safety, activity, and immune correlates of anti- PD-1 antibody in cancer. New England Journal of Medicine 2012;366(26):2443- 2454. 195. Ribas A, Flaherty K. BRAF targeted therapy changes the treatment paradigm in melanoma. Nature Reviews in Clinical Oncology 2011;8:426-433. 196. Long G, Wilmott J, Haydu L, et al. Effects of BRAF inhibitors on human melanoma tissue before treatment, early during treatment, and on progression. Pigment Cell Melanoma Research 2013;26(4):499-508. 197. Mangana J, Levesque M, Karpova M, et al. Sorafenib in melanoma. Expert opinion on investigational drugs 2012;21(4):557-568. 198. Ott P, Hamilton A, Min C, et al. A phase II trial of sorafenib in metastatic melanoma with tissue correlates. Plos One 2010;5(12):1-9. 199. Tsai J, Lee J, Wang W, et al. Discovery of a selective inhibitor of oncogenic B-Raf kinase with potent antimelanoma activity. PNAS 2008;105(8):3041-3046. 200. Heidorn S, Milagre C, Whittaker S, et al. Kinase-Dead BRAF and oncogenic RAS cooperate to drive tumor progression through CRAF. Cell 2010;140:209-221. 201. Bollag G, Tsai J, Zhang J, et al. Vemurafenib: the first drug approved for BRAF- mutant cancer. Nature Reviews Drug Discovery 2012;11:873-886. 202. Yang H, Higgins B, Kolinsky K, et al. RG7204 (PLX4032), a selective BRAFV600E inhibitor, displays potent antitumor activity in preclinical melanoma models. Cancer Research 2010;70(13):5518-5527. 203. Haq R, Shoag J, Andreu-Perez P, et al. Oncogenic BRAF regulates oxidative metabolism via PGC1alpha and MITF. Cancer Cell 2013;23:302-315. 204. Su F, Viros A, Milagre C, et al. RAS mutations in cutaneous squamous-cell carcinomas in patients treated with BRAF inhibitors. The New England Journal of Medicine 2012;366(3):207-215.

371

205. Hauschild A, Grob J, Demidov L, et al. Dabrafenib in BRAF-mutated metastatic melanoma: a multicentre, open-label, phase 3 randomised controlled trial. The Lancet 2012;380:358-365. 206. Joseph E, Pratilas C, Poulikakos P, et al. The RAF inhibitor PLX4032 inhibits ERK signaling and tumor cell proliferation in a V600E BRAF selective manner. PNAS 2010;107(33):14903-14908. 207. Beck D, Niessner H, Smalley K, et al. Vemurafenib potently induces endoplasmic retuculum stress-mediated apoptosis in BRAFV600E melanoma cells. Science Signaling 2013;6(260):1-11. 208. Bollag G, Hirth P, Tsai J, et al. Clinical efficacy of a RAF inhibitor needs broad target blockade in BRAF-mutant melanoma. Nature 2010;467:596-599. 209. Haferkamp S, Borst A, Adam C, et al. Vemurafenib induces senescence features in melanoma cells. Journal of Investigative Dermatology 2013;133:1601-1609. 210. Jiang C, Lai F, Tay K, et al. Apoptosis of human melanoma cells induced by inhibition of B-RAFV600E involves preferential splicing of bim s. Cell Death and Disease 2010;1(4):1-14. 211. Kwong L, Costello J, Liu H, et al. Oncogenic NRAS signaling differentially regulates survival and proliferation in melanoma. Nature Medicine 2012;18(10):1503-1510. 212. Eggermont A, Robert C. A new paradigm tumor for drug development. Nature Reviews in Clinical Oncology 2011;9:74-76. 213. Ascierto P, Schadendorf D, Berking C, et al. MEK162 for patients with advanced melanoma harbouring NRAS or Val600 BRAF mutations: a non-randomised, open- label phase 2 study. The Lancet 2013;14:249-256. 214. Ascierto P, Grimaldi A, Acquavella N, et al. Future perspectives in melanomas research. Meeting report from the "Melanoma Bridge. Napoli, December 2nd-4th 2012". Journal of Translational Medicine 2013;11(137):1-29. 215. Aziz S, Davies M, Pick E, et al. Phosphatidylinositol-3-kinase as a therapeutic target in melanoma. Clinical Cancer Research 2009;15(9):3029-3036. 216. Marone R, Erhart D, Mertz A, et al. Targeting melanoma with dual phosphoinositide 3-kinase/mammalian target of rapamycin inhibitors. Molecular Cancer Research 2009;7(4):601-613. 217. Greger J, Eastman S, Zhang V, et al. Combinations of BRAF, MEK, and PI3K/mTOR inhibitors overcome acquired resistance to the BRAF inhibitor GSK2118436 Dabrafenib, mediated by NRAS or MEK mutations. Molecular Cancer Therapeutics 2012;11(4):909-920. 218. Flaherty K, Robert C, Hersey P, et al. Improved survival with MEK inhibition in BRAF-mutated melanoma. New England Journal of Medicine 2012;367(2):107-114. 372

219. Lo R. Melanoma prognostics and personalized therapeutics at a crossroad. The Journal of Investigative Dermatology 2013;133:292-295. 220. de Bono J, Ashworth A. Translating cancer research into targeted therapeutics. Nature 2010;467:543-549. 221. Cox A, Der C. The RAF inhibitor paradox revisited. Cancer Cell 2012;21:147-149. 222. Smalley K, Nathanson K, Flaherty K. Genetic subgrouping of melanoma reveals new opportunities for targeted therapy. Cancer Research 2009;69(8):3241-3244. 223. Hatzivassiliou G, Haling J, Chen H, et al. Mechanism of MEK inhibition determines efficacy in mutant KRAS-versus BRAF-driven cancers. Nature 2013;501(7466):232- 236. 224. Ko J, Fisher D. A new era: melanoma genetics and therapeutics. Journal of Pathology 2010;223:241-250. 225. Fecher L, Cummings S, Keefe M, et al. Towards a molecular classification of melanoma. The Journal of Clinical Oncology 2007;25(12):1606-1620. 226. Chapman P, Hauschild A, Robert C, et al. Improved survival with vemurafenib in melanoma with BRAF V600E mutation. The New England Journal of Medicine 2011;364(26):2507-2516. 227. Jang S, Atkins M. Which drug, and when, for patients with BRAF-mutant melanoma? The Lancet 2013;14:e60-e69. 228. Gibney G, Messina J, Fedorenko I, et al. Paradoxical oncogenesis-the long-term effects of BRAF inhibition in melanoma. Nature Reviews Clinical Oncology 2013;10(7):390-399. 229. Sosman J, Kim K, Schuchter L, et al. Survival in BRAF V600-mutant advanced melanoma treated with vemurafenib. The New England Journal of Medicine 2012;366(8):707-714. 230. Tawbi H, Kirkwood J. BREAKing a path for progress -dabrafenib confirms class effect. Nature Reviews Clinical Oncology 2012;9:497-497. 231. Falchook G, Long G, Kurzrock R, et al. Dabrafenib in patients with melanoma, untreated brain metastases, and other solid tumours: a phase 1 dose-escalation trial. The Lancet 2012;379:1893-1901. 232. Falchook G, Lewis K, Infante J, et al. Activity of the oral MEK inhibitor trametinib in patients with advanced melanoma: a phase 1 dose-escalation trial. The Lancet 2012;13(8):782-789. 233. Kim K, Kefford R, Pavlick A, et al. Phase II study of the MEK1/MEK2 inhibitor trametinib in patients with metastatic BRAF-mutant cutaneous melanoma previously

373

treated with or without a BRAF inhibitor. Journal of Clinical Oncology 2013;31(4):482-489. 234. Ribas A, Hodi F, Callahan M, et al. Hepatotoxicity with combination of vemurafenib and ipilimumab. New England Journal of Medicine 2013;364(14):1365-1366. 235. Smalley K, Haass N, Brafford P, et al. Multiple signaling pathways must be targeted to overcome drug resistance in cell lines derived from melanoma metastases. Molecular Cancer Therapeutics 2006;5(5):1136-1144. 236. Holzel M, Bovier A, Tuting T. Plasticity of tumour and immune cells: a source of heterogeneity and a cause for therapy resistance? Nature Reviews Cancer 2013;13:365-376. 237. Frederick D, Piris A, Cogdill A, et al. BRAF inhibition is associated with enhanced melanoma antigen expression and a more favorable tumor microenvironment in patients with metastatic melanoma. Clinical Cancer Research 2013;19(5):1225-1231. 238. Infante J, Fecher L, Flachook G, et al. Safety, pharmacokinetic, pharmocodynamic, and efficacy data for the oral MEK inhibitor trametinib: a phase 1 dose-escalation trial. The Lancet 2012;13:773-781. 239. Posch C, Moslehi H, Feeney L, et al. Combined targeting of MEK and PI3K/mTOR effector pathways is necessary to effectively inhibit NRAS mutant melanoma in vitro and in vivo. PNAS 2013;110(10):4015-4020. 240. Gupta S, Ramjuan A, Haiko P, et al. Binding of ras to phosphoinositide 3-kinase p110alpha is required for ras-driven tumorigenesis in mice. Cell 2007;129(5):957- 968. 241. Espona-Fiedler M, Soto-Cerrato V, Hosseini A, et al. Identification of dual mTORC1 and mTORC2 inhibitors in melanoma cells: prodigiosin vs obatoclax. Biochemical Pharmacology 2012;83(4):489-496. 242. Ji Z, Kumar R, Taylor M, et al. Vemurafenib synergizes with nutlin-3 to deplete survivin and suppress melanoma viability and tumor growth. Clinical Cancer Research 2013;19(16):4383-4391. 243. Margolin K, Ernstoff M, Hamid O, et al. Ipilimumab in patients with melanoma and brain metastases: an open label, phase 2 trial. The Lancet 2012;13:459-465. 244. Boni A, Cogdill A, Dang P, et al. Selective BRAFV600E inhibition enhances T-cell recognition of melanoma without affecting lymphocyte function. Cancer Research 2010;70(13):5213-5219. 245. Kwong L, Chin L. The brothers RAF. Cell 2010;140:180-182. 246. Hatzivassiliou G, Song K, Yen I, et al. RAF inhibitors prime wild-type RAF to activate the MAPK pathway and enhance growth. Nature 2010;464:431-436.

374

247. Gowrishankar K, Snoyman S, Pupo G, et al. Acquired resistance to BRAF inhibition can confer cross-resistance to combined BRAF/MEK inhibition. Journal of Investigative Dermatology 2012;132:1850-1859. 248. Lau E, Ronai Z. RAF-isotype switching: from B to C through PDE. Nature Structural & Molecular Biology 2011;18(5):517-518. 249. Tahiri A, Roe K, Ree A, et al. Differential inhibition of ex-vivo tumor kinase activity by vemurafenib in BRAF(V600E) and BRAF wild-type metastatic malignant melanoma. Plos One 2013;8(8):1-12. 250. Wilmott J, Tembe V, Howle J, et al. Intratumoral molecular heterogeneity in a BRAF- mutant, BRAF inhibitor-resistant melanoma: a case illustrating the challenges for personalized medicine. Molecular Cancer Therapeutics 2012;11(12):2704-2708. 251. Shi H, Moriceau G, Kong X, et al. Melanoma whole-exome sequencing identifies V600EB-RAF amplification-mediated acquired B-RAF inhibitor resistance. Nature Communications 2012;3(724):1-8. 252. Sullivan R, Flaherty K. Resistance to BRAF-targeted therapy in melanoma. European Journal of Cancer 2013;49(6):1297-1304. 253. Thakur M, Salangsang F, Landman A, et al. Modelling vemurafenib resistance in melanoma reveals a strategy to forestall drug resistance. Nature 2013;000:1-6 254. Roesch A, Vultur A, Bogeski I, et al. Overcoming intrinsic multidrug resistance in melanoma by blocking the mitochondrial respiratory chain of slow-cycling JARID1Bhigh cells. Cancer Cell 2013;23:811-825. 255. Nathanson K, Martin A, Wubbenhorst B, et al. Tumor genetic analyses of patients with metastatic melanoma treated with BRAF inhibitor Dabrafenib (GSK2118436). Clinical Cancer Research 2013;19(17):4868-4878. 256. Trunzer K, Pavlick A, Schuchter L, et al. Pharmacodynamic effects and mechanisms of resistance to vemurafenib in patients with metastatic melanoma. Journal of Clinical Oncology 2013;31:1-15. 257. Wilson T, Fridlyand J, Yan Y, et al. Widespread potential for growth-factor-driven resistance to anticancer kinase inhibitors. Nature 2012;487:505-510. 258. Yadav V, Zhang X, Liu J, et al. Reactivation of mitogen-activated protein kinase (MAPK) pathway by FGF receptor 3 (FGFR3)/Ras mediates resistance to vemurafenib in human B-RAF V600E mutant melanoma. The Journal of Biological Chemistry 2012;287(33):28087-28098. 259. Nazarian R, Shi H, Wang Q, et al. Melanomas acquire resistance to B-RAF(V600E) inhibition by RTK or N-RAS upregulation. Nature 2010;468:973-979. 260. Straussman R, Morikawa T, Shee K, et al. Tumour micro-environment elicits innate resistance to RAF inhibitors through HGF secretion. Nature 2012;487:500-504. 375

261. Drummer R, Flaherty K. Resistance patterns with tyrosine kinase inhibitors in melanoma: new insights. Current Opinions in Oncology 2012;24(2):150-154. 262. Wagle N, Emery C, Berger M, et al. Dissecting therapeutic resistance to RAF inhibition in melanoma by tumour genomic profiling. Journal of Clinical Oncology 2011;29(22):3085-3096. 263. Villanueva J, Infante, JR., Krepler, C, et al. Concurrent MEK2 mutation and BRAF amplification confer resistance to BRAF and MEK inhibitors in melanoma. Cell Reports 2013;4:1-10. 264. Chen B, Tardell C, Higgins B, et al. BRAFV600E negatively regulates the AKT pathway in melanoma cell lines. Plos One 2012;7(8):1-9. 265. Abel E, Basile K, Kugel C, et al. Melanoma adapts to RAF/MEK inhibitors through FODD3-mediated upregulation of ERBB3. Journal of Clinical Investigation 2013;123(5):2155-2168. 266. Becker T, Boyd S, Mijatov B, et al. Mutant B-RAF-Mcl-1 survival signaling depends on the STAT3 transcription factor. Oncogene 2013:1-9. 267. Liu F, Cao J, Wu J, et al. Stat3 targeted therapies overcome the acquired resistance to vemurafenib in melanomas. Journal of Investigative Dermatology 2013. 268. Koop A, Satzger I, Alter M, et al. Intermittent B-Raf inhibitor therapy is a fesible option- report of a patient with metastatic melanoma. British Journal of Dermatology 2013. 269. Thakur M, Stuart D. The evolution of melanoma resistance reveals therapeutic opportunities. Cancer Research 2013;73(20):1-5. 270. AACR. Intermittent vemurafenib prevents resistance in melanoma. Nature Reviews Clinical Oncology 2013;10(250). 271. Romano E, Pradervand S, Paillusson A, et al. Identification of multiple mechanisms of resistance to vemurafenib in a patient with BRAFV600E-mutated cutaneous melanoma successfully rechallenged after progrerssion. Clinical Cancer Research 2013. 272. Poulikakos P, Solit D. Resistance to MEK inhibitors: should we co-target upstream? Science Signaling 2011;4(166):1-3. 273. Emery C, Vijayendran K, Zipser M, et al. MEK1 mutations confer resistance to MEK and B-RAF inhibition. PNAS 2009;106(48):20411-20416. 274. Gogas H, Eggermont A, Hauschild A, et al. Biomarkers in melanoma. Annals of Oncology 2009;20:vi8-vi13. 275. Ollila S, Makela T. The tumour suppressor kinase LKB1: lessons from mouse models. Journal of Molecular Cell Biology 2011;3:330-340. 376

276. Nakada Y, Stewart T, Pena C, et al. The LKB1 tumous suppressor as a biomarker in mouse and human tissues. Plos One 2013;8(9):e73449. 277. Rothberg B, Bracken M, Rimm D. Tissue biomarkers for prognosis in cutaneous melanoma : a systematic review and meta-analysis. Journal of the National Cancer Institute 2009;101(7):452-474. 278. Larson A, Konat E, Alani R. Melanoma biomarkers: current status and vision for the future. Nature Clinical Practice Oncology 2009;6(2):105-117. 279. Chen G, Cheng Y, Tang Y, et al. Role of Tip60 in human melanoma cell migration, metastasis, and patient survival. Journal of Investigative Dermatology 2012;132:2632-2641. 280. Rothberg B, Berger A, Molinaro A, et al. Melanoma prognostic model using tissue microarrays and genetic algorithms. Journal of Clinical Oncology 2009;27(34):5772- 5780. 281. Fang M, Xia F, Mahalingam M, et al. MEN1 is a melanoma tumour suppressor that preserves genomic integrity by stimulating transcription of genes that promote homologous recombination-directed DNA repair. Molecular and Cellular Biology 2013;33(13):2635-2647. 282. Anghel S, Correa-Rochal R, Budinska E, et al. Breast cancer suppressor candidate-1 (BCSC-1) is a melanoma tumor suppressor that down regulates MITF. Pigment Cell Melanoma Research 2012;25:482-487. 283. Poetsch M, Dittberner T, Cowell J, et al. TTC4, a novel candidate tumor suppressor gene at 1p31 is often mutated in malignant melanoma of the skin. Oncogene 2000;19:5817-5820. 284. Fung J, Smith R, Brown M, et al. Identification and characterization of a novel melanoma tumors suppressor gene on human chromosome 6q21. Clinical Cancer Research 2009;15(3):797-803. 285. Marin I. Origin and diversification of TRIM ubiquitin ligases. Plos One 2012;7(11):1- 10.

286. Ozato K, Shin D, Chang T, et al. TRIM family proteins and their emerging roles in innate immunity. Nature Reviews Immunology 2008;8:849-860. 287. Han K, Lou D, Sawyer S. Identification of a genomic reservoir for new TRIM genes in primate genomes. PLOS Genetics 2011;7(12):1-14. 288. Versteeg G, Rajsbaum R, Sanchez-Aparicio M, et al. The E3-Ligase TRIM family of proteins regulates signaling pathways triggered by innate immune pattern-recognition receptors. Immunity 2013;38:384-398. 289. Napolitano L, Meroni G. TRIM family: pleiotropy and diversification through homomultimer and heteromultimer formation. IUBMB Life 2012;64(1):64-71. 377

290. Sardiello M, Cairo S, Fontanella B, et al. Genomic analysis of the TRIM family reveals two groups of genes with distinct evolutionary properties. BMC Evolutionary Biology 2008;8(225):1-22. 291. Pasquier L. Fish 'n' TRIMs. Journal of Biology 2009;8(50):50.1-50.5. 292. Boudinot P, van der Aa L, Jouneau L, et al. Origin and evolution of TRIM proteins: new insights from the complete TRIM repertoire of zebrafish and pufferfish. Plos One 2011;6(7):1-18. 293. Perfetto L, Gherardini P, Davey N, et al. Exploring the diversity of SPRY/B30.2- mediated interactions. Trends in Biochemical Sciences 2013;38(1):36-46. 294. Nisole S, Stoye J, Saib A. TRIM family proteins: retroviral restriction and antiviral defence. Nature Reviews Microbiology 2005;3:799-808. 295. Gack M. TRIMming flavivirus infection. Cell Host & Microbe 2011;10:175-177. 296. Sugiura T. The cellular level of TRIM31, an RBCC protein overexpressed in gastric cancer, is regulated by multiple mechanisms including the ubiquitin-proteasome system. Cell Biology International 2011;35:657-661. 297. Hatakeyama S. TRIM proteins and cancer. Nature Reviews. Cancer 2011;11:792-804. 298. Chu Y, Yang X. SUMO E3 ligase activity of TRIM proteins. Oncogene 2011;30(9):1108-1116. 299. Bertrand-Vallery V, Belot N, Dieu M, et al. Proteomic profiling of human keratinocytes undergoing UVB-induced alternative differentiation reveals TRIpartite motif protein 29 as a survival factor. Plos One 2010;5(5):1-13. 300. Muller H, Woods G. Ultraviolet radiation effects on the proteome of skin cells. Advances in Experimental Medicine and Biology 2013;990. 301. Cho M. Tripartite motif-containing 29 expression in squamous cell carcinoma. Annals of Dermatology 2012;24(4):491-493. 302. Horn E, Albor A, Liu Y, et al. RING protein trim32 associated with skin carcinogenesis has anti-apoptotic and E3-ubiquitin ligase properties. Carcinogenesis 2004;25(2):157-167. 303. Espinosa A, Dardalhon V, Brauner S, et al. Loss of the lupus autoantigen ro52/trim21 induces tissue inflammation and systemic autoimmunity by disregulating the IL-23- Th17 pathway. Journal of Experimental Medicine 2009;206(8):1661-1671. 304. Munding C, Keller M, Niklaus G, et al. The estrogen-responsive b box protein: a novel enhancer of interleukin-1beta secretion. Cell Death and Differentiation 2006;13:1938-1949.

378

305. Beer H, Munding C, Dubois N, et al. The estrogen-responsive b box protein. The Journal of Biological Chemistry 2002;277(23):20740-20749. 306. Taylor R, Lubick K, Robertson S, et al. TRIN79alpha, an interferon-stimulated gene product, restricts tick-borne encephalitis virus replication by degrading the viral RNA polymerase. Cell Host & Microbe 2011;10:185-196. 307. Jefferies C, Wynne C, Higgs R. Antiviral TRIMs: friend or foe in autoimmune and autoinflammatory disease? Nature Reviews Immunology 2011;11:617-625. 308. McNab F, Rajsbaum R, Stoye J, et al. Tripartite-motif proteins and innate immune regulation. Current Opinion in Immunology 2011;23:46-56. 309. Munir M. TRIM proteins: another class of viral victims. Science Signaling 2010;3(118):1-4. 310. James L, Keeble A, Khan Z, et al. Structural basis for PRYSPRY-mediated tripartite motif (TRIM) protein function. PNAS 2007;104(15):6200-6205. 311. Towers G. The control of viral infection by tripartite motif proteins and cyclophilin A. Retrovirology 2007;4(40):1-10. 312. Stremlau M, Owens C, Perron M, et al. The cytoplasmic body component TRIM5alpha restricts HIV-1 infection in old world monkeys. Nature 2004;427:848- 853. 313. Barr S, Smiley J, Bushman F. The interferon response inhibits HIV particle production by induction of TRIM22. Plos Pathogens 2008;4(2):1-11. 314. Si Z, Vandegraaff N, O'Huigin C, et al. Evolution of a cytoplasmic tripartite motif (TRIM) protein in cows that restricts retroviral infection. PNAS 2006;103(19):7454- 7459. 315. Gack M, Shin Y, Joo C, et al. TRIM25 RING-finger E3 ubiquitin ligase is essential for RIG-I-mediated antiviral activity. Nature 2007;446:916-921. 316. Kawai T, Akira S. Regulation of innate immune signalling pathways by the tripartite motif (TRIM) family proteins. EMBO Molecular Medicine 2011;3:513-527. 317. Kamitani S, Ohbayashi N, Ikeda O, et al. KAP1 regulates type I interferon/STAT1- mediated IRF-1 gene expression. Biochemical and Biophysical Research Communications 2008;370:366-370. 318. Allouch A, Di Primio C, Alpi E, et al. The TRIM family protein KAP1 inhibits HIV-1 integration. Cell Host & Microbe 2011;9:484-495. 319. Zhang S, Guo J, Wu J, et al. Identification and characterization of multiple TRIM proteins that inhibit hepatitis B virus transcription. Plos One 2013;8(8):1-12.

379

320. Sjostrand M, Brauner S, Kvarnstrom M, et al. TRIM genes are part of the interferon signature observed in patients with primary Sjogren's syndrome. Annals of the Rheumatic Diseases 2012;71(Suppl1):A1-A93. 321. Bell J, Malyukova A, Holien J, et al. TRIM16 acts as an E3 ubiquitin ligase and can heterodimerize with other TRIM family members. Plos One 2012;7(5):1-9. 322. Cambiaghi V, Giuliani V, Lombardi S, et al. TRIM proteins in cancer. Advances in Experimental Medicine and Biology 2012;770:77-91. 323. Khetchoumian K, Teletin M, Tisserand J, et al. Loss of TRIM24 (Tif1alpha) gene function confers oncogenic activity to retinoic acid receptor alpha. Nature Genetics 2007;39(12):1500-1506. 324. Yuan Z, Villagra A, Peng L, et al. The ATDC (TRIM29) protein binds p53 and antagonizes p53-mediated functions. Molecular and Cellular Biology 2010;12:3004- 3015. 325. Kano S, Miyajima N, Fukuda S, et al. Tripartite motif protein 32 facilitates cell growth and migration via degradation of Abl-Interactor 2. Cancer Research 2008;68(14):5572-5580. 326. Zhang J, Hu M, Wang Y, et al. TRIM32 protein modulates type I interferon induction and cellular antiviral response by targeting MITA/STING protein for K63-linked ubiquitination. Journal of Biological Chemistry 2012;287(34):28646-28655. 327. Sugiura T, Miyamoto K. Characterization of TRIM31, upregulated in gastric adenocarcinoma, as a novel RBCC protein. Journal of Cellular Biochemistry 2008;105:1081-1091. 328. Khatamianfar V, Valiyeva F, Rennie P, et al. TRIM59, a novel multiple cancer biomarker for immunohistochemical detection of tumorigenesis. BMJ Open 2012;2(5):1-10. 329. Tsukamoto H, Kato T, Enomoto A, et al. Expression of ret finger protein correlates with outcomes in endometrial cancer. Cancer Science 2009;100(10):1895-1901. 330. Iwakoshi A, Murakumo Y, Kato T, et al. RET finger protein expression is associated with prognosis in lung cancer with epidermal growth factor receptor mutations. Pathology International 2012;62:324-330. 331. Lee J, Shan J, Zhong J, et al. RFP-mediated ubiquitination of PTEN modulates its effects on AKT activation. Cell Research 2013;23:552-564. 332. Dho S, Kwon K. The RET finger protein induces apoptosis via its RING finger-B- box-coiled-coil motif. The Journal of Biological Chemistry 2003;278(34):31902- 31908. 333. Zoumpoulidou G, Broceno C, Li H, et al. Role of the tripartite motif protein 27 in cancer development. Journal of the National Cancer Institute 2012;104:941-952. 380

334. Krutzfeldt M, Ellis M, Weekes D, et al. Selective ablation of retinoblastoma protein function by the RET finger protein. Molecular Cell 2005;18:213-224. 335. Lott S, Chen N, Chandler D, et al. DEAR1 is a dominant regulator of acinar morphogenesis and an independent predictor of local recurrence-free survival in early-onset breast cancer. PLoS Medicine 2009;6(5):1-16. 336. Uhlen M, Oksvold P, Fagerberg L, et al. Towards a knowledge based Human Protein Atlas. Nature Biotechnology 2010;28(12):1248-1250. 337. Liu H, Golder-Novoselsky E, Seto M, et al. The novel estrogen-responsive b box protein (EBBP) gene is tamoxifen regulated in cells expressing an estrogen receptor DNA-binding domain mutant. Molecular Endocrinology 1998;12:1733-1748. 338. Rajsbaum R, Stoye J, O'Garra A. Type I interferon-dependant and -indepandant expression of tripartite motif proteins in immune cells. European Journal of Immunology 2008;38:619-630. 339. Uchil P, Hinz A, Siegel S, et al. TRIM protein-mediated regulation of inflammatory and innate immune signaling and its association with antiretroviral activity. Journal of Virology 2013;87(1):257-272. 340. Asaoka K, Ikeda K, Hishinuma T, et al. A retrovirus restriction factor TRIM5alpha is transcriptionally regulated by interferons. Biochemical and Biophysical Research Communications 2005;338:1950-1956. 341. Kong H, Anderson D, Lee C, et al. Cutting edge: autoantigen ro52 is an interferon inducible E3 ligase that ubiquitinates IRF-8 and enhances cytokine expression in macrophages. Journal of Immunology 2007;179:26-30. 342. Yang K, Shi H, Liu X, et al. TRIM21 is essential to sustain IFN regulatory factor 3 activation during antiviral response. Journal of Immunology 2009;182:3782-3792. 343. Yap M, Stoye J. TRIM proteins and the innate immune response to viruses. Advances in Experimental Medicine and Biology 2012;770:93-104. 344. Carthagena L, Bergamaschi A, Luna J, et al. Human TRIM gene expression in response to interferons. Plos One 2009;4(3):1-13. 345. Xue Q, Zhou Z, Lei X, et al. TRIM38 negatively regulates TLR3-mediated IFN-beta signaling by targeting TRIF for degradation. Plos One 2012;7(10):1-10. 346. Zhao W, Wang L, Zhang M, et al. Tripartite motif-containing protein 38 negatively regulates TLR3/4- and RIG-I-mediated IFN- production and antiviral response by targeting NAP1. The Journal of Immunology 2012;188(11):5311-5318. 347. Wang J, Liu B, Wang N, et al. TRIM56 is a virus- and interferon- inducible E3 ubiquitin ligase that restricts pestivirus infection. Journal of Virology 2011;85(8):3733-3745.

381

348. Tsuchida T, Zou J, Saitoh T, et al. The ubiquitin ligase TRIM56 regulates innate immune responses to intracellular double-stranded DNA. Immunity 2010;33:765-776. 349. Forbes S, Bindal N, Bamford S, et al. COSMIC: mining complete cancer genomes in the catalogue of somatic mutations in cancer. Nucleic Acids Research 2010;39:945- 950. 350. Marshall G, Bell J, Koach J, et al. TRIM16 acts as a tumour suppressor by inhibitory effects on cytoplasmic vimentin and nuclear E2F1 in neuroblastoma cells. Oncogene 2010;29:6172-6183. 351. Bell J, Malyukova A, Kavallaris M, et al. TRIM16 inhibits neuroblastoma cell proliferation through cell cycle regulation and dynamic nuclear localization. Cell Cycle 2013;12(6):889-898. 352. Kim P, Rahmanto A, Tan O, et al. TRIM16 overexpression induces apoptosis through activation of caspase-2 in cancer cells. Apoptosis 2013;18:639-651. 353. Raif A, Marshall G, Bell J, et al. The estrogen-responsive b box protein (EBBP) restores retinoid sensitivity in retinoid-resistant cancer cells via effects on histone acetylation. Cancer Letters 2009;277:82-90. 354. Cheung B, Koach J, Tan O, et al. The retinoid signalling molecule, TRIM16, is repressed during squamous cell carcinoma skin carcinogenesis in vivo and reduces skin cancer cell migration in vitro. The Journal of Pathology 2012;226(3):451-462. 355. Cheung B, Bell J, Raif A, et al. The estrogen-responsive b box protein is a novel regulator of the retinoid signal. The Journal of Biological Chemistry 2006;281(26):18246-18256. 356. Chakravarti N, Lotan R, Diwan A, et al. Decreased expression of retinoid receptors in melanoma: entailment in tumorigenesis and prognosis. Clinical Cancer Research 2007;13(16):4817-4824. 357. Fountain J, Karayiorgou M, Taruscio D, et al. Genetic and physical map of the interferon region on chromosome 9p. Genomics 1992;14:105-112. 358. Rakosy Z, Vizkeleti L, Ecsedi S, et al. Characterization of 9p21 copy number alterations in human melanoma by fluorescence in situ hybridization. Cancer Genetics and Cytogenetics 2008;182:116-121. 359. Palmieri G, Cossu A, Botti G, et al. Definition of the role of chromosome 9p21 in sporadic melanoma through genetic analysis of primary tumours and their metastases. British Journal of Cancer 2000;83(12):1707-1714. 360. Conway C, Beswick S, Elliott F, et al. Deletion at chromosome arm 9p in relation to BRAF/NRAS mutations and prognostic significance for primary melanoma. Genes, Chromosomes & Cancer 2010;49:425-438.

382

361. Maccioni L, Rachakonda P, Bermejo J, et al. Variants at the 9p21 locus and melanoma risk. BMC Cancer 2013;13(1):1-15. 362. Olopade O, Buchhagen D, Malik K, et al. Homozygous loss of the interferon genes defines the critical region on 9p that is deleted in lung cancers. Cancer Research 1993;53:2410-2415. 363. Wakabayashi T, Natsume A, Hashizume Y, et al. A phase I clinical trial of interferon- beta gene therapy for high-grade glioma: novel findings from gene expression profiling and autopsy. Journal of Gene Medicine 2008;10:329-339. 364. Dunn G, Koebel C, Schreiber R. Interferons, immunity and cancer immunoediting. Nature Reviews Immunology 2006;6:836-848. 365. Merkel C, Medrano R, Barauna V, et al. Combined p19Arf and interferon-beta gene transfer enhances cell death of B16 melanoma in vitro and in vivo. Cancer Gene Therapy 2013:1-9. 366. Ryuke Y, Mizuno M, Natsume A, et al. Growth inhibition of subcutaneous mouse melanoma and induction of natural killer cells by liposome-mediated interferon-beta gene therapy. Melanoma Research 2003;13(4):349-356. 367. Kageshita T, Mizuno M, Ono T, et al. Growth inhibition of human malignant melanoma transfected with the human interferon-beta gene by means of cationic liposomes. Melanoma Research 2001;11:337-342. 368. Roh M, Zheng Z, Kim H, et al. Difference of interferon-alpha and interferon-beta on melanoma growth and lymph node metastasis in mice. Melanoma Research 2013;23(2):114-124. 369. Wang X, Liu P, Liu H, et al. Delivery of interferons and siRNA targeting STAT3 using lentiviral vectors suppresses the growth of murine melanoma. Cancer Gene Therapy 2012;19:822-827. 370. Eggermont A, Suciu S, Testori A, et al. Long-term results of the randomized phase III trial EORTC 18991 of adjuvant therapy with pegylated interferon alfa-2b versus observation in resected stage III melanoma. Journal of Clinical Oncology 2012;30(31):3810-3818. 371. Mocellin S, Pasquali S, Rossi C, et al. Interferon alpha adjuvant therapy in patients with high-risk melanoma: a systemic review and meta-analysis. Journal of the National Cancer Institute 2010;102:493-501. 372. Hauschild A, Gogas H, Tarhini A, et al. Practical guidelines for the management of interferon-alpha-2b side effects in patients receiving adjuvant treatment for melanoma: expert opinion. Cancer 2008;112(5):982-994.

383

373. Chawla-Sarkar M, Leaman D, Borden E. Preferential induction of apoptosis by interferon (IFN)-beta compared with IFN-alpha2: correlation with TRAIL/Apo2L induction in melanoma cell lines. Clinical Cancer Research 2001;7:1821-1831. 374. Borden E. Augmentation of effects of interferon-stimulated genes by reversal of epigenetic silencing: potential application to melanoma. Cytokine & Growth Factor Reviews 2007;18:491-501. 375. Qin X, Beckhan C, Brown J, et al. Human and mouse IFN-beta gene therapy exhibits different anti-tumor mechanisms in mouse models. Molecular Therapy 2001;4(4):356-364. 376. Jablonska J, Leschner S, Westphal K, et al. Neutrophils responsive to endogenous IFN-beta regulate tumor angiogenesis and growth in a mouse tumor model. The Journal of Clinical Investigation 2010;120(4):1151-1164. 377. Reu F, Bae S, Cherkassky L, et al. Overcoming resistance to interferon-induced apoptosis of renal carcinoma and melanoma cells by DNA demethylation. Journal of Clinical Oncology 2006;24(23):3771-3779. 378. Satomi H, Wang B, Fujisawa H, et al. Interferon-beta from melanoma cells suppresses the proliferations of melanoma cells in an autocrine manner. Cytokine 2002;18(2):108-115. 379. Borden E, Jacobs B, Hollovary E, et al. Gene regulatory and clinical effects of interferon-beta in patients with metastatic melanoma: a phase II trial. Journal of Interferon & Cytokine Research 2011;31(5):433-440. 380. Dunn I, Haggerty T, Kono M, et al. Enhancement of human melanoma antigen expression by IFN-beta. Journal of Immunology 2007;179:2134-2142. 381. Nunez N, Andreani V, Crespo M, et al. IFNB produced by TLR4-activated tumor cells is involved in improving the antitumoral immune response. Cancer Research 2012;72(3):592-603. 382. Natsumi A, Ishii D, Wakabayashi T, et al. IFN-beta down-regulates the expression of DNA repair gene MGMT and sensitizes resistant glioma cells to temozolomide. Cancer Research 2005;65(17):7573-7579. 383. Rapprich H, Hagedorn M. Intralesional therapy of metastatic spreading of melanoma with beta-interferon. Journal of the German Society of Dermatology 2006;4(9):743- 746. 384. Aoyagi S, Hata H, Homma E, et al. Sequential local injection of low-dose interferon- beta for maintenance therapy in stage II and III melanoma: a single-institution matched case-control study. Oncology 2012;82:139-146. 385. Ford E, Thanos D. The transcriptional code of human IFN-beta gene expression. Biochimica et Biophysica Acta 2010;1799:328-336.

384

386. Panne D, Maniatis T, Harrison S. An atomic model of the interferon-beta enhanceosome. Cell 2007;129:1111-1123. 387. Pan Y, Nussinov R. The role of response elements organization in transcription factor selectivity: the IFN-beta enhanceosome example. PLoS Computational Biology 2011;7(6):1-14. 388. Merika M, Williams A, Chen G, et al. Recruitment of CBP/p300 by the IFNbeta enhanceosome is required for synergistic activation of transcription. Molecular Cell 1998;1:277-287. 389. Falvo J, Parekh B, Lin C, et al. Assembly of a functional beta interferon enhanceosome is dependent on ATF-2-c-jun heterodimer orientation. Molecular and Cellular Biology 2000;20(13):4814-4825. 390. Panne D. The enhanceosome. Current Opinion in Structural Biology 2008;18:236- 242. 391. Kim T, Kim T, Lee W, et al. Signaling pathways to the assembly of an interfeon-beta enhanceosome. Journal of Biological Chemistry 2000;275(22):16910-16917. 392. Livak K, Schmittgen T. Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) method. Methods 2001;25(4):402-408. 393. Alam H, Sehgal L, Kundu S, et al. Novel functions of keratins 5 and 14 in proliferation and differentiation of stratified epithelial cells. Molecular Biology of the Cell 2011;22(21):4068-4078. 394. Hafner M, Wenk J, Nenci A, et al. Keratin 14 Cre transgenic mice authenticate keratin 14 as an oocyte-expressed protein. Genesis 2004;38:176-181. 395. Coulombe P, Kopan R, Fuchs E. Expression of keratin K14 in the epidermis and hair follicle: insights into complex programs of differentiation. The Journal of Cell Biology 1989;109:2295-2312. 396. Cummins D, Cummins J, Pantle H, et al. Cutaneous malignant melanoma. Mayo Clinic Proceedings 2006;81(4):500-507. 397. Gogas H, Kirkwood J, Sondak V. Chemotherapy for metastatic melanoma. Cancer 2007;109(3):455-464. 398. Lee H, Zhang P, Herrmann A, et al. Acetylated STAT3 is crucial for methylation of tumour-suppressor gene promoters and inhibition by resveratrol results in demethylation. PNAS 2012;109(20):7769-7769. 399. Yu X, Vazquez A, Levine A, et al. Allele-specific p53 mutant reactivation. Cancer Cell 2012;21:614-625. 400. Finger E, Cheng C, Williams T, et al. CTGF is a therapeutic target for metastatic melanoma. Oncogene 2013;33(9):1093-1100. 385

401. Passeron T, Valencia J, Namiki T, et al. Upregulation of sox9 inhibits the growth of human and mouse melanomas and restores their sensitivity to retinoic acid. Journal of Clinical Investigation 2009;119(4):954-963. 402. Bell J, Malyukova, A., Kavallaris, M., Marshall, GM., Cheung, BB. TRIM16 inhibits neuroblastoma cell proliferation through cell cycle regulation and dynamic nuclear localization. Cell Cycle 2013;12(6):889-898. 403. Talantov D, Mazumder A, Yu J, et al. Novel genes associated with malignant melanoma but not benign melanocytic lesions. Clinical Cancer Research 2005;11(20):7234-7242. 404. In't Hout F, Haydu L, Murali R, et al. Prognostic importance of the extent of ulceration in patients with clinically localized cutaneous melanoma. Annals of Surgery 2012;255(6):1165-1170. 405. Schilling B, Sondermann W, Zhao F, et al. Differential influence of vemurafenib and dabrafenib on patients' lymphocytes despite similar clinical efficacy in melanoma. Ann Oncol 2014;00:1-7. 406. Bassi C, Ho J, Srikumar T, et al. Nuclear PTEN controls DNA repair and sensitivity to genotoxic stress. Science 2013;341:395-399. 407. Chang C, Mulholland D, Valamehr B, et al. PTEN nuclear localization is regulated by oxidative stress and mediates p53-dependent tumor suppression. Molecular and Cellular Biology 2008;28(10):3281-3289. 408. Trotman L, Wang X, Alimonto A, et al. Ubiquitination regulates PTEN nuclear import and tumor suppression. Cell 2007;128:141-156. 409. Bykov V, Issaeva N, Shilov A, et al. Restoration of the tumour suppressor function to mutant p53 by a low-molecular-weight compound. Nature Medicine 2002;8(3):282- 288. 410. Wang W, Rastinejad F, El-Deiry W. Restoring p53-dependent tumour suppression. Cancer Biology & Therapy 2003;2:S55-S63. 411. Ji Q, Hao X, Meng Y, et al. Restoration of tumour suppressor miR-34 inhibits human p53-mutant gastric cancer tumorspheres. BMC Cancer 2008;8(266):1-12. 412. Lu M, Breyssens H, Salter V, et al. Restoring p53 function in human melanoma cells by inhibiting MDM2 and cyclin B1/CDK1-phosphorylated nuclear iASPP. Cancer Cell 2013;23:618-633. 413. Wang X, Trotman L, Koppie T, et al. NEDD4-1 is a proto-oncogenic ubiquitin ligase for PTEN. Cell 2007;128:129-139. 414. Fouladkou F, Landry T, Kawabe H, et al. The ubiquitin ligase Nedd4-1 is despensable for the regulation of PTEN stability and localization. PNAS 2008;105(25):8585-8590.

386

415. Kito Y, Bai J, Goto N, et al. Pathobiological properties of the ubiquitin ligase Nedd4L in melanoma. International Journal of Experimental Pathology 2013;95:24-28. 416. Massiah MA, Simmons BN, Short KM, et al. Solution structure of the RBCC/TRIM B-box1 domain of human MID1: B-box with a RING. Journal of Molecular Biology 2006;358(2):532-45. 417. Kudo-Saito C, Shirako H, Takeuchi T, et al. Cancer metastasis is accelerated through immunosuppression during Snail-induced EMT of cancer cells. Cancer Cell 2009;15(3):195-206. 418. Chen Y, Gelfond J, McManus L, et al. Reproducibility of quantitative RT-PCR array in miRNA expression profiling and comparison with microarray analysis. BMC Genomics 2009;10:1-10. 419. Gaj S, Eijssen L, Mensink R, et al. Validating nutrient-related gene expression changes from microarrays using RT2 PCR arrays. Genes & Nutrition 2008;3:153- 157. 420. de Visser K, Eichten A, Coussens L. Paradoxical roles of the immune system during cancer development. Nature Reviews Cancer 2006;6(1):24-37. 421. Lai KC, Lu CC, Tang YJ, et al. Allyl isothiocyanate inhibits cell metastasis through suppression of the MAPK pathways in epidermal growth factorstimulated HT29 human colorectal adenocarcinoma cells. Oncol Rep 2013;31(1):189-96. 422. Huang C, Jacobson K, Schaller M. MAP kinases and cell migration. Journal of Cell Science 2004;117:4619-4628. 423. Tsao H, Chin L, Garraway LA, et al. Melanoma: from mutations to medicine. Genes Dev 2012;26(11):1131-55. 424. Straume O, Sviland L, Akslen L. Loss of nuclear p16 protein expression correlates with increased tumor cell proliferation (Ki-67) and poor prognosis in patients with vertical growth phase melanoma. Clinical Cancer Research 2000;6(5):1845-1853. 425. Cheng Y, Chen G, Martinka M, et al. Prognostic significance of Fbw7 in human melanoma and its role in cell migration. Journal of Investigative Dermatology 2013;133:1794-1802. 426. Shen Z, Wen X, Lan F, et al. The tumor suppressor gene LKB1 is associated with prognosis in human breast cancer. Clinical Cancer Research 2002;8(7):2085-2090. 427. Tang Y, Cheng Y, Martinka M, et al. Prognostic significance of KAI1/CD82 in human melanoma and its role in cell migration and invasion through the regulation of ING4. Carcinogenesis 2013;35(1):86-95. 428. Schlenk R, Dohner K. Impact of new prognostic markers in treatment decisions in acute myeloid leukemia. Current Opinion in Hematology 2009;16(2):98-104.

387

429. Kerr D, Shi Y. Tailoring treatment and trials to prognosis. Nature Reviews Clinical Oncology 2013;10:429-430. 430. Hait WN. Anticancer drug development: the grand challenges. Nature Reviews Drug Discovery 2010;9(4):253-4. 431. Honda K, Yanai H, Takaoka A, et al. Regulation of the type I IFN induction: a current view. International Immunology 2005;17(11):1367-1378. 432. McMasters K, Edwards M, Ross M, et al. Ulceration as a predictive marker for response to adjuvant interferon therapy in melanoma. Annals of Surgery 2010;252:460-466. 433. Cheon D, Orsulic S. Mouse models of cancer. Annual Review of Pathology: Mechanisms of Disease 2011;6:95-119. 434. Taneja P, Zhu S, Maglic D, et al. Transgenic and knockout mice models to reveal the functions of tumour suppressor genes. Clinical Medicine Insights: Oncology 2011;5:235-257. 435. Sampath H, Flowers M, Liu X, et al. Skin-specific deletion of stearoyl-CoA desaturase-1 alters skin lipid composition and protects mice from high fat diet- induced obesity. Journal of Biological Chemistry 2009;284(30):19961-19973. 436. Cui C, Kunisada M, Piao Y, et al. Dkk4 and Eda regulate distinctive developmental mechanisms for subtypes of mouse hair. Plos One 2010;5(4):e10009. 437. Qiao W, Li AG, Owens P, et al. Hair follicle defects and squamous cell carcinoma formation in Smad4 conditional knockout mouse skin. Oncogene 2006;25(2):207-17. 438. Lee G, Lee K, Choi K, et al. Phenotype of a calbindin-D9k gene knockout is compensated for by the induction of other calcium transporter genes in a mouse model. Journal of Bone Mineral Research 2007;22(12):1968-1978. 439. Barbaric I, Miller G, Dear T. Appearances can be deceiving: phenotypes of knockout mice. Breifings in Functional Genomics and Proteomics 2007;6(2):91-103. 440. Kitano H. Biological robustness. Nature Reviews Genetics 2004;5(11):826-837. 441. Bonaldo P, Sandri M. Cellular and molecular mechanisms of muscle atrophy. Disease Models & Mechanisms 2013;6:25-39. 442. Abel E, Angel J, Kiguchi K, et al. Multi-stage chemical carcinogenesis in mouse skin: fundamentals and applications. Nature Protocols 2009;4(9):1350-1362. 443. Azarova A, Lyu Y, Lin C, et al. Roles of DNA topoisomerase II isozymes in chemotherapy and secondary malignancies. PNAS 2007;104(26):11014-11019. 444. Sundberg J, Sundberg B, Beamer W. Comparison of chemical carcinogen skin tumour induction efficacy in inbred, mutant, and hybrid strains of mice: morphologic

388

variations of induced tumours and absence of a papillomavirus cocarcinogen. Molecular Carcinogenesis 1997;20:19-32. 445. Schramek D, Sendoel A, Segal J, et al. Direct in vivo RNAi screen unveils myosin IIa as a tumour suppressor of squamous cell carcinoma. Science 2014;343(309-313). 446. Goel VK, Ibrahim N, Jiang G, et al. Melanocytic nevus-like hyperplasia and melanoma in transgenic BRAFV600E mice. Oncogene 2009;28(23):2289-98. 447. Lee E, Muller W. Oncogenes and tumour suppressor genes. Cold Spring Harbor Perspectives in Biology 2010;2(a003236). 448. Bhawan J, Whren K, Panova I, et al. Keratin 16 expression in epidermal melanocytes of normal human skin. American Journal of Dermatopathology 2005;27(6):476-481. 449. Vogel W, Guislain A, Kvistborg P, et al. Ipilimumab-induced sarcoidosis in a patient with metastatic melanoma undergoing complete remission. Journal of Clinical Oncology 2011;30(2):e7-e10. 450. Marsh Durban V, Deuker MM, Bosenberg MW, et al. Differential AKT dependency displayed by mouse models of BRAFV600E-initiated melanoma. Journal of Clinical Investigation 2013;123(12):5104-18. 451. Woodworth C, Michael E, Smith L, et al. Strain-dependent differences in malignant conversion of mouse skin tumors is an inherent property of the epidermal keratinocyte. Carcinogenesis 2004;25(9):1771-1778. 452. Lu TL, Chang JL, Liang CC, et al. Tumor spectrum, tumor latency and tumor incidence of the Pten-deficient mice. Plos One 2007;2(11):e1237. 453. Sachs N, Secades P, van Hulst L, et al. Loss of integrin alpha3 prevents skin tumor formation by promoting epidermal turnover and depletion of slow-cycling cells. Proc Natl Acad Sci U S A 2012;109(52):21468-73. 454. Sachs N, Secades P, van Hulst L, et al. Reduced susceptibility to two-stage skin carcinogenesis in mice with epidermis-specific deletion of CD151. Journal of Investigative Dermatology 2014;134(1):221-8. 455. Becks L, Prince M, Burson H, et al. Aggressive mammary carcinoma progression in Nrf2 knockout mice treated with 7,12-dimethylbenz[a]anthracene. BMC Cancer 2010;10:540. 456. Seiberg M. Keratinocyte-melanocyte interactions during melanosome transfer. Pigment Cell Research 2001;14(4):236-242. 457. Kippenberger S, Bernd A, Bereiter-Hahn J, et al. The mechanism of melanocyte dendrite formation: the impact of differentiating keratinocytes. Pigment Cell Research 1998;11(1):34-37.

389

458. Jamal S, Schneider R. UV-induction of keratinocyte endothelin-1 downregulates E- cadherin in melanocytes and melanoma cells. Journal of Clinical Investigation 2002;110:443-452. 459. Fujisawa H, Kondo S, Wang B, et al. The expression and modulation of IFN-alpha and IFN-beta in human keratinocytes. Journal of Interferon & Cytokine Research 1997;17(12):721-725. 460. Kariko K, Bhuyan P, Capodici J, et al. Small interfering RNAs mediate sequence- independent gene suppression and induce immune activation by signaling through toll-like receptor 3. Journal of Immunology 2004;172:6545-6549. 461. Petrova E, Lopez-Gay J, Rhiner C, et al. Flower-deficient mice have reduced susceptibility to skin papilloma formation. Disease Models & Mechanisms 2012;5:553-561. 462. Conde-Perez A, Larue L. Human relevance of NRAS/BRAF mouse melanoma models. European Journal of Cell Biology 2013;171-9339(13):00069-1. 463. Ackermann J, Frutschi M, Kaloulis K, et al. Metastasizing melanoma formation caused by expression of activated N-RasQ61R on an INK4a-deficient background. Cancer Research 2005;65(10):4005-4011. 464. Berger M, Hodis E, Heffernan T, et al. Melanoma genoma sequencing reveals frequent PREX2 mutations. Nature 2012;485:502-506. 465. Tsai K, Vin H, Leung M, et al. Suppression of apoptosis by BRAF inhibitors through off-target inhibition of JNK signaling. 2012 ASCO Annual Meeting. Chicargo, Illinois 2012. 466. Hambley TW, Hait WN. Is anticancer drug development heading in the right direction? Cancer Research 2009;69(4):1259-62. 467. Muller PY, Milton MN. The determination and interpretation of the therapeutic index in drug development. Nature Reviews Drug Discovery 2012;11(10):751-61. 468. Lipinski C. Lead- and drug-like compounds: the rule-of-five revolution. Drug Discovery Today: Technologies 2004;11(007):337-341. 469. Menzies A, Haydu L, Carlino M, et al. Inter-and intra-patient heterogeneity of response and progression to targeted therapy in metastatic melanoma. Plos One 2014;9(1). 470. Chou TC. Theoretical basis, experimental design, and computerized simulation of synergism and antagonism in drug combination studies. Pharmacology Reviews 2006;58(3):621-81. 471. Le K, Blomain ES, Rodeck U, et al. Selective RAF inhibitor impairs ERK1/2 phosphorylation and growth in mutant NRAS, vemurafenib-resistant melanoma cells. Pigment Cell Melanoma Research 2013;26(4):509-17. 390

472. Abel EV, Aplin AE. FOXD3 is a mutant B-RAF-regulated inhibitor of G(1)-S progression in melanoma cells. Cancer Research 2010;70(7):2891-900. 473. Collado M, Serrano M. Senescence in tumours: evidence from mice and humans. Nature Reviews Cancer 2010;10(1):51-7. 474. Kuilman T, Michaloglou C, Mooi WJ, et al. The essence of senescence. Genes Dev 2010;24(22):2463-79. 475. Shi H, Moriceau G, Kong X, et al. Melanoma whole-exome sequencing identifies (V600E)B-RAF amplification-mediated acquired B-RAF inhibitor resistance. Nature Communications 2012;3:724. 476. Franken N, Rodermond H, Stap J, et al. Clonogenic assay of cells in vitro. Nature Protocols 2006;1(5):2315-2319. 477. Keseru G, Makara G. Hit discovery and hit-to-lead approaches. Drug Discovery Today 2006;11(15/16):741-748. 478. McCubrey J, Steelman L, Chappell W, et al. Roles of the Raf/MEK/ERK pathway in cell growth, malignant transformation and drug resistance. Biochimica et Biophysica Acta 2007;1773:1263-1284. 479. Houben R, Ortmann S, Drasche A, et al. Prolieration arrest in B-Raf mutant melanoma cell lines upon MAPK pathway activation. Journal of Investigative Dermatology 2009;129:406-414. 480. Woods D, Parry D, Cherwinski H, et al. Raf-induced proliferation or cell cycle arrest is determined by the level of Raf activity with arrest mediated by p21Cip1. Molecular and Cellular Biology 1997;17(9):5598-5611. 481. Zhang Z, Miao L, Lv C, et al. Wentilactone B induces G2/M phase arrest and apoptosis via the Ras/Raf/MAPK signaling pathway in human hepatoma SMMC- 7721 cells. Cell Death and Disease 2013;4:e657. 482. Gelzo M, Granato G, Albano F, et al. Evaluation of cytotoxic effects of 7- dehydrocholesterol on melanoma cells. Free Radicals Biology Medicine 2014. 483. Rappa G, Fodstad O, Lorico A. The stem cell-associated antigen CD133 (Prominin-1) is a molecular therapeutic target for metastatic melanoma. Stem Cells 2008;26(12):3008-17. 484. Yuan G, Wang C, Ma C, et al. Oncogenic function of DACT1 in colon cancer through the regulation of beta-catenin. Plos One 2012;7(3):e34004. 485. Kunz M, Moeller S, Koczan D, et al. Mechanisms of hypoxic gene regulation of angiogenesis factor Cyr61 in melanoma cells. Journal of Biological Chemistry 2003;278(46):45651-60.

391

486. Babic AM, Kireeva ML, Kolesnikova TV, et al. CYR61, a product of a growth factor- inducible immediate early gene, promotes angiogenesis and tumor growth. Proc Natl Acad Sci U S A 1998;95(11):6355-60. 487. Walker TM, Van Ginkel PR, Gee RL, et al. Expression of angiogenic factors Cyr61 and tissue factor in uveal melanoma. Arch Ophthalmol 2002;120(12):1719-25. 488. Si W, Kang Q, Luu HH, et al. CCN1/Cyr61 is regulated by the canonical Wnt signal and plays an important role in Wnt3A-induced osteoblast differentiation of mesenchymal stem cells. Molecular Cell Biology 2006;26(8):2955-64. 489. Lindner DJ, Wu Y, Haney R, et al. Thrombospondin-1 expression in melanoma is blocked by methylation and targeted reversal by 5-Aza-deoxycytidine suppresses angiogenesis. Matrix Biology 2013;32(2):123-32. 490. Tsuchida R, Osawa T, Wang F, et al. BMP4/Thrombospondin-1 loop paracrinically inhibits tumor angiogenesis and suppresses the growth of solid tumors. Oncogene 2013. 491. Zhou ZQ, Cao WH, Xie JJ, et al. Expression and prognostic significance of THBS1, Cyr61 and CTGF in esophageal squamous cell carcinoma. BMC Cancer 2009;9:291. 492. Di L, Kerns EH, Ma XJ, et al. Applications of high throughput microsomal stability assay in drug discovery. Comb Chem High Throughput Screen 2008;11(6):469-76. 493. Pradhallad A, Sun C, Huang S, et al. Unresponsiveness of colon cancer to BRAF(V600E) inhibition through feedback activation of EGFR. Nature Medicine 2012;483:626-633. 494. Lai F, Guo S, Jin C, et al. Cotargeting histone deacetylases and oncogenic BRAF synergistically kills human melanoma cells by necrosis independently of RIPK1 and RIPK3. Cell Death and Differentiation 2013. 495. Hsu M, Meier F, Nesbit M, et al. E-Cadherin expression in melanoma cells restores keratinocyte-mediated growth control and down-regulates expression of invasion- related adhesion receptors. American Journal of Pathology 2000;156(5):1515-1525. 496. Hausmann S, Marq JB, Tapparel C, et al. RIG-I and dsRNA-induced IFNbeta activation. Plos One 2008;3(12):e3965. 497. Sloot S, Fedorenko I, Smalley K, et al. Long-term effects of BRAF inhibitors in melanoma treatment: friend or foe? Expert Opinion on Pharmacotherapy 2014;Ahead of print.

392

393